SECTION 2 Background to medicine

2.1 Science in medicine When, how, and what 33
2.1 Science in medicine: When, how, and what 33
ESSENTIALS
Science has always been part of Western medicine, although what
counts as scientific has changed over the centuries, as have the con-
tent of medical knowledge, the tools of medical investigation, and
the details of medical treatments. This brief overview develops a his-
torical typology of medicine since antiquity. It divides the ‘kinds’ of
medicine into five sections:
(1) Bedside medicine, developed by the Hippocratic doctors in
classical times, has its modern counterpart in primary care.
(2) Library medicine, associated with the scholastic mentality of
the Middle Ages, still surfaces in the problems of information storage
and retrieval in the computer age.
(3) Hospital medicine, central to French medicine of the early
19th century, placed the diagnostic and therapeutic functions of the
modern hospital centre stage in care and teaching.
(4) Social medicine is about prevention, both communal and in-
dividual, and is especially visible in our notion of ‘lifestyle’ and its
impact on health.
(5) Laboratory medicine has its natural home in the research es-
tablishment and is a critical site for the creation of medical know-
ledge, setting the standards for both medical science and scientific
medicine.
Introduction
At least since the Hippocratic era, Western medicine has always as-
pired to be scientific. What has changed is not so much the aspir-
ations but what it has meant to be ‘scientific’.
‘Science is the father of knowledge, but opinion breeds ignorance’,
opined the Hippocratic treatise The Canon, and Hippocratic practi-
tioners developed an approach to health, disease, and its treatment
based on systematic observation and cumulative experience. Even
the word ‘physic’, whence physician as well as physicist, derives from
the Greek for ‘nature’. Further, Hippocratic medicine was experi-
mental, that word stemming from the same classical roots which
gave us ‘experience’.
Words, however, can be slippery, as philosophers as divergent as
Francis Bacon and Ludwig Wittgenstein have stressed. The science
and experiment of the Hippocratics can still inspire, but they are not
our science and experiment. During the past two or three centuries,
an armoury of sciences and technologies has come to underpin
medical practice. This essay briefly describes these, within the con-
text of distinctive and perennial features of medical practice (i.e. suf-
fering individuals whose problems and diseases demand attention).
A historical typology of Western medicine
The history of Western medicine can be divided into five ‘kinds’ of
medicine: bedside, library, hospital, social, and laboratory, with the
latter extended to include technology (Table 2.1.1). Each approach
to medical care and knowledge emerged at a particular historical
period, but each still has relevance to us today.
Bedside medicine
Bedside medicine can be equated with the vision of the Hippocratics,
with its emphasis on the individual patient, a tendency towards
holism, and an abiding concern with the patient within his or
her own unique environment. These are some of the reasons why
Hippocrates (Fig. 2.1.1) is still claimed as the dominant father figure
by both orthodox and alternative medical practitioners.
Library medicine
What can be called ‘library’ medicine dominated in the Middle
Ages, when learned medicine retreated into the universities and
scholars sometimes assumed that everything worth discovering
had been uncovered by the ancients, and everything worth being
revealed could be found in the Bible. The millennium between
the sacking of Rome and the discovery of the New World is often
dismissed as a sterile period scientifically, but the physicians of
the period, linguistically erudite and philosophically inclined,
would have been surprised to be described as unscientific.
They simply believed that the road to knowledge was through
the book, and—​were they able to—​would point out that this so-​
called sterile period gave us the hospital and university.
2.1
Science in medicine: When, how, and what
William F. Bynum
34
SECTION 2  Background to medicine
These medical men also sometimes engaged with nature, although
it is undeniable that nature rather than words became an increasing
source of truth and knowledge during the Scientific Revolution, a
period stretching roughly from just before Andreas Vesalius (1514–​
1564) to Isaac Newton (1642–​1727). Around 1600, it was becoming
apparent to many that the Greeks had not left behind a complete
and accurate account of the nature of the world, and that scientific
knowledge was cumulative. This ‘Battle of the Books’, the debate over
whether the ancients or the moderns knew more, was decided in fa-
vour of the moderns.
Many of the outstanding scientific achievements of the era
were in astronomy and physics, but medicine, both in its theory
and its practice, was also affected. Theory has always been easier
to change than practice, of course, and it was famously remarked
that William Harvey’s discovery of the circulation of the blood
had no impact on therapeutics. Harvey (1578–​1657) also notori-
ously lamented that his practice fell off mightily following the
discovery, his patients fearing that he was ‘crack-​brained’. The
fear that too close an identification with science was detrimental
to patient confidence recurs in medical history, and is still part
of the delicate negotiations between the profession and its public,
and to the status of academic medicine.
Within the discipline of medicine itself there have always been
individuals—​some of them, like Thomas Sydenham (1624–​1689),
eminently successful—​who believed that experimental science had
little to offer to patient care. But these ‘artists’ of medicine could still
invoke the authority of Hippocrates, with its older connotations
of knowledge and experience. Sydenham himself did not demur
from his being dubbed ‘the English Hippocrates’. During the early
modern period, the whole spectrum of the sciences—​mathematics,
physics, chemistry, the life sciences (not yet called biology)—​made
their ways into formulations of health and disease. Iatrophysics,
iatromathematics, and iatrochemistry all had their advocates in
the 17th and 18th centuries as approaches to medical theory and
practice.
That these systems tended to encourage speculation to run ahead
of evidence was recognized at the time, and this was part of the
reason why ‘hospital medicine’ had little recourse to those discip-
lines we now call ‘basic medical sciences’. The founders of French
hospital medicine, Xavier Bichat (1771–​1802), J.  N. Corvisart
(1755–​1821), and R. T. H. Laennec (1781–​1826), often referred to
chemistry, physiology, and the like as sciences that were ‘accessory’
to medicine.
Hospital medicine
The medicine that developed in the Paris hospitals, after the re-
opening in 1794 of the medical schools closed by the Revolution a
couple of years earlier, emphasized above all the study of disease in
the sick patient. In a sense, this was Hippocratic medicine writ large,
but with some significant differences.
First, the hospital offered the curious doctor a vast arena for
observing disease. The equivalent of a lifetime’s experience of a
lone practitioner in the community could be experienced in a few
Fig. 2.1.1  A statue of Hippocrates, originally from Kos Odeion but
now in the Archaeological Museum of Kos. This is a late Hellenistic
period copy of a classical prototype. No contemporary likeness of
Hippocrates exists, but several busts and statues were created later on
in the classical period.
Copyright © D. A. Warrell.
Table 2.1.1  A historical typology of Western medicine
CHARACTERISTICS
Object of inquiry
Form and site of education
Goal
Example
T
Y
P
E
S
Bedside
Whole patient
Apprenticeship
Therapy
Hippocrates (c.460–​370 bce)
Library
Text
Scholastic, linguistic,
University
Preservation, recovery,
commentary
Constantine the African (d. before 1098)
Hospital
Patient, organ, lesion
Hospital
Diagnosis
R. T. H. Laennec (1781–​1826)
Social
Population, statistic
Community
Prevent
John Simon (1816–​1904)
Laboratory
Animal model
Laboratory
Understand
Claude Bernard (1813–​1878)
Technology
Molecule
Automated laboratory
The above + profit
Kary Mullis (b. 1944)
Adapted from W. F. Bynum, History of Medicine: a very short introduction, OUP, 2008.
2.1  Science in medicine: When, how, and what
35
months of hospital work. Hospitals offered the possibility of defining
disease on the basis of hundreds of cases.
Secondly, Hippocratic humoralism gradually disappeared as the
dominant explanatory framework of health and disease, replaced
by the primacy of the lesion, located in the solids: the organs and
tissues, and by the mid-​19th century, cells. In this new orientation,
disease was literally palpable, its lesions to be discovered in life by
the systematic use of physical examination—​Corvisart rediscovered
percussion, Laennec invented the stethoscope—​and these findings
to be correlated after death by routine autopsy. French high priests
of hospital medicine brought diagnosis to a new stage and replaced
the older symptom-​based nosologies with a more objective, demon-
strable one of lesions.
The third feature of hospital medicine was what Pierre Louis
(1787–​1872) called the numerical method, the use of numbers to
guide both disease classification and therapeutic evaluation.
The philosophy underlying early 19th-​century French medicine
was most systematically expounded by one of the many American
students who studied in Paris, Elisha Bartlett, in his Philosophy of
Medical Science (1844). The medical science whose philosophy
he chronicled was one of facts. Bartlett argued that all systems of
medicine, past and present, were speculative, vague, and useless.
Cullen, Brown, Broussais, and Hahnemann were all consigned to
the historical dustbin. The new medicine was one of systematic
observation and collection of facts, which, when properly com-
pared and organized, could provide an objective understanding of
disease and a rational basis for its treatment. Bartlett’s philosophy
was essentially undiluted Baconian inductivism applied to medi-
cine. Unsurprisingly, he counted Hippocrates as well as Pierre Louis
among his heroes.
One consequence of the lesion-​based medicine was the recogni-
tion that not much of what doctors did actually altered the natural
history of disease. Therapeutic scepticism, or even nihilism, flour-
ished among doctors whose lives were spent, as Laennec put it,
‘among the dead and dying’. It was less likely to be expressed among
doctors concerned with earning a living treating private, paying pa-
tients, but the concern with medicine’s therapeutic impotency also
fuelled the movement to prevent disease.
Social medicine
The fourth kind of medicine, social, also flourished in the 19th cen-
tury. Just as hospitals existed long before ‘hospital medicine’, so epi-
demics and preventive measures were not invented by the public
health movement of the 1830s. Nevertheless, the preventive infra-
structures developed partly in response to the cholera pandemics
still exist, although of course much changed.
The chief architect of the British public health movement, Edwin
Chadwick (1800–​1890), was a lawyer who thought that, on the
whole, doctors were overrated (he was neither the first nor the last
lawyer to hold that opinion). He held that filth spread via the foul
smells (miasma) of rotting organic matter caused epidemic dis-
eases. His solutions were engineering ones—​clean water and effi-
cient waste disposal—​which he argued would leave the world an
altogether more pleasant and healthier place. His ideas were formed
during the 1830s and early 1840s, and they remained more or less
fixed for the rest of his long life, which extended well into the bac-
teriological age. Nevertheless, Chadwick also invoked science in his
public health reform programme, above all the science of statistical
investigation. His use of statistics can easily be shown to have been
naive, but it was ardent. In his own sphere of enquiry, Chadwick
was as much in awe of the unadorned ‘fact’ as was his contemporary
Bartlett. A later generation of Medical Officers of Health and others
concerned with disease prevention (or containment) would develop
new investigative techniques, more sophisticated statistics and, es-
pecially, new theories of disease causation and transmission. But the
early public health movement was firmly based on the science of
its time.
Laboratory medicine
The final locus of medicine, the laboratory, was also largely a product
of the 19th century, though of course laboratories (a place where one
worked, especially to mutate lead into gold) had existed for much
longer. A leading exponent of the laboratory, and one of its most
thoughtful philosophers, had experienced Paris hospital medicine
as a medical student. Claude Bernard’s Introduction to the Study of
Experimental Medicine (1865) is at once an intriguing account of his
own brilliant career and a sophisticated analysis of the philosophy
of experimentation within the life sciences (Fig. 2.1.2). Hospitals,
he argued, are merely the gateways to medical knowledge, and bed-
side clinicians can be no more than natural historians of disease. To
understand the causes and mechanisms of disease, it is necessary
to go into the sanctuary of the laboratory, where experimental con-
ditions can be better controlled. There are in nature no uncaused
causes:  determinism is the iron law of the universe, extending
equally to living systems and inorganic ones. However, organ-
isms present special experimental problems, and it is only through
isolating particular features, and holding other parameters as con-
stant as possible, that reliability and reproducibility can be achieved.
Bernard identified three primary branches of experimental medi-
cine:  physiology, pathology, and therapeutics. His own research
programme touched all three pillars: his research on the roles of
the liver and pancreas in sugar metabolism contributed to under-
standing normal physiology as well as diseases such as diabetes;
his investigations of the sites of action of agents such as curare and
carbon monoxide foreshadowed structural pharmacology and drug
receptor theory; his work on the functions of the sympathetic nerves
buttressed his own more general notion of the constancy of milieu
interieur as the precondition to vital action (and freedom), a pre-
cursor of Walter Cannon’s concept of homeostasis. Bernard stands
supreme as the quintessential advocate of the laboratory.
Who was the first modern medical
experimentalist?
When Bernard wrote, experimental medical science was still a
fledgling activity, best developed in the universities of the German
states and principalities. The German university ideal of medical
education was to be extolled by the American educational re-
former Abraham Flexner (1866–​1959) in the early 20th century.
It was in the reformed and newly created German universities that
the forms of modern scientific research were established. Research
careers were created; copublication in specialist journals became
common; scientific societies flourished. The microscope became
the symbol of the medical scientist even as the stethoscope was
becoming the hallmark of the forward-​looking clinician. In the
36
SECTION 2  Background to medicine
hands of scientists like Schwann, Virchow, and Weismann, the
modern cell theory was developed and applied to medicine and
biology more generally. These researchers established the drive
to push units of analysis further and further. Eduard Buchner’s
identification of cell-​free ferments in 1897 firmly established the
importance of subcellular functions. Pasteur, Koch, Ehrlich, von
Behring, and others advanced new notions of the causes of disease,
the body’s response to infection, and the possibilities of new drugs
to combat disease. Any of these scientists might arguably be the
answer to the parlour-​game question: Who was the first modern
medical scientist?
The German-​speaking lands perfected the modern forms of sci-
entific research, but a good case can be made for a Frenchman to
be crowned the first thoroughly modern experimentalist within
medicine. François Magendie (1783–​1855) (Fig. 2.1.3) was a child
of the Enlightenment and product of the French Revolution. One
of several eminent individuals (Thomas Malthus was another)
raised according to the anarchic principles espoused by Jean-​
Jacques Rousseau, Magendie did not learn to read or write until
he was 10. His subsequent precocity was such that he was ready
for medical studies by the age of 16, learned anatomy and surgery
as an apprentice, and made his way through the Paris hospital
system. Although he never lost interest in practical medical issues,
his reputation was established primarily within the laboratory.
Fig. 2.1.2  Claude Bernard and his pupils. Oil painting, 1889 by Léon Augustin Lhermitte.
Courtesy of the Wellcome Collection, Creative Commons Attribution (CC BY 4.0).
Fig. 2.1.3  François Magendie. Lithograph by N. E. Maurin.
From Burgess R (1973). Portraits of doctors and scientists in the Wellcome Institute,
London, no. 1870.2, by courtesy of the Wellcome Library, London.
2.1  Science in medicine: When, how, and what
37
His monographs on physiology and pharmacology marked new
beginnings, and his life manifests three emblematic qualities
which make him one of us.
First, he valued facts above theories, evidence above rhetoric.
However, he went beyond Bartlett and the high priests of hospital
medicine in insisting that in experiment, and not simply observa-
tion, lay the real future of medical knowledge. Like his pupil Claude
Bernard, Magendie was a deft experimentalist. He used animals
(and occasionally patients) to probe into a whole range of prob-
lems in physiology, pathology, and pharmacology:  the functions
of the spinal nerves, the physiology of vomiting, important facets
of absorption, digestion, circulation, nutrition, and the actions of
drugs and poisons. He described anaphylaxis a century before it was
named. He was as philosophically naive as Bernard was sophisti-
cated: of course he had theories, but his image of himself as a rag-​
picker with a spiked stick, gathering isolated experimental facts
where he found them, is a telling one.
Secondly, he was modern in sometimes backing the wrong horses.
He judged cholera and yellow fever to be non​contagious, was suspi-
cious of anaesthesia, and sometimes claimed more than we might
for his newly introduced therapeutic substances, such as strychnine
and veratrine. Magendie could often be mistaken in his beliefs; so
can we.
Thirdly, Magendie was the scientist who first expunged the
double-​faced Janus from the medical mentality. William Harvey
worshipped Aristotle, Albrecht von Haller was steeped in his-
tory, and Isaac Newton popularized the pious conceit of pygmies
standing on the shoulders of giants. Until the 19th century, doc-
tors routinely looked to the past, not simply for inspiration but for
useful information. Magendie looked only in one direction: the
future. He had no sense of history and no use for it. He meant
what he said when he insisted that most physiological ‘facts’ had
to be verified by new experiments, and he undertook to provide
a beginning. He made the laboratory the bedrock of medicine.
With Magendie, the history of medicine became an antiquarian
discipline.
What happened next?
Like everyone, Magendie was of his time. Nevertheless, his values
were symptomatic of important themes within 19th-​century medi-
cine and medical science. By the beginning of the First World War,
most of the structures and the fundamental concepts of modern
medicine were in place. Of course, both medical science and med-
ical practice have been utterly transformed since. But the impulse of
experimentation and its variable translation into practice were there.
We have gone far beyond the cell in our analytical procedures, and
our medical, surgical, and therapeutic armamentaria are vastly more
sophisticated and powerful.
Technological medicine
Our medicine is fundamentally different in one important respect,
even if the trend was already evident in the 19th century:  the
fusion of science and technology. Science and technology have
become so intertwined that the older distinctions between them
are blurred. Technology made a real but minimal impact on
19th-​century medicine. Some instruments, such as Helmholtz’s
ophthalmoscope, came into clinical medicine through the labora-
tory; and German experimental scientists were eager to exploit
the latest equipment such as kymographs, sphygmographs, and
the profusion of artefacts (Petri dishes, autoclaves, and so on),
which Koch and his colleagues devised for the bacteriological
laboratory.
Most important of all was probably X-​rays, discovered by
Roentgen in late 1895. This made an immediate impact on med-
ical diagnosis, and the associated science of radioactivity soon was
felt within therapeutics (Fig. 2.1.4). Significantly, perhaps, the pi-
oneers of the radioactive phenomena—​Roentgen, Becquerel, the
Curies—​got their Nobel Prizes in physics or chemistry. Hounsfield
and Cormack got theirs for computer-​assisted tomography in
medicine or physiology. More recently, Kary Mullis’s Nobel Prize
was for a technological development within molecular biology.
Both medical science and medical practice are now inseparably
rooted in technology. So is modern life, another reflection of a per-
ennial historical truth: medical knowledge and medical practice are
products of wider social forces with unique historical individualities.
Fig. 2.1.4  Photograph of a radiograph (x-ray negative) taken on
22 December 1895 by Wilhelm Röntgen, the discover of x-rays. It shows
his wife’s left hand; she wears a large ring on her 4th digit.
Courtesy of the Science Museum, London, Creative Commons Attribution
(CC BY 4.0).
38
SECTION 2  Background to medicine
FURTHER READING
Ackerknecht EH (1967). Medicine at the Paris hospital, 1784–​1848.
Johns Hopkins University Press, Baltimore, MD.
Bynum WF (1994). Science and the practice of medicine in the nine-
teenth century. Cambridge University Press, Cambridge.
Bynum WF (2008). History of medicine:  a very short introduction.
Oxford University Press, Oxford.
Bynum WF, Bynum H (eds) (2007). Dictionary of medical biography.
Greenwood, Westport, CT.
Bynum WF, Porter R (eds) (1993). Companion encyclopedia of the his-
tory of medicine. Routledge, London.
Bynum WF, et al. (2006). The Western medical tradition, 1800 to 2000.
Cambridge University Press, Cambridge.
Conrad LI, et al. (1995). The Western medical tradition, 800 bc to ad
1800. Cambridge University Press, Cambridge.
Cooter R, Pickstone J (eds) (2000). Medicine in the 20th century.
Harwood Academic Publishers, Amsterdam.
King LS (1982). Medical thinking:  a historical preface. Princeton
University Press, Princeton, NJ.
Reiser SJ (1978). Medicine and the reign of technology. Cambridge
University Press, Cambridge.
Weatherall DJ (1995). Science and the quiet art: medical research and
patient care. Oxford University Press, Oxford.

2.10 Medicine quality, physicians, and patients 12
2.10 Medicine quality, physicians, and patients 124
ESSENTIALS
Poor-​quality medicines negate the enormous advantages of modern
pharmaceuticals and lead to avoidable morbidity and mortality; loss
of confidence in medicines and healthcare systems; economic losses
for patients, their families, governments, and the pharmaceutical in-
dustry; and—​for anti-​infectives—​engender pathogen drug resistance.
There are many examples of both falsified (due to fraudulent pro-
duction) and substandard (due to in-​factory negligence) medicines,
vaccines, diagnostic tests, and devices. The evidence base informing
our understanding of their epidemiology is poor, but it is becoming
clearer that the distribution of poor-​quality medical products varies
greatly around the world in both time and place. Formal chem-
ical analysis is highly sophisticated, expensive, and rarely available in
many countries. New portable diagnostic devices offer hope that they
could empower medicine inspectors and pharmacists to obtain ob-
jective evidence for selection of suspicious samples for formal analysis.
Caveat emptor (buyer beware) is vital: patients, pharmacists, nurses,
and physicians should observe their medicines and be alert when the
packaging, medicine taste, or side effects change; when they are sur-
prisingly inexpensive; and when they are associated with unexpected
treatment failure and puzzling clinical syndromes. Much greater inter-
national effort is needed for the sharing of data about poor-​quality
medicines so that detection can be acted upon and patients protected.
Introduction
Most patients, doctors, and pharmacists tend to assume that the
medicines they take or provide are of good quality and that poor-​
quality medicines are a problem of the distant past or are far away.
However, there have been numerous problems with deliberate falsi-
fication of medicines and poor-​quality production since medicines
were first manufactured. Attempts to ensure that countries have a
safe and efficacious medicine supply have led to great improvements
in medicine regulation in the wealthier world over the last century.
However, poor-​quality medicines are still with us. There is growing,
but belated, concern that much of the developing world’s supply of
medical products, including medicines, vaccines, diagnostic tests
and devices, in particular its supply of anti-​infective drugs, are of
poor quality. There are also less frequent but important problems in
wealthier countries. The enormous investment in pharmaceuticals,
research, development, registration, clinical trials, and trade will
ultimately be pointless and harmful if the medicines patients take
contain insufficient, incorrect, or dangerous ingredients.
Definitions
There has been considerable debate and confusion over definitions
for the different types of poor-​quality medicines. This has arisen be-
cause of poor science and because of tension between the defence of
commercial interests and the public health importance of enhanced
access to good quality medicines. Here, we use the term falsified to
mean poor-​quality medicines that are deliberately and fraudulently
produced, substandard for those that are poor quality because of un-
intentional but negligent errors in factory production and degraded
for those that left the factory as good quality but deteriorated through
not following Good Pharmacy Practice in the distribution chain. The
distinction is important as the underlying human motivation and regu-
latory actions needed for falsified and substandard medicines differ.
The term falsified is preferred to ‘counterfeit’ as the latter is argued
to be primarily an intellectual property and trade term rather than
concerned with patient outcome and public health. The use of falsi-
fied and substandard terminology was agreed at the World Health
Assembly in 2017.
Substandard medicines usually contain either too little or too
much active pharmaceutical ingredient (API) or are of poor bio-
availability, leading to low blood levels. Falsified medicines often,
but not always, contain no API or the wrong API. The amount of ac-
tive ingredient is not sufficient information to determine accurately
whether a medicine is falsified. Inspection of the packaging, that re-
veals deception for falsified medicines, is usually required.
Epidemiology
Products of both the innovative and generics industry, in the private
and public sectors, for both expensive and inexpensive medicines
2.10
Medicine quality, physicians, and patients
Paul N. Newton
2.10  Medicine quality, physicians, and patients
125
are circulating. There is considerable confusion over the extent of
the problem. Many of the data have been collected using ‘conveni-
ence’ sampling and have often been interpreted and extrapolated un-
critically. In order to be able to estimate, with confidence intervals,
the proportion of the medicine supply in a community or area that
is poor quality and to follow change through time and assess the
impact of interventions, random sampling is required. Random lot
quality assurance sampling may provide a cost-​effective alternative
to estimate whether the prevalence of poor-​quality medicines ex-
ceeds a given threshold.
The available evidence, although it should be viewed with caution,
suggests that there is an important and neglected global problem,
but that it is focal through time and space. Hotspots of poor-​quality
medicines will inevitably lead to focal excess mortality and morbidity.
An epidemic of falsified oral artesunate containing no, or minimal,
artesunate afflicted mainland Southeast Asia until recently and there
remain severe problems with falsified artemisinin combination
therapy across Central and West Africa. The WorldWide Antimalarial
Resistance Network tabulates and maps accessible reports of the
quality of antimalarials (http://​www.wwarn.org/​aqsurveyor).
In the United States, over 700 patients were infected with
Exserohilum rostratum, by being given glucocorticoid injections
for chronic musculoskeletal pain from one compounding phar-
macy, contaminated with this fungus. Sixty-​one patients died. Some
c.500 children died from renal failure after ingesting paracetamol
containing the renal toxin diethylene glycol in south Asia, Africa,
and Central America. In Singapore, 150 non​diabetic patients were
admitted with severe hypoglycaemia because they had taken fal-
sified erectile dysfunction medicines containing sulfonylureas.
Contaminated substandard magnesium sulphate led to Serratia
septicaemia in United States and gentamicin-​resistant Pseudomonas
aeruginosa in gentamicin eye drops in Mauritius led to severe
eye infections, probably due to errors in manufacture rather than
fraud. The national visceral leishmaniasis control programme in
Bangladesh used ‘miltefosine’ that contained no miltefosine.
Vaccines have also been affected by both falsification and sub-
standard production, including falsified rabies and influenza vac-
cines. Expiry date tampering, to falsely lengthen the duration that
medicines and diagnostic tests can be legitimately sold, also occurs.
Diagnostic tests have been falsified, as have surgical implants, such
as cardiac stents, and insecticide-​treated bednets that contained no
insecticide. Dental and surgical equipment and electronic compo-
nents have been falsified, causing patient harm.
Detection
The formal evaluation of the content and dissolution of medicines re-
quires sophisticated equipment and consumables with highly trained
staff and quality assurance/​quality control (QA/​QC). Medicine regu-
latory authority (MRA) laboratories in the wealthy world are likely to
be equipped with mass spectrometry and high-​performance liquid
chromatography equipment with dissolution apparatus, for example.
However, in most low- or middle-income countries such equipment,
along with the necessary technical capacity, consumables, and QA/​
QC, are not available. In 2018 in Southeast Asia and Africa there are
only 5 and 7 countries, respectively, with World Health Organization
(WHO) prequalified laboratories for medicine quality analysis.
Forensic techniques such as pollen analysis are rarely available.
Over the last decade a plethora of portable devices have been in-
vented and developed for the rapid evaluation of medicine quality in
the field. These are based on principles such as Raman, near infrared,
and image analysis. However, they have not yet been independently
compared for diagnostic accuracy and cost-​effectiveness, so the jury
is still out. However, the hope is that these will empower medicine in-
spectors and pharmacists to screen medicine quality in the distribu-
tion chain and objectively select which samples should be escalated
for formal analysis, rather than just relying on eyes and experience.
Impact
Poor-​quality medicines can lead to serious individual health con-
sequences, such as treatment failure due to suboptimal dosing, ad-
verse reactions, and increased morbidity and mortality. They can
undermine public confidence in health programmes and waste
scarce resources. Patients might also suffer from surprising ad-
verse effects from unexpected active ingredients. For example, fal-
sified co-​trimoxazole containing diazepam, metronidazole made
of chloroquine, diazepam containing haloperidol, oseltamivir con-
taining vitamin C, ceftazidime made of reused vials containing
streptomycin and ofloxacin made of penicillin have been described.
Therefore, it is crucial to check a patient’s medication when faced
with unexpected clinical syndromes.
The world clearly has a problem, but the extent is uncertain, in
part because the collation of such publicly accessible data has only
recently started. There are clear examples where patients have been
harmed by poor-​quality medicines. However, objective estimates of
impact on patient outcome have been difficult. Logically, if a medi-
cine has been shown to reduce mortality by 30% but the medicine
taken has zero efficacy (because it contains no API), mortality will
not be reduced by that 30%. In order to reliably conclude that a poor-​
quality medicine has harmed, evidence is needed that (a) a patient
has suffered demonstrable harm, (b) proof that the patient took a
particular medicine and (c)  chemical analysis of that medicine
demonstrating that it does not contain what it is stated to contain (or
contains toxins). In much of the world, such linked-​up evidence has
been difficult to obtain.
The first evidence-​based estimate from modelling of the excess
mortality suggested 122 350 under-​five malaria deaths per year were
associated with consumption of poor-​quality antimalarials, repre-
senting 3.75% of all under-​five deaths in a sample of 39 countries.
This is a first approximation and will be refined as the objective evi-
dence improves.
Poor-​quality medicines also have wider consequences on public
health with loss of confidence in healthcare, spread of drug resistant
pathogens, economic losses for producers and traders in the genuine
articles, and increased burden for health workers.
Modelling strongly suggests that underdosing is an important
contributor to antimicrobial resistance. Therefore, if patients con-
sume medicines containing low %API (usually substandard medi-
cines) that result in blood levels within the window of selection, the
risks of engendering resistance are high. Medicines containing none
of the stated API (usually falsified medicines) may also contribute to
drug resistance for some pathogens, such as malaria if they result in
hyperparasitaemia and high gametocyte burdens.
126
section 2  Background to medicine
Clinical trials have naively been regarded as immune from the
issues of poor-​quality medicines and medical devices but there is
growing realization that their quality should be checked and docu-
mented or trials risk wrongly informing policy.
Interventions
For the individual health worker and patient—​caveat emptor is vital.
Patients may notice when their medications change in appearance,
taste, or side effects. Greater awareness of the problem and suspicion
in the face of puzzling clinical syndromes, treatment failure, and
medicines that seem surprisingly inexpensive are needed.
The problems of poor-​quality medicines cannot be viewed in iso-
lation. They are enmeshed with many other complex health system
problems, especially the affordability and accessibility of medi-
cines and the often-​limited capacity of MRAs in L/​MICs. MRAs are
the keystones for important interventions but only 20% of WHO
member states are reported to have well-​developed drug regulation
and 30% have ‘either no drug regulation or a capacity that hardly
functions’. Strengthening MRAs, improving the oversight and QA/​
QC of medicine manufacturing collaboration with other key stake-
holders such as procuring agencies, and facilitating patient access to
affordable medicines are likely to be key factors in improving quality.
A key neglected issue has been the woeful sharing of informa-
tion between and within countries. There is no global system for
the mandatory reporting, assessment, and dissemination of infor-
mation on suspicious medicines. The more functional MRAs issue
alerts and the revitalized WHO’s new Rapid Alert System (http://​
www.who.int/​medicines/​publications/​drugalerts/​en/​) facilitates in-
formation sharing on poor-​quality medicines between MRAs. The
e-​drug (http://​lists.healthnet.org/​mailman/​listinfo/​e-​drug) and e-​
med (http://​lists.healthnet.org/​mailman/​listinfo/​e-​med) systems re-
port medicine quality problems in English and French, respectively.
In most of the world there is no legal duty for health workers, aca-
demics, or the pharmaceutical industry to report suspicions of poor-​
quality medicines to the national MRA. There is clearly an ethical
duty to do so but this needs to be become a legal duty. The Access to
Medicine Index (http://​www.accesstomedicineindex.org/​) includes
monitoring of the reporting of falsified medicines by the pharma-
ceutical industry to WHO and MRAs.
Areas of uncertainty and controversy
There are many other related important global issues that need to
be addressed, such as the lack of an international coding system for
identifying pills and capsules, poor-​quality instruction sheets with
medicines in small fonts and in user-​unfriendly language.
There is an urgent need for clear understanding that generic
medicines per se are not counterfeit, and a need for an international
convention making medicine falsification an international, extra-
ditable, crime. We need better understanding as to how to engage
with societies about medicine quality problems so that they do not
inappropriately stop good quality medicines in the face of warnings
of poor-​quality products. More evidence is needed to understand
how best to empower drug inspectors with the optimal new portable
devices. It will be key to have an international capacity building and
financial mechanism to support MRAs that lack resources to pro-
tect the medicine supply and better objective-​collated data to inform
policy.
FURTHER READING
Attaran A, et al. (2012). How to achieve international action on falsified
and substandard medicines: a consensus statement. BMJ, 345, e7381.
Caudron JM, et al. (2008). Substandard medicines in resource-​poor
settings: a problem that can no longer be ignored. Trop Med Int Hlth,
13, 1062–​72.
Fernandez FM, et al. (2011). Poor quality drugs: grand challenges in
high throughput detection, countrywide sampling, and forensics
in developing countries. Analyst, 136(15), 3073–​82.
Godlee F (2018). Why aren't medical devices regulated like drugs?
BMJ, 363, k5032.
Newton PN, et al. (2006). Counterfeit anti-​infective medicines. Lancet
Infect Dis, 6, 602–​13.
Newton PN, et al. (2014). Falsified medicines in Africa and public
health—​‘No Action—​Talk Only’. Lancet Global Health, 2, e509–​10.
Vickers S, et al. (2018). Field detection devices for medicines quality
screening: a systematic review. BMJ Global Health, 3, e000725.
World Health Organization (2017). WHO Global Surveillance and
Monitoring System for substandard and falsified medical prod-
ucts. World Health Organization: Geneva. http://apps.who.int/
medicinedocs/en/m/abstract/Js23373en/

2.11 Preventive medicine 127
2.11 Preventive medicine 127
ESSENTIALS
Causes of death
Most deaths before age 80  years are preventable but preventive
medicine cannot offer immortality.
Childhood and early adult life
Deaths from infectious diseases and trauma usually reflect poverty and
political instability. Prevention requires political action to reduce the
risk of war and improve the supply of food, clean water, sanitation, and
shelter. Preventive medicine can help by controlling spread of infection
through vaccination, health education, control of insect vectors, and
treatment of disease carriers to prevent onward transmission.
Middle age
The commonest cause of premature death globally is vascular
disease—​mainly heart attacks and stroke. The main causes are poor
diet, obesity, and tobacco smoking. Political action is important to
make it easy for people to take exercise and eat healthily, and to make
it difficult to buy and smoke tobacco.
Preventive medicine
Preventive medicine can identify and treat people at increased risk
of death from vascular disease (particularly those with diabetes, high
blood pressure, and high blood lipids) and will save many lives by
doing this effectively, but the need for medicines to treat vascular
disease in individuals is a measure of public health failure. The co-
existence of obesity and starvation as major causes of preventable
mortality in many countries is a growing public health challenge.
Many effective preventive interventions such as legislation to make
seatbelts compulsory or tax tobacco should be targeted at the whole
population, whereas preventive medicine provided by clinicians must
target individuals. This often requires screening to detect early signs
of disease (e.g. HIV infection, cancer) or markers of risk of disease
(e.g. high blood pressure, intrauterine growth delay), but clinicians
must be clear that something effective can be done to ameliorate the
condition detected before any screening is undertaken.
Failure of evidence-​based preventive interventions
The usual reason for failure is lack of effective implementation, with
the three most important issues being: (1) poor population coverage,
meaning that only a small proportion of the at-​risk population re-
ceive the intervention; (2) inadequate staff training; and (3) inad-
equate quality control.
Preventive medicine and curative medicine
Preventive medicine is an important and integral part of good
curative medicine. All doctors have a responsibility to think about
why someone is ill. Whatever disease is diagnosed, the question of
whether it could have been prevented, and whether the risk of pro-
gression can be reduced, must be addressed. For example, every
clinician who diagnoses a stroke must ask themselves whether a pre-
vious clinical opportunity to measure and control blood pressure has
been missed, and reflect on this in regard to their future practice re-
lating to other patients.
Introduction
In his millennium address, Nelson Mandela reminded the world
that ‘we close the century with most people still languishing in
poverty, subjected to hunger, preventable disease, illiteracy, and
insufficient shelter’. The health gap between rich and poor nations
remains shameful. For example, overall life expectancy in Sierra
Leone is currently estimated as 46 years compared to 81 years in
the United Kingdom. But even in the economically developed
world, many people still die prematurely. In 2013, approximately
23% (114 740 out of 506 790) of all deaths registered in England
and Wales were from causes considered avoidable through good-​
quality healthcare or wider public health interventions. Again,
there is a marked gap between rich and poor. Men in the most
advantaged areas of England lived on average 9 years longer, and
experienced 19 more years of good health, than men living in
the least advantaged areas. It is naive to think that medicine will
remedy this situation. The fundamental step in achieving good
health remains the elimination of poverty, with consequent access
to food, sanitation, education, and shelter. The power of medicine
lies in the scientific understanding it provides of the disease pro-
cess. Preventive medicine uses this understanding both to try to
reduce the risk of disease and to detect and treat appropriately
emergent disease before it does damage.
2.11
Preventive medicine
David Mant
128
section 2  Background to medicine
What is the scope for prevention? Fig. 2.11.1 shows the number
expected to die at different ages if 10 000 men were subject to the
age-​specific death rates in the United Kingdom in 2010 compared to
1870. The dramatic fall in deaths during childhood and early adult-
hood demonstrates unequivocally that such deaths are preventable.
Moreover, most deaths before age 80 in high-​income countries
are from diseases which are amenable to preventive intervention.
Migrant studies show that genetic advantage does not explain the
high average life expectancy in Japan (81 years in men and 87 years
in women). It therefore seems reasonable to expect that effective pre-
ventive medicine can make death before age 80 years uncommon.
However, the proportion of men or women who can expect to live
over 100 years remains very low (1.1% in England in 2011/​12). The
achievable objective therefore seems to be better quality of life before
death, and avoidance of premature death, rather than extension of
life beyond 100 years. Preventive medicine cannot offer immortality.
Preventive strategies
Identifying and reducing risk
The main difference between preventive and curative medicine is the
focus on risk. The leading global risk factors for preventable mor-
tality in 2013, estimated by the Global Burden of Disease study, are
shown in Table 2.11.1. The most disturbing fact remains the occur-
rence of both starvation and obesity on the same list. Although there
has been progress in some areas since 1990 (e.g. the number of at-
tributable deaths due to child starvation, unclean water, and unsafe
sex have more than halved), other issues such as deaths due to air
pollution, substance abuse (including alcohol and tobacco), and oc-
cupational hazards have increased. Although there has been a shift
from environmental towards individually mediated risks, individual
behaviour is always constrained by sociocultural and economic fac-
tors largely outside individual control.
On a global scale, the dietary and behavioural risk factors leading
to chronic disease now predominate, whether risk is measured in
terms of deaths or disability. The geographical differences in risk
factors for ill health mainly reflect national wealth. The individual
behaviours which promote hypertension and cardiovascular disease
(eating an unhealthy diet, smoking tobacco, taking too little exer-
cise) are now the most important risk factors in middle-​income as
well as high-​income countries. However, Table 2.11.2 shows that
lack of food, unsafe water, and unsafe sexual practices still remain
the top three causes of years of healthy life lost in the low-​income
countries of sub-​Saharan Africa.
The prevention paradox
Preventive medicine aims to reduce the risk of disease (or the risk
of further morbidity and mortality in those who develop disease)
so its benefit is the absence of disease in the future rather than the
present. Absence of something is a difficult benefit to champion,
particularly to the individual. As Geoffrey Rose pointed out many
years ago, not only is the benefit intangible but many people must
take precautions in order to prevent illness in only a few. Even in
a country where diphtheria is common, several hundred children
must be immunized to prevent one death. Rose called this the ‘pre-
vention paradox’—​a preventive measure, which brings large benefits
to the community may offer little to each participating individual.
0
50
100
150
200
250
300
Number of deaths
Age at death
Age at death of 10 000 people born in the UK experiencing the age-
specific death rates of 1870 and 2010
0–5
6–10
11–15
16–20
21–24
25–30
31–35
36–40
41–45
46–50
51–55
56–60
61–65
65–70
71–75
76–80
81–85
86–90
91–95
96–100
100–105
105–110
1870
2010
Fig. 2.11.1  Numbers dying at different ages if 10 000 men were subject from birth to the UK
mortality rates in 1870 and 2010.
Source of data: Office of National Statistics Life Tables, United Kingdom; the figure is based on one originally
drawn by Doll R in British Medical Journal 1982; 286: 445–​53.
2.11  Preventive medicine
129
The risk paradox
The importance of the risk paradox is best illustrated by an example.
One of the leading global risk factors for vascular disease is a high
level of cholesterol in the blood. Fig. 2.11.2 shows historical data
from the United Kingdom on the prevalence of high blood chol-
esterol, the death rate associated with each cholesterol level, and
the proportion of all deaths attributable to cholesterol occurring
at each level. The risk paradox is that although those with a blood
cholesterol greater than 7.5 mmol/​litre are at highest individual risk
of disease, they account for only 8% of total deaths. The group of
people in whom most deaths occur (22%) is that with only a mod-
estly increased cholesterol level of 5.5–​6.0 mmol/​litre. This is simply
because of the number of people involved. There are far fewer in
the high-​risk group than in the moderate risk group. Targeting
Table 2.11.1  Leading risk factors for global mortality in 2013
Risk factor
Attributable deaths
each year (millions)
Comment
Environmental
Ambient air pollution
3.1
Particulates 2.9 m; ozone pollution 0.2 m; overall 33% increase in attributable deaths since 1990
Indoor smoke from solid fuels
2.9
No reduction since 1990
Unsafe water, sanitation, and
handwashing
1.3
The most important of these issues remains unsafe water (1.2 m attributable deaths) but the number
of attributable deaths from this cause has almost halved since 1990
Occupational risks
0.7
Top three issues: 1. air quality (gases/​fumes/​dust); 2. asbestos; 3. industrial injuries. Overall 27%
increase in attributable deaths since 1990
Dietary
Unhealthy diet
11.3
Top three issues: 1. High sodium; 2. Lack of fruit; 3. Lack of whole grain cereals
Child and maternal malnutrition
1.7
Top three issues: 1. undernutrition (starvation); 2. suboptimal breastfeeding; 3. iron deficiency
Obesity (high BMI)
4.4
Overeating is compounded by low physical activity (attributable deaths 2.2 m)
Other behavioural
Tobacco smoke
6.1
Smoking 5.8 m, passive smoking 0.3 m; overall 18% increase in attributable deaths from tobacco use
since 1990
Alcohol and drug use
3.2
Alcohol 2.8 m, other drugs 0.4 m; overall 51% increase in attributable deaths since 1990
Unsafe sex and sexual abuse
1.8
Unsafe sex 1.5 m, abuse of partner or child 0.3 m; deaths from unsafe sex more than halved since 1990
Metabolic
High systolic blood pressure
10.3
Main modifiable risk factors for high blood pressure are dietary (especially salt intake)
High blood cholesterol
4.4
Mean cholesterol levels vary between WHO Regions, but not more than 2.0 mmol/​litre in any age
group
High fasting blood glucose
4.0
The main risk factor for diabetes is obesity
Data from: Global Burden of Disease Study 2013. Published in The Lancet online 11 September 2015 (http://​dx.doi.org/​10.1016/​50140-​6736(15)00128-​2).
Table 2.11.2  Top 10 risk factors for ill health (disability-​adjusted
life years attributable to level 2 risk factors) in 2013 in sub-​Saharan
Africa and globally
Rank
Sub-​Saharan Africa
Global
1
Child and maternal nutrition
Dietary risks
2
Unsafe sex
High systolic blood pressure
3
Unsafe water sanitation and
hand washing
Child and maternal nutrition
4
Air pollution
Tobacco smoke
5
Alcohol and drug use
Air pollution
6
High systolic blood pressure
High body mass index
7
Dietary risks
Alcohol and drug use
8
High body mass index
High fasting blood glucose
9
High fasting blood glucose
Unsafe water, sanitation, and
hand washing
10
Tobacco smoke
Unsafe sex
Data from: Global Burden of Disease Study 2013. Published in The Lancet online 11
September 2015 (http://​dx.doi.org/​10.1016/​50140-​6736(15)00128-​2).
CHD deaths/1000/6 year
Prevalence
20
10
0
4
5
6
7
8
4%
8%
17%
22%
19%
13%
9%
8%
0
10
20
Serum cholesterol (mmol)
Fig. 2.11.2  Proportion of coronary heart disease deaths attributable
to raised serum cholesterol (percentages above columns). Columns and
right axis show population distribution of cholesterol levels. The broken
line and left axis show the attributable mortality.
Reproduced from Rose G. Rose’s Strategy of Preventive Medicine, 2008. With
permission from Oxford University Press, Oxford.
130
section 2  Background to medicine
preventive medicine at just the high-​risk group will often have rela-
tively little impact on the total number of deaths in the population.
Primary and secondary prevention
Use of these two terms is sometimes inconsistent (e.g. antenatal care
is usually characterized as secondary prevention although preg-
nancy is not a disease) but usually they can be interpreted as follows:
•	Primary prevention—​interventions to reduce the risk of disease
in healthy people (e.g. use of seat belts to prevent injury in car
accidents; tobacco control to prevent the occurrence of smoking-​
related disease; immunization against infectious disease).
•	Secondary prevention—​interventions to prevent avoidable mor-
bidity in people with disease (e.g. treatment of vascular disease
with aspirin; screening for early cancer).
Primary prevention could be perceived as always better than sec-
ondary prevention in the same way that prevention is better than
cure. However, this is oversimplistic for three reasons: (1) many ap-
parently healthy people will have undetected disease; (2) some inter-
ventions fall into both categories (e.g. stopping smoking reduces the
progression as well as onset of many smoking-​related diseases);
(3) primary prevention may be less effective than secondary preven-
tion, particularly when the intervention is aimed at the individual.
Strategic choices
In choosing priority strategies for risk prevention, the World Health
Organization (WHO) has recommended that in general it is more
effective:
•	to focus on population-​based rather than individual interventions
•	to prioritize primary rather than secondary prevention
•	to control distal before proximal risks to health
This advice makes good sense from a public health perspective—​it is
self-​evidently better to prioritize provision of clean water before fo-
cusing on early recognition and treatment of dehydration from diar-
rhoea caused by dirty water. However, making strategic choices about
preventive medicine aimed at individual patients is less straightfor-
ward. It is much easier to persuade individuals to change their behav-
iour when they perceive themselves at high risk (e.g. to convince them
to stop them smoking after they develop a smoking-​related illness).
Primary prevention is seldom delivered cost-​effectively to healthy
people by doctors or nurses in routine clinical practice. The one ob-
vious exception is vaccination, where involvement of clinicians is
often important in delivering vaccines and ensuring high coverage
of the population at risk. Clinicians are also in an ideal position to re-
inforce some key public health messages, such as stopping smoking
and hand washing, with brief opportunistic advice during routine
consultations (see next). However, several large-​scale clinical trials
have shown that it is not cost-​effective for doctors and nurses to ex-
pend scarce resources trying to convince otherwise healthy individ-
uals to change eating or exercise habits determined primarily by the
cultural and socioeconomic context in which they live.
Defining and identifying the at-​risk population
Public and individual intervention
Some public health interventions do not require anything other
than a broad geographical definition of risk. Interventions are
targeted at whole populations and do not require identification of
individuals within that population—​for example, road accidents
can be reduced by seatbelt legislation and tobacco consumption
by taxation without identifying the individual driver or smoker.
However, some primary preventive strategies (e.g. vaccination)
and most examples of secondary prevention (e.g. screening) have
to be delivered to specific individuals at risk. You not only need to
know that people with high blood pressure are at risk of stroke,
you need to know who has high blood pressure. These individuals
can be defined in one of three ways: by demographic, phenotypic,
or familial characteristics. Within each group of at-​risk people so
defined, further subpopulations may be identifiable as at particu-
larly high risk.
Demographic risk
This is the most common way to define the target group for pre-
ventive medicine services. For example, the United States Preventive
Task Force defines the target group for most of the preventive
services it recommends in terms of age. The other demographic risk
characteristic commonly used by the Task Force to define the target
population is gender. It is obvious why it recommends that breast
and cervical cancer screening should only be targeted at women,
but gender-​specific recommendations are also made for aortic an-
eurysm (based on differential risk) and gonorrhoea (based on the
differential likelihood of asymptomatic infection). Some preventive
programmes may also target specific ethnic or racial groups—​for
example, in Australia, interventions to reduce the risk of rheumatic
fever associated with group A streptococcal infection of the throat
have been restricted to the indigenous Aboriginal population who
are at exceptionally high risk.
Phenotypic risk
A phenotype is a set of observable characteristics of an individual
or group. Many epidemiological risk factors for disease are phys-
ical characteristics (e.g. obesity, hyperlipidaemia); other phenotypic
categories often used to define at-​risk target populations are behav-
iour (e.g. smoking, driving) and disease states (diabetes mellitus, an-
gina). As some phenotypic risks interact (e.g. smoking and exposure
to asbestos interactively increase risk of mesothelioma), multiple
risk assessment is an increasingly common practice. Several clin-
ical tools have been developed to help estimate multifactorial risk in
everyday practice, such as the New Zealand risk charts for cardio-
vascular disease.
Familial risk
Recent advances in genetic technology have increased our ability to
characterize familial risk accurately, and further advance is likely.
At present, most preventive medicine programmes in this area use
genetic assessment to refine assessment of individual risk in pheno-
typically identified high-​risk individuals or families (e.g. cystic
fibrosis, neurofibromatosis) or demographically defined popula-
tions (e.g. pregnant women at risk of giving birth to a child with
Down’s syndrome). However, the characterization of risk based on
population-​based genetic screening is already technically feasible in
the economically developed world, increasing the potential power of
preventive medicine but also raising important ethical issues about
how society deals with accurate predictions of high disease risk, par-
ticularly when evident at or before birth.
2.11  Preventive medicine
131
Registration, screening, and case-​finding
It is much easier to identify specific individuals at risk when uni-
versal health registration is in place and its accuracy is systematically
maintained. It not only allows efficient provision of primary preven-
tion strategies, such as vaccination, but screening for disease risk
is also much more efficient when based on an accurate population
register. In the absence of a population health register, it is neces-
sary to rely on case-​finding. This requires identification of at-​risk
individuals during routine clinical work (normally in clinical con-
sultations, but sometimes through contact or family tracing). It is
less efficient than systematic population screening, but sometimes
provides better access to socially disadvantaged groups who may re-
spond poorly to screening invitations or have no registered address.
It also allows some interventions to be given at a particularly appro-
priate moment (e.g. smoking cessation advice at a consultation for
cough or contraceptive advice after termination of pregnancy).
Interventions to modify risk
The importance of public health
The marked improvements in health which have been achieved in
economically developed countries over the past 150 years are not
attributable to medicine. Life expectancy has doubled mainly be-
cause of environmental control of infectious pathogens (through
sanitation and control of insect vectors) and a lifestyle that reduces
individual susceptibility to infectious disease (better food, shelter,
and education). So although medical science can play an important
role in guiding public health policy by improving understanding of
the mechanisms of disease, and specific medical interventions allow
us to treat disease when it occurs, the role of preventive medicine
should not be overestimated. In particular, the medical profession
should not take upon itself responsibilities for public health which
are more appropriately assumed by governments and other social
and environmental agencies.
The preventive responsibilities of all doctors
Preventive medicine is an important and integral part of good cura-
tive medicine. All doctors have a responsibility to think about why
someone is ill. Whatever cause is identified (physiological, social,
or psychological), the question about whether the cause can be pre-
vented (and the risk of future disease reduced) should be addressed.
Clinicians should be held professionally accountable if they can be
shown to have missed a previous clinical opportunity to measure
blood pressure in an individual who subsequently develops a stroke.
Doctors who work in a primary care role (particularly those with a
registered population) have the added responsibility to ask them-
selves whether the risk should be addressed at a population rather
than just an individual patient level.
Changing behaviour
Most of the leading risk factors for preventable mortality cited
earlier can be influenced by individual behaviour: smoking, diet,
exercise, alcohol consumption, hand washing, and unsafe sex. The
most effective way to influence such behaviour is usually through
public health policy, but individual practitioners can play an im-
portant complementary role. People do listen to their doctors, and
several clinical trials have shown brief advice on behaviour modi-
fication to be cost-​effective, even though the impact may be small
(e.g. in most studies only about 1 in 30 smokers given brief advice
to stop smoking actually quit). Brief advice is most effective if prac-
tical in nature (giving guidance on how change can be achieved) and
if backed up by written advice to take home (see Chapters 26.6.1,
26.6.2, and 26.6.3 for further discussion). More intensive inter-
ventions may sometimes be more effective but tend to be less cost-​
effective. In each case, it is important to take account of the scientific
evidence about effectiveness and the local socioeconomic context.
Immunization
Vaccination is a very effective preventive strategy. Vaccination
against smallpox has led to global eradication of the disease; eradi-
cation of polio is a feasible global objective. Vaccination against
many diseases, particularly diseases of childhood such as measles,
diphtheria, and polio, has led to rapid and dramatic falls in disease
incidence. Fig. 2.11.3 shows the impact of introduction of the Hib
vaccine in 1992 on the incidence of Haemophilus influenzae infec-
tion in England and Wales.
Table 2.11.3 shows the current routine vaccine schedule in the
United Kingdom. Every individual in the population is routinely
vaccinated against 14 different organisms. At-​risk groups may be re-
quired to have specific vaccines to protect themselves (e.g. health
workers against hepatitis B), and the general population and trav-
ellers to other countries can choose and pay for a range of vaccines
for other diseases including hepatitis A, typhoid, and rabies. It is not
uncommon for a United Kingdom resident to have received vaccin-
ation against 15–​20 microorganisms during their lifetime.
Several new and important vaccines are on the horizon (e.g. against
malaria), but the existence of an effective vaccine does not guarantee
the success of an immunization programme. This depends on the ef-
fective delivery of the vaccine to the at-​risk population. Programmes
are often limited in their effect by affordability (some vaccines are
unaffordable without external funding in many low-​income coun-
tries), acceptance (parental anxiety about the adverse effects often
limits uptake), and delivery (vaccines may lose potency if stored
outside a refrigerator). There are also potential problems with the
Number of reports
250
200
150
100
50
0
Years
Hib vaccine introduced
1995
1994
1993
1992
1991
1990
1989
up to 1 year of age
all other ages
Fig. 2.11.3  Effect of introduction of Haemophilus influenzae type b
(Hib) vaccination in the United Kingdom on laboratory reports of Hib
infection.
Reproduced by permission of the Controller, Her Majesty’s Stationery Office.
132
section 2  Background to medicine
antigenic variability of organisms over time (e.g. influenza) and the
difficulty of immunizing at an age young enough to prevent mor-
bidity but old enough to stimulate an immune response (e.g. mea-
sles). Nevertheless, immunization is probably the most important
medical contribution to primary disease prevention.
Screening
The issue of screening is dealt with in Chapter 2.12. Two-​thirds of the
preventive services for adults recommended for implementation by
the United States Preventive Services Task Force involve screening.
The purpose of screening is to identify disease at an early and curable
stage. The most important criterion that has to be met for screening
to be ethical is that the condition identified can be treated. Good
intention is not enough. Screening will do harm if it identifies con-
ditions that we can do nothing about, either because of lack of ef-
fective interventions or lack of resources. It is particularly important
to assess the effectiveness of screening interventions in randomized
trials, as people whose disease was detected by screening will appear
to clinicians to do well even if the screening is ineffective. This is
because early diagnosis will lead to longer survival irrespective of
treatment, and illness detected by screening will always tend to be
more benign in its natural history than illness detected clinically.
Prophylactic treatment
Although most people think of medicines as cures for current
illness, many medicines are prescribed with a view to preventing
future illness. Antibiotics are given before surgery to prevent
postoperative infection, antimalarials to prevent malaria in travel-
lers, anticoagulants to prevent stroke in people with atrial fibrilla-
tion, and lipid-​lowering agents to prevent heart attacks in people at
high risk of cardiovascular disease. The duration of treatment may
also be extended beyond the initial treatment phase to achieve a pre-
ventive effect. Antidepressants are continued after cure to prevent
relapse, angiotensin converting enzyme (ACE) inhibitors to prevent
worsening of ventricular dysfunction, and uricosuric agents to pre-
vent further episodes of gout.
It must be clear from these examples that many drugs have the po-
tential to be used for prevention as well as cure. In some cases (e.g.
treatment of diabetes) the distinction between prevention and cure
is unhelpful: treatment aims to prevent morbidity in both the short
and long terms. However, in all the examples given, prescribing is
limited to a defined high-​risk group. Prophylactic treatment with
drugs is less helpful when a high-​risk population cannot be easily
defined. It is usually inappropriate to use prophylactic treatment to
reduce population risk for three reasons: the strategy is seldom cost-​
effective, increasing the reliance of the population on medicine is an
adverse social outcome, and uncommon adverse effects can easily
outweigh any benefit. However, a case can be made for treating high-​
risk populations defined only by age and gender with a ‘polypill’
combining certain drugs with a low side-​effect profile that have been
shown to be effective in preventing cardiovascular disease.
Environmental change
Most environmental causes of disease are best modified on a public
health rather than an individual basis. Such factors include the safety
of the workplace, environmental pollution, transport safety, food
hygiene, and provision of clean water. However, some diseases have
environmental causes, which need to be recognized and avoided by
the individual patient. On a global scale, avoidance of insect and
other disease vectors (e.g. by netting) and attention to nutritional
hygiene (e.g. by hand washing and filtering water) are probably the
most important. In economically developed countries, the most
common diseases amenable to individual environmental interven-
tion are those associated with atopy, such as asthma and eczema.
Not all patients have an identifiable allergenic cause for their symp-
toms and, even if one is identified, avoidance (e.g. of house dust mite
in asthma) may not be easy. But dramatic improvement can occur,
and treating contact dermatitis without giving advice on contact
Table 2.11.3  Routine immunization programme in the United Kingdom according to age
2 months
3 months
4 months
1 year
2 years, then
annually until
16 years
3 ½ years
Girls
12–​14 years
14 years
65 years
65 years
then
annually
70 years
Diphtheria
✔
✔
✔
✔
✔
Tetanus
✔
✔
✔
✔
✔
Polio
✔
✔
✔
✔
✔
Pertussis
✔
✔
✔
✔
H. influenzae B
✔
✔
✔
✔
Rotavirus
✔
✔
Pneumococcus
✔
✔
✔
✔
Meningococcus C
✔
✔
Measles
✔
✔
Mumps
✔
✔
Rubella
✔
✔
Influenza
✔
✔
Human
Papillomavirus
✔
Shingles
✔
Data from: Department of Health/​Public Health England. Immunisation against infectious disease (The Green Book) Chapter 11. HMSO, London (updated February 2015). https://​
www.gov.uk/​government/​uploads/​system/​uploads/​attachment_​data/​file/​400554/​2902222_​Green_​Book_​Chapter_​11_​v2_​4.pdf
2.11  Preventive medicine
133
avoidance, or treating louse bites without giving advice on how to
rid clothes of lice, is bad medical practice.
What interventions work?
Public health interventions
It is impossible to list here all the public health initiatives that are
known to be effective. There is no doubt that provision of clean
water and sanitation; avoidance of war; progressive taxation; pro-
vision of education; fiscal policy to reduce tobacco and alcohol
use; food policies to reduce community intake of salt, saturated fat,
and excess calories; and education to promote safe sex, are all ef-
fective public health interventions. Their effective implementation
depends on political will at a national and international level. In
general, the effectiveness of public health interventions will reflect
government attitude to regulation (e.g. whether the presumption of
market-​led development and free trade dominates individual and
societal health concerns) and the understanding of health risk by
the general public, politicians, and health practitioners. At an inter-
national level, the World Health Organization cites the Framework
Convention for Tobacco Control as an exemplar of a very effective
government-​led international public health initiative. This covers
advertising, regulation, taxation, and smoke-​free zones, as well as
the individual treatment of addiction.
Interventions aimed at the individual
As with public health interventions, it is impossible to list here all
the preventive interventions targeted at individuals which have
been shown to be effective. Many are in any case better seen as
part of good routine clinical care (and are included in the relevant
chapters on specific diseases). Nevertheless, one accessible source
of regularly updated evidence on individual interventions is the
United States Preventive Services Task Force and Table 2.11.4 lists
Table 2.11.4  Preventive interventions recommended for adults by the United States Preventive Services Task Force as providing important
health benefit on the basis of research evidence rated as A (high certainty of substantial net benefit) or B (high certainty of moderate net
benefit or moderate certainty of moderate to substantial net benefit)
Men
Women
Timing and target groups
Screening interventions
Alcohol misuse screening and
counselling
✔
✔
Counselling about safe-​drinking for all; specific help for heavy drinkers identified by screening
Aortic aneurysm
✔
Ultrasound at age 65–​75 in ever smokers
BRCA-​related cancer
✔
Women with family history suggestive of BRCA1 or BRCA2 gene mutations
Breast cancer
✔
Biennial mammography from ages 50 to 74 years
Cervical cancer
✔
Cytology every 3 years age 21–​65 years or combined cytology/​HPV testing every 5 years
Chlamydia infection
✔
All sexually active women age <25 years (after age 25 only high-​risk groups)
Colorectal cancer
✔
✔
Age 50–​75 years using faecal occult blood (FOB) testing, sigmoidoscopy, or colonoscopy
Depression
✔
✔
Recommendation states that screening is appropriate only when accurate diagnosis, treatment,
and follow-​up possible
Diabetes mellitus
✔
✔
Adults with sustained hypertension (BP >135/​80)
Risk of falling in older adults
✔
✔
Community-​dwelling adults 65+ years; physical therapy and/​or vitamin D supplements
recommended for high-​risk groups
Gonorrhoea infection
✔
At-​risk groups
Hepatitis C infection
✔
✔
At-​risk groups and those born between 1945 and 1965
High blood pressure
✔
✔
All adults age 18+ years
HIV infection
✔
✔
At-​risk groups
Intimate partner violence
✔
Women of childbearing age (guideline also includes advice on identifying risk of abuse of elderly)
Lipid disorders
✔
✔
All men aged 35+; women and younger men 20+ if high cardiovascular disease (CVD) risk
Lung cancer
✔
✔
Heavy smokers (current or within 15 years of cessation)
Obesity
✔
✔
Assuming availability of intensive multifactorial behavioural intervention when BMI 30+
Osteoporosis
✔
Women aged 65+ (plus women at equivalent risk for reasons other than age)
Syphilis infection
✔
✔
At-​risk groups
Other interventions
Aspirin
✔
✔
Adults at increased CVD risk
Breast cancer chemoprevention
✔
Women age 35+ years at high risk of breast cancer
Dietary counselling
✔
✔
Adults at high risk of CVD
Tobacco use counselling
✔
✔
All who use tobacco (assuming all patients are asked about tobacco use opportunistically at clinical
consultations)
Data from: Recommendations of the U.S. Preventive Services Task Force. US Department of Health and Human Services, Washington, DC; October 2015 (http://​www.
uspreventiveservicestaskforce.org/​Page/​Name/​uspstf-​a-​and-​b-​recommendations/​Individual interventions).
134
section 2  Background to medicine
the conditions for which evidence of effectiveness is rated by them
as A (high certainty of substantial net benefit) or B (high certainty
of moderate net benefit or moderate certainty of moderate to sub-
stantial net benefit). In interpreting this evidence, it is important
to recognize that the recommendations reflect the specific context
of the United States; differences in the prevalence of the target
condition and the availability and effectiveness of interventions
to ameliorate risk identified impact on their relevance to other
countries.
The recommended preventive interventions (other than vac-
cination) for two important at-​risk groups—​pregnant women
and children—​are listed separately in Table 2.11.5. Again, in
interpreting this evidence it is important to consider its local ap-
plicability. For example, the appropriateness of treating individual
children with fluoride to prevent dental caries depends both on
whether the local diet promotes caries and whether fluoride is
added to drinking water as a public health measure. Similarly,
screening for sickle cell disease and counselling about skin cancer
prevention is only appropriate in at-​risk ethnic groups, and
screening for depression in adolescents is only appropriate if re-
sources are available to make an accurate diagnosis and manage
the problem effectively.
Implementation issues
Cultural constraints
Most behaviour aimed at preventing disease has a strong sociocul-
tural component and reflects prevalent attitudes and norms in so-
ciety. Preventive interventions are severely constrained by this social
context. Convincing people to stop smoking, eat less salt, drink less
beer, or drive more slowly is difficult if everyone else is doing the op-
posite. For example, the data in Fig. 2.11.4 (showing that the mean
blood cholesterol level in Finland in the 1990s was almost twice
that in Japan), supported by migrant studies showing that this dif-
ference was dietary rather than genetic in origin, led some enthu-
siastic policy makers to assume that medical advice to individuals
to modify their diet could cut cholesterol levels by half. However,
even in clinical trials, dietary advice from health professionals in a
community setting has seldom achieved a reduction in blood chol-
esterol of more than 3–​5%. Studies of salt restriction (to lower blood
pressure) show a similar result. Intensive intervention and support
are needed for an individual patient to achieve a physiologically
significant reduction in intake and many find such a diet unpal-
atable. Countercultural change is difficult to achieve. The adverse
Table 2.11.5  Preventive interventions for pregnant women and children (other than vaccination) recommended by the US Preventive
Services Task Force as providing important health benefit on the basis of research evidence rated as A (high certainty of substantial net
benefit) or B (high certainty of moderate net benefit or moderate certainty of moderate to substantial net benefit)
Children
Pregnant women
Timing and target groups
Alcohol misuse screening and counselling
✔
First antenatal visit
Asymptomatic bacteriuria screening
✔
12–​16 weeks’ gestation
Breastfeeding promotion
✔
During pregnancy and after delivery
Congenital hypothyroidism screening
✔
Newborn
Deafness
✔
Newborn
Dental caries prevention (fluoride)
✔
From first tooth eruption to 5 years
Depression (major depressive disorders) screening
✔
Children aged 12+ if diagnostic and treatment provision in place
Folic acid supplementation
✔
All women planning or capable of pregnancy
Gestational diabetes screening
✔
After 24 weeks’ gestation
Gonococcal ophthalmia neonatorum prophylaxis
(antibiotic drops)
✔
Newborn
Infection screening (chlamydia, gonorrhoea, Hep B,
Hep C, HIV, syphilis)
✔
✔
All at-​risk of specific infections (all pregnant women for HIV and
syphilis);
Iron deficiency anaemia screening and dietary iron
supplementation
✔
✔
Screen all pregnant women; supplement at-​risk children
6–​12 months
Obesity
✔
Children age 6+ years
Pre-​eclampsia prevention (aspirin)
✔
After 12 weeks’ gestation in women at high risk
PKU (phenylketonuria) screening
✔
Newborn
Rhesus (D) incompatibility screening
✔
First antenatal visit and repeat at 24–​28 weeks if Rh (D) negative
Sexually transmitted infection, behavioural counselling
✔
Sexually active adolescents
Sickle cell disease screening
✔
Newborn
Skin cancer avoidance counselling
✔
Children and young adults age 10–​24 years with fair skin
Tobacco use counselling
✔
✔
School age children and pregnant women who smoke
Visual-​impairment screening
✔
Ages 3–​5 years
Data from: Recommendations of the U.S. Preventive Services Task Force. US Department of Health and Human Services, Washington, DC; October 2015 (http://​www.
uspreventiveservicestaskforce.org/​Page/​Name/​uspstf-​a-​and-​b-​recommendations/​Individual interventions).
2.11  Preventive medicine
135
consequence is the common use of medication (in this case statins)
to treat what is effectively a sociocultural problem.
Time constraints
One lesson we often forget is that things change over time. The North
Karelia project is of more than historical interest. It was a large-​scale,
long-​term programme to reduce mortality from cardiovascular
disease in northern Finland, started in 1972, which involved both
public health and individual intervention. Fig. 2.11.5 compares
mortality from cardiovascular disease in North Karelia with that in
10 other provinces in Finland before and during the intervention
by plotting two regression lines. The difference in slopes of these
two lines shows that the intervention was to some extent effective.
However, far more impressive in magnitude is the absolute fall in
mortality over time both in North Karelia and in the other prov-
inces. The lessons for preventive medicine are twofold: the effect of
medical intervention may be small compared to the effect of other
economic and social influences; and the change in baseline risk and
social context over time may be so rapid that it will substantially in-
fluence the absolute benefit of any preventive intervention.
Programme effectiveness
Many of the interventions cited earlier are known to work because
they have been tested in clinical trials. However, clinical trials
are often done in settings far removed from everyday life.
Participants are compliant, those delivering the intervention are
highly trained, the technology is of high specification, and quality
control is rigorous. These conditions will not hold under ordinary
working conditions. When recommended preventive interven-
tions fail, the most common reason is lack of effective implemen-
tation of the programme, rather than lack of effectiveness of the
intervention itself.
The three most important implementation issues that determine
programme effectiveness are:
•	 Coverage—​the population at risk needs to receive the intervention.
•	Delivery—​the way the intervention is delivered (e.g. maintenance
of equipment, the training of staff, and the storage of biological
materials) must be high quality.
•	Quality assurance—​performance indicators (e.g. the maximum
number of cases missed by screening) need to be predefined and
monitored.
Failure to implement any one of these issues can stop a poten-
tially effective intervention delivering any benefit. Two well-
​documented examples from the United Kingdom are the resur-
gence in whooping cough when vaccine population coverage
fell (after media publicity about potential adverse effects of the
pertussis vaccine) and the missed cases of uterine cervix which
occurred before adequate quality standards were defined and
monitored for cervical sampling and cytological assessment in the
national screening programme.
Conclusion
Preventive medicine is an integral part of clinical practice for all
doctors. It is our responsibility as clinicians not only to cure the
presenting illness but also to take action where possible to prevent
future morbidity. However, we must display both humility and as-
sertiveness in our approach. We need to be humble in our approach
to patients and to recognize that medicine is not the main deter-
minant of health. At the same time, we must display assertiveness
in our advocacy of prevention. In the United Kingdom, the Royal
College of Physicians’ reports, the campaigning of medical charities,
and the decision by virtually all doctors to stop smoking have all
played an important part in influencing both public and political
opinion against tobacco use. As a profession, we can make a unique
and powerful contribution to the prevention of premature death by
identifying and publicizing the existence and causes of ill health. We
also have a unique and powerful responsibility to act as advocates
for our patients in ensuring that these causes are addressed and the
risk to their health is ameliorated. Good clinical practice entails pre-
ventive medicine, but good preventive medicine is more than just
good clinical practice.
Total serum cholesterol (mg/100 ml)
80
60
40
20
0
100
200
300
400
500
South Japan
East Finland
Relative frequency (%)
Fig. 2.11.4  Distribution of serum cholesterol in southern Japan and
eastern Finland.
Reproduced from Rose G. Rose’s Strategy of Preventive Medicine, 2008. With
permission from Oxford University Press, Oxford.
1969
Year
1000
Mortality/1000
900
800
700
600
500
400
82
81
80
79
78
77
76
75
74
73
72
71
70
North Karelia
Ten other provinces
Fig. 2.11.5  The North Karelia project. Age-​standardized annual
mortality from cardiovascular disease in men aged 35–​64 years in
Finland, 1969–​1982.
Redrawn from original data published by Tuomilehto J et al., British Medical
Journal 1986: 293: 1068–​71.
136
section 2  Background to medicine
FURTHER READING
Department
of
Health/​Public
Health
England
(2015).
Immunisation Against Infectious Disease (The Green Book).
HMSO, London (https://​www.gov.uk/​government/​collections/​
immunisation-​against-​infectious-​disease-​the-​green-​book)
Rose G (1992). The strategy of preventive medicine. Oxford University
Press, Oxford.
US Preventive Services Task Force (2014). Guide to Clinical
Preventive Services—​Recommendations of the US Preventive
Services Task Force. (http://​www.uspreventiveservicestaskforce.
org/​)
World Health Organization (2003). World Health Report 2002.
Reducing risks, promoting healthy life. World Health Organization,
Geneva. http://​www.who.int/​whr/​2002
World Health Organization (2013). Global Action Plan for the Prevention
and Control of Non-​communicable Diseases 2013–​2020. (http://​
apps.who.int/​iris/​bitstream/​10665/​94384/​1/​9789241506236_​eng.
pdf?ua=1)

2.12 Medical screening 137
2.12 Medical screening 137
ESSENTIALS
Medical screening is the systematic application of a test or inquiry to
identify individuals at sufficient risk of a specific disorder to benefit
from further investigation or direct preventive action (these individ-
uals not having sought medical attention on account of symptoms
of that disorder). Key to this definition is that the early detection of
disease is not an end in itself; bringing forward a diagnosis without
altering the prognosis is useless and may be harmful.
Criteria for screening
Before a potential screening test is introduced into practice it must be
shown to prevent death or serious disability from the disease to an ex-
tent sufficient to justify the human and financial costs. To this end, three
screening parameters need to be determined: (1) the detection rate
(sensitivity); (2) the false-​positive rate (equivalent to the specificity); and
(3) the odds of being affected given a positive screening result (equiva-
lent to the positive predictive value). Where a detection rate cannot be
directly determined (e.g. in cancer screening, or if the efficacy of the
intervention is uncertain), a randomized trial is needed to show that
screening and subsequent treatment reduce disease-​specific mortality.
Circumstances where screening is not appropriate
Screening tests should not be practised simply because they seem
intuitively useful: chest radiography to screen for lung cancer and
manual breast self-​examination to screen for breast cancer were as-
sumed to be worthwhile, but randomized trials showed they did not
significantly reduce mortality from the cancer. Screening for prostate
cancer is widely practised, yet it does harm (from hazardous treat-
ment) with evidence of a relatively modest reduction in mortality
from the disease. Causal risk factors, even important ones like serum
cholesterol and blood pressure for cardiovascular disease, usually dis-
criminate poorly between individuals who will and will not develop
the disease they cause, because most of the population is ‘exposed’.
Particular disorders where screening is justified
The number of disorders for which medical screening has been
shown to be worthwhile is perhaps surprisingly small, but includes:
(1) antenatal screening (e.g. various single-​gene disorders, Down’s
syndrome, neural tube defects, and some infections such as hepa-
titis B and HIV that may be asymptomatic in the mother but cause
disease when transmitted to the fetus); (2)  neonatal screening
(e.g. congenital hypothyroidism, certain inborn errors of me-
tabolism such as phenylketonuria, and congenital deafness); (3)
adult screening—​in respect of cancer this has been shown to be
worthwhile for only three cancers—​breast, cervical, and colorectal;
screening individuals with diabetes mellitus prevents blindness
from retinopathy; screening men around the age of 65 prevents
death from ruptured abdominal aortic aneurysm; and screening
young women for chlamydia infection prevents pelvic inflamma-
tory disease and its complications (including infertility).
Future prospects
Tests that arise out of technological development in the absence
of a clear case of medical need (e.g. whole-​body scanning using
MRI or CT), should not be ‘sold’ to the public in the belief that they
are helpful. As with all screening methods, their value needs to be
shown before they are introduced into practice. Determining when
medical screening is an effective method of preventing serious dis-
ease and disability is one of the most challenging areas in medical
research.
Introduction
There is scarcely a medical discipline that does not include some
aspect of screening. It has made significant inroads into the pre-
vention of disease, but is often used inappropriately in circum-
stances where there is insufficient evidence that it benefits health.
Determining when screening is an effective method of prevention
is one of the most challenging areas in medical research today, re-
quiring an understanding of the principles of screening, the path-
ology, natural history, and epidemiology of the diseases concerned,
and quantitative information on the efficacy of the screening tests
and the remedies available.
2.12
Medical screening
Nicholas Wald and Malcolm Law1
1  © The author.
138
section 2  Background to medicine
Medical screening contains three elements:
1.	 Identifying individuals at sufficiently high risk of having or
developing a specific disorder to benefit from further investiga-
tion or direct preventive action.
2.	 It is systematically offered to a population that has not sought
medical attention for symptoms of the relevant disease. It is usu-
ally initiated by medical authorities, not the patient.
3.	 Its purpose is to benefit screened individuals. On this basis, mass
testing activities such as surveillance for HIV infection or pre-​
employment examinations to test fitness for work are not classi-
fied as medical screening.
The following definition has been widely used and encapsulates
these three elements:
Medical screening is the systematic application of a test or inquiry, to
identify individuals at sufficient risk of a specific disorder to benefit
from further investigation or direct preventive action, among those
who have not sought medical attention on account of symptoms of
that disorder.
Worthwhile screening aims to prevent death or disability from
specific disorders. Screening that simply brings forward a diag-
nosis without altering the prognosis is useless and may be harmful,
prompting needless anxiety, and possibly hazardous interven-
tions. The early detection of disease is not an end in itself. As
with any medical treatment, screening needs to be shown to offer
medical benefit and to be acceptably safe. Many medical disorders
are not candidates for screening, because they are too trivial or
because treatment is no more effective following screening than
following clinical presentation. The value of a screening test, in
which the benefits are considered in the light of the human and
financial costs, needs to be determined before it is introduced into
practice.
Requirements for a worthwhile screening test
See Box 2.12.1.
The disorder
The disorder needs to be clinically well defined and should, wher-
ever possible, be specified independently of the screening test. The
disorder should not be an ‘abnormal’ value of the screening test
being offered, such as a value lying outside the 95% range. This cre-
ates a circularity (‘tautological screening’) that makes it impossible
to determine whether screening is genuinely preventing disease or is
just causing overdiagnosis. It is necessary to know the distribution of
values of the screening test in individuals who have (or will develop)
the clinical disease and in individuals who do not, in order to assess
the value of the test.
An example is hypertension, or high blood pressure, an asymp-
tomatic condition that increases risk of a heart attack or stroke. If
hypertension were regarded as the medical disorder being screened
for, then all the ‘hypertensives’ would have blood pressure above the
cut-​off and all the ‘non​hypertensives’ below it (i.e. a perfect test).
The apparent screening perfection is a tautological misconception.
A high blood pressure measurement is the result of a screening test
(blood pressure measurement) for the clinical diseases (stroke and
myocardial infarction) caused by high blood pressure. In fact, blood
pressure measurement, although widely practised, is not a good
screening test for stroke or myocardial infarction. Many people
who will not have a stroke or myocardial infarction have high blood
pressure, and many who do will not. This is considered in further
detail next.
Prevalence or incidence
To derive an estimate of the odds of being affected among indi-
viduals with a positive screening result (see next), the prevalence,
or incidence, of the disorder needs to be known. Prevalence is the
number of cases of a disorder in a defined population at a given point
in time, incidence is the number of new cases occurring in a de-
fined population over a specified period. If the disorder is very rare
screening may not be justifiable, unless it can easily be incorporated
into an existing screening protocol. If the disorder is very common
(e.g. heart attacks and strokes), screening may be pointless and a
population-​wide preventive strategy may be needed.
Natural history
Screening should be restricted to disorders that are medically
important (i.e. associated with serious morbidity or premature
mortality).
Remedy
A remedy or treatment must be available that is more effective or ac-
ceptable following screening than at clinical presentation. Offering
an effective treatment is insufficient; the treatment must be more
effective or acceptable if delivered early.
Screening test
The screening test should be simple and safe. Some screening tests
are so simple or performed so routinely that they are not recognized
as such. For example, asking a woman’s age was once the antenatal
screening test for Down’s syndrome. A routine blood count includes
Box 2.12.1  Requirements for a worthwhile screening test
1	 Disorder: well defined
2	 Prevalence/​incidence: known
3	 Natural history: medically important disorder
4	 Remedy or treatment: more effective or acceptable than at clinical
presentation
5	 Screening test: simple and safe
6	 Test performance specified:
(a)	 Detection rate can be determined:  detection rate and false-​
positive rate known and acceptable. For a quantitative screening
test, the distributions of test values in affected and unaffected
individuals should be known, the extent of overlap sufficiently
small, and a suitable cut-​off level defined
(b)	Detection rate cannot be determined: randomized trial evidence
shows that the combined effect of screening and treatment is
sufficiently effective in preventing death and disability from the
disease being screened for, with an acceptably low proportion of
individuals requiring further investigation
7	 Financial: overall cost acceptable to achieve the health benefit
8	 Facilities: available or can easily be installed, including for diagnosis
and treatment
9	 Acceptability:  procedures following a positive result are generally
agreed and acceptable to the screening authorities and the screened
individuals
2.12  Medical screening
139
the antenatal screening test for β-​thalassaemia (mean corpuscular
volume), so the issue is not one of introducing the test but of system-
atically interpreting a test already carried out.
The purpose of testing generally defines whether it is a screening
or diagnostic test. If the aim is to identify a high-​risk group for fur-
ther investigation or preventive treatment, it is a screening test; if it
is to make a diagnosis, it is a diagnostic test. Screening tests indicate
a probability of having or developing a disorder, whereas diagnostic
tests usually indicate whether an individual is affected or unaffected.
The accuracy of each type of test does not itself define what type of
test it is. Sometimes mass testing, perceived as screening, is in fact
diagnosis (e.g. obstetric ultrasonography used routinely to diagnose
anencephaly). Screening tests usually apply to healthy populations,
but this is not always the case (e.g. screening for retinopathy among
people with diabetes).
Screening test performance
It is useful to separate screening tests for which detection rates can
be determined from screening tests for which this is not possible.
Detection rate can be determined
The performance of screening and diagnostic tests is defined by
three parameters: (1) the detection rate; (2) the false-​positive rate;
and (3) the odds of being affected given a positive result (OAPR).
Detection rate
The detection rate of a test (or test sensitivity) is the proportion of
affected individuals with positive test results (Table 2.12.1).
An advantage of the term detection rate over sensitivity is that it
avoids confusion with the usage of sensitivity in analytical biochem-
istry, where it means the minimum detectable amount in an assay. In
cancer screening, ‘detection rate’ is often taken to mean the preva-
lence of detected cancers at a screening examination.
False-​positive rate
The false-​positive rate is the proportion of unaffected individuals
with positive test results (Table 2.12.1).
The complement of the false-​positive rate is the specificity, which
is 100% minus the false-​positive rate (e.g. a false-​positive rate of 3%
is the same as a specificity of 97%). Advantages of the term false-​
positive rate over specificity are that (1) it is more easily understood
and remembered; (2) it focuses attention on the group to be offered
further medical intervention; and (3) a 10% false-​positive rate is
twice as ‘bad’ as one of 5%, whereas this is concealed within the cor-
responding specificity values of 90% and 95%.
Odds of being affected given a positive result (OAPR)
The OAPR is the ratio of the number of affected to unaffected indi-
viduals among those with positive test results (i.e. true positives:false
positives in the population in question).
The OAPR in Table 2.12.1 would be a:b if the numbers in
Table 2.12.1 came directly from screening everyone in a study popu-
lation. In practice this is uncommon because the disorder being
screened for is rare and so it is sensible to estimate the detection
rate on all the affected individuals but only a small sample of un-
affected individuals. Because of this sampling difference, tables like
Table 2.12.1 cannot be used to estimate the OAPR. It is best esti-
mated indirectly using estimates of the prevalence of the disorder
from one source and estimates of the detection rates and false-​
positive rates of the screening test from another source. This can be
done using a flow diagram such as that in Fig. 2.12.1, in which the
detection rate (80%) is applied to the number of affected individ-
uals (prevalence 1%) and the false-​positive rate (4%) to the number
of unaffected. Then the ratio of true-​positive to false-​positive tests
performed will be an unbiased estimate of the OAPR in a total
population. The OAPR is 1:5 after the screening test, and 38:1 after
the diagnostic test (detection rate 95%, false-​positive rate 0.5%). If
the prevalence of the disorder were halved (0.5%) the OAPRs would
be halved to 1:10 and 19:1, respectively. Thus, the less common the
disorder, the less likely people with positive results will be affected.
The OAPR can be expressed as a probability (‘true positives/​all
positives’) which is known as the positive predictive value (PPV). In
the example in Fig. 2.12.1) the OAPR, 80:400 = 1:5, is equivalent to
a predictive value of 1/​(1+5) = 1/​6 ​-​ 17%. The OAPR is more useful
than the PPV because it is numerically easier to compute when tests
are performed in sequence (Fig. 2.12.1), and it provides a better
impression of the relative performance of tests. In the example, the
OAPR of 38:1 is equivalent to a PPV of 97% (38/​39). If the detection
Table 2.12.1  Algebraic summary of detection and false-​positive
rates of qualitative tests or quantitative tests using a specified cut-​off
Test result
Affected
Unaffected
Positive
TRUE POSITIVES
a
FALSE POSITIVES
b
Negative
FALSE NEGATIVES
c
TRUE NEGATIVES
d
Total
a + c
b + d
Detection rate (sensitivity)
a
a+ c
False-​positive rate
(1 –​ specificity)
b
b+ d
10000
individuals
100
affected
Screening
test
Diagnostic
test
DR = 80%
DR = 95%
80+ve
400+ve
OAPR  =  80:400
1:5
76:2
38:1
9900
unaffected
76+ve
2+ve
FPR = 0.5%
FPR = 4%
Fig. 2.12.1  Flowchart to show the performance of screening and
diagnostic tests. The critical first step in constructing such a flowchart
is to separate individuals into affected and unaffected, not into
screen-​positive and screen-​negative.
Reproduced from Wald, N. An Introduction to Epidemiology in Medicine.
London: Royal Society of Medicine Press, 2004.
140
section 2  Background to medicine
rate of the screening test were halved (to 40%) the OAPR would
also be halved (to 19:1) but the PPV, 95% (19/​20), appears only a
little lower.
A good screening test has a high detection rate, a low false-​positive
rate, and a high OAPR (e.g. 1:10 is better than 1:50). Stating the de-
tection rate for a test is uninformative unless a false-​positive rate (or
specificity) is also stated. Screening performance is assessed by spe-
cifying the detection rate for a given false-​positive rate, or specifying
the false-​positive rate for a given detection rate.
The detection rates and false-​positive rates are independent of the
prevalence of the disease for tests that measure a consequence of the
disease (e.g. the antenatal markers of Down’s syndrome) but may not
be independent for a screening test that is a measure of a cause of
the relevant disease. For example, when screening for an autosomal
recessive disease such as cystic fibrosis by testing for a known DNA
mutation in the gene for the disease, a higher gene prevalence will
necessarily be linked to a higher disease prevalence.
The OAPR is always dependent on the prevalence. The higher the
prevalence the higher the OAPR, even if the detection rate and false-​
positive rate are constant.
Estimates of the detection rate and false-​positive rate can be ap-
plied from one population to others because they are generally inde-
pendent of the prevalence of the disorder. This is not the case with
the OAPR, which depends on the prevalence.
For a qualitative (or categorical) test, such as the presence or ab-
sence of a cystic fibrosis mutation among a given panel of muta-
tions tested for, there is only one detection rate and false-​positive
rate. This is not the case with quantitative (or non​categorical) tests,
such as maternal serum α-​fetoprotein (AFP) for open spina bifida
screening, which yield numerical results. In such cases, the detection
rate and false-​positive rate depend on the screening cut-​off level used
to distinguish positive from negative results. No single pair of detec-
tion and false-​positive rates summarizes the performance of tests;
both will vary as the cut-​off is changed. For example, at cut-​off level
A in the relative frequency distributions in Fig. 2.12.2 the test will
have a detection rate given by the area under the curve for affected
subjects to the right of cut-​off level A (95%), and a false-​positive rate
given by the area under the curve for unaffected subjects to the right
of the same cut-​off level (10%). The higher the cut-​off level (say, B or
C) the lower the detection rate and false-​positive rate.
It is common to summarize the performance of a test as a receiver–​
operator characteristic (ROC) curve. This is a plot of the detection
rate against the false-​positive rate, with both scales plotted from 0%
to 100%. In such a graph a useless test is represented by the diagonal,
indicating that the detection rate and the false-​positive rate are al-
ways the same. As the screening test improves, the ROC curve bows
out from the diagonal towards the axes. A perfect screening test
clings to the detection rate axis up to 100% while the false-​positive
rate remains at zero. The area under a ROC curve is sometimes used
to indicate the performance of a screening test, but it is not a satis-
factory measure of this. It is better to state detection rate for specified
false-​positive rate or vice versa. A weakness of a ROC curve is that
for screening tests that are potentially useful, the area of the graph
that is informative is restricted to a small portion, namely the part
covering false-​positive rates up to about 10% and detection rates
(from 40% to 100%) between about 50% and 100%. Fig. 2.12.3 illus-
trates detection rates plotted against false-​positive rates (from 0% to
10%) in multiple marker antenatal screening for Down’s syndrome,
showing the improvements in screening that have been made over
the past 20 years.
Good screening tests are usually early manifestations of the dis-
ease being screened for, while causes of a disease that are highly
prevalent in a community are usually poor screening tests. Causal
risk factors such as blood pressure for stroke are important aetio-
logically and account for a large proportion of the disease they cause
because they are usually common (e.g. most adults over 55 can
be said to have a high blood pressure) yet many escape the conse-
quences (e.g. a stroke). This means that causal risk factors usually do
not discriminate well between individuals who will and who will not
develop the disease.
Table 2.12.2 shows the detection rate for a 5% false-​positive rate
(DR5) for various risk ratio estimates between the top and bottom
fifths of the distribution of a risk factor. Even a ‘strong’ risk factor
with a fivefold risk ratio between the top and bottom quintile groups
(fifths) of the distribution (typical of low-​density lipoprotein (LDL)
cholesterol and myocardial infarction) has only a 14% detection rate
Affected
A
Unaffected
Cut-off A at 6 units
DR
FPR
95%
10%
Cut-off B at 6.5 units
DR
FPR
90%
2.5%
Cut-off C at 7 units
DR
FPR
80%
1%
Unaffected
Unaffected
Test variable (arbitrary units)
2
3
4
5
6
7
8
9
10
11
B
C
Affected
Affected
Fig. 2.12.2  Hypothetical example of the detection rate and false-​
positive rate of a screening test at three different cut-​off levels. The
implied vertical axis is the percentage of individuals at different levels
of the screening test variables, considered separately for affected and
unaffected individuals.
Reproduced from Wald, N. An Introduction to Epidemiology in Medicine.
London: Royal Society of Medicine Press, 2004.
2.12  Medical screening
141
for a 5% false-​positive rate. An interquintile risk ratio of around
1000 is necessary to achieve a detection rate of at least 75% for a 5%
false-​positive rate.
The OAPR can be determined by using the flowchart method il-
lustrated in Fig. 2.12.1. It can also be determined using the likeli-
hood ratio (LR) which is a measure of the ‘concentrating’ power of a
test (Fig. 2.12.4). For a group of people with values of the screening
variable above a specified cut-​off (i.e. all screen positives), this is the
proportion of the area for ‘affecteds’ to the right of the cut-​off div-
ided by the proportion of the area for ‘unaffecteds’ to the right of the
cut-​off. Fig. 2.12.4a, which is equivalent to the detection rate divided
by the false-​positive rate (DR/​FPR). It is the number of times indi-
viduals with positive results are more likely to have the disorder for
which they are being tested compared with the general population
(individuals who have not been tested).
That is, the OAPR is the likelihood ratio multiplied by the preva-
lence of the disorder (expressed as an odds):
OAPR
LR
prevalenceasanodds.
×
So (see Fig. 2.12.4a for example), if the detection rate is 80% and
the false-​positive rate is 1%, then the LR is 80%/​1%, or 80. If the
prevalence of the disorder were 1:1000, then
OAPR
80
1:1000
80 :1000
1:1000 / 80
1:12.5.
×
=
For an individual with the screening variable at some specific value,
the likelihood ratio is the height of the relative distribution curve for
‘affecteds’ at the test value for that individual divided by the height of
the curve for ‘unaffecteds’ at the same test value. So, for example, an
100
90
80
60
50
40
0
1
2
3
4
5
False-positive rate (%)
Integrated test
9.4
8.6
Serum integrated test
8.7
7.3
Combined test
8.6
7.2
Quadruple test
8.3
6.4
Triple test
6
7
8
9
10
70
Detection rate (%)
7.7
5.6
Fig. 2.12.3  Antenatal screening for Down’s syndrome: detection rates
and false-​positive rates for specified screening tests. The integrated test
consists of the ultrasound marker nuchal translucency and pregnancy-​
associated plasma protein A (PAPP-​A) measured in the first trimester,
and AFP, unconjugated oestriol (uE3), human chorionic gonadotropin
(hCG), and inhibin-​A in the second trimester. The serum integrated
test is the same as the integrated test without nuchal translucency. The
combined test consists of nuchal translucency, PAPP-​A, and hCG in the
first trimester. The quadruple test consists of AFP, uE3, hCG, and inhibin-​A
in the second trimester. The triple test is the same as the Quadruple test
without inhibin-​A. All the tests include maternal age.
Reproduced from Wald NJ, et al. (2004). SURUSS in perspective. Br J Obstet Gynaecol,
111, 521–​31, with permission from Wiley-​Blackwell.
Table 2.12.2  Detection rate for a 5% false-​positive rate (DR5)
according to relative risk between top and bottom fifths of the
distribution in unaffected individuals
Relative risk between top and bottom fifths of
the distribution in unaffected individuals
DR5 (%)
1
5
2
8
3
11
5
14
10
20
40
36
80
45
800
71
2000
79
10 000
89
Reproduced from Wald NJ, Hackshaw AK, Frost CD, ‘When can a risk factor be used as
a worthwhile screening test?’, BMJ, 319, 1562–​65 © 1999 with permission from BMJ
Publishing Group.
Affected
Affected
a
b
Unaffected
Unaffected
(a) Likelihood ratio for groups
(b) Likelihood ratio for individuals
DR = 80%
FPR = 1%
LR = 80%/1% = 80
FPR = 1%
LR = a/b = 12/1 = 12
DR = 80%
Test variable (arbitrary units)
2
3
4
5
6
7
8
9
10
11
12
Fig. 2.12.4  Likelihood ratio for groups and for individuals.
Reproduced from Wald, N. An Introduction to Epidemiology in Medicine.
London: Royal Society of Medicine Press, 2004.
142
section 2  Background to medicine
individual with a test result of 7 (arbitrary units) in Fig. 2.12.4b has
a likelihood ratio of 12, and so
OAPR
12
1:1000
12 :1000
1:1000 /12
1: 83.
×
=
In this way the likelihood ratio is used to estimate the risk for an
individual.
Fig. 2.12.5, showing the distribution of diastolic blood pressure
in men who did and did not subsequently die of a stroke, illustrates
how a particular blood pressure measurement of, say, 105 mm Hg
in a 70-​year-​old man, can be converted into a risk of developing a
stroke. At a diastolic blood pressure of 105 mm Hg the likelihood
ratio is 3. The annual risk of a fatal stroke in all 70-​year-​old men
regardless of blood pressure is 2:1000 (about 0.2%), so if his dia-
stolic blood pressure is 105 mm Hg the risk is 3 × 2:1000 or 6:1000
(about 0.6%).
To establish whether a quantitative screening test is worthwhile,
the overlapping distributions of the values of the screening test in
people with and without the disorder must be examined. If the two
distributions are widely separated, as in the example in Fig. 2.12.6
(ultrasound measurement of the diameter of the abdominal aorta
as a screening test for aneurysm likely to rupture), the test is good.
If they substantially overlap, as in the example in Fig. 2.12.7 (serum
cholesterol as a screening test for future death from ischaemic heart
disease or blood pressure as a screening test for stroke, Fig. 2.12.5),
it is not.
Detection rate cannot be determined
Determining the detection rate is straightforward when all indi-
viduals can be found to be either affected or unaffected. This is not
always possible, notably in cancer screening, because if a lesion is
found and a treatment carried out, one cannot know if that lesion
would have become a clinical case had treatment not been given,
or if it is ‘overdiagnosis’. The problem arises for any progressive dis-
order in which the clinical outcome is not determined in a uniform
way among all individuals, as would be the case if all screening re-
search were initially observational without intervention dependent
on the result of the screening test. Sometimes such an observational
approach is possible, for example, by storing serum samples in a
population of adults (without testing them at the time of collection),
and later identifying those who did and did not develop a cancer,
retrieving the serum samples, and testing them on a case–​control
basis; this provides an unbiased estimate of the screening perform-
ance of the test. Such an approach is not practical with tests based on
imaging, such as mammograms, which could not ethically be taken
and stored without being examined at the time. In such circum-
stances, it may never be possible to know the screening performance
of the test. The solution is to perform a randomized trial of screening
(and treatment) versus no screening. If this shows that mortality
from the disease is reduced, the combined effect of screening and
treatment is known, though the relative contributions of the two in
achieving the health benefit may not be.
Cancer screening must prolong survival (the time between diag-
nosis and death) to be effective, but because of two biases, prolonged
survival alone is insufficient evidence that screening genuinely
Men who
die of stroke
3
1
Men who do not
die of stroke
Diastolic blood pressure (mm Hg)
40
50
60
70
80
90
100
110
120
130
140
150
Fig. 2.12.5  Likelihood ratio of a fatal stroke in a man with a diastolic
blood pressure of 105 mm Hg.
Reproduced from Wald, N. An Introduction to Epidemiology in Medicine.
London: Royal Society of Medicine Press, 2004.
Ruptured
aneurysms
(n = 163)
Unruptured
aortas
(n = 3897)
Maximum aortic diameter (cm)
0
2
4
6
8
10
12
14
16
Fig. 2.12.6  Aortic diameter and ruptured aortic aneurysm. The
distribution of less than 2 cm is not real but simply represents the lower
half of the Gaussian distribution of more than 2 cm, which is based
on data.
Data from Law MR, Morris J, Wald NJ (1994). Screening for abdominal aortic
aneurysms. J Med Screen, 1, 110–​116.
Died of IHD
Did not die of IHD
Serum cholesterol (mmol/litre)
15%
5%
2
3
4
5
6
7
8
9
10
11
Fig. 2.12.7  Relative distributions of serum cholesterol in men who
subsequently died of ischaemic heart disease (IHD) and in men who
did not.
Reproduced from ‘A strategy to reduce cardiovascular disease by more than 80%’,
NJ Wald & MR Law, British Medical Journal 2003, 326; 7404 with permission from
BMJ Publishing Group Ltd.
2.12  Medical screening
143
improves prognosis. The first bias, lead time bias, is the prolongation
of survival from bringing forward the date of diagnosis, even though
the date of death is unchanged. The second bias, length time bias,
arises because cancer screening involves periodic examinations (say
3-​yearly). So screening will detect slowly growing tumours more
readily than rapidly growing ones because rapidly growing ones are
more likely to develop and proceed to clinical presentation within
the interval between two consecutive screening examinations, and
thereby escape detection at screening. Survival with such rapidly
growing screen-​detected cancers will inevitably be shorter than
average. This is biased sampling. Both biases can be avoided by com-
paring mortality from the specific cancer (the number of deaths
divided by the number of people at risk) between screened and
unscreened groups rather than comparing survival once a cancer
is diagnosed. The biases are avoided because mortality measures
deaths, whereas survival measures time.
Disease-​specific mortality could be subject to bias if the screened
and unscreened groups were at different risks of developing the
disease. For example, women of higher socioeconomic status may
be more likely to develop breast cancer and more likely to ac-
cept screening. So breast cancer mortality could still be higher in
screened women even if screening were effective. The only way to
reliably avoid such selection bias is to carry out a randomized con-
trolled trial to be sure that like is compared with like.
Financial considerations
Having determined that the first six requirements for a worthwhile
screening test are met (see Box 2.12.1), the financial considerations
need to be assessed. Screening programmes should seek to minimize
the cost for a given outcome (i.e. to maximize the cost-​effectiveness).
If the most medically effective form of screening is also the most cost-​
effective, it should be the programme of choice, provided it is affordable.
If the best screening policy is not the most cost-​effective, a judgement is
needed on whether the extra health gain justifies the extra cost.
Facilities
Medical screening effectively ‘creates’ patients by identifying indi-
viduals at sufficient risk of a disorder to be offered further tests or
treatment when they had no prior suggestion that they may have the
disorder. This necessarily creates anxiety and a demand for medical
attention, and an obligation to ensure that facilities exist for the ne-
cessary investigation, treatment, and support. Screening should not
be implemented until such arrangements have been made. Screening
therefore needs to be offered in the context of programmes that are
capable of meeting all the related needs of the people being screened.
Acceptability
Medical screening, including the treatment or remedy, must be ac-
ceptable to the population concerned and to the professional staff
involved. The purpose, the benefits, and the limitations of screening
need to be understood and regarded as important from the perspec-
tive of each individual who is offered screening. The decision not
to be screened needs to be respected and programmes should not
be driven by targets that set high uptake rates, though of course,
if the rates are very low it would call into question the need for
the screening programme. A  key element in the acceptability of
screening is individual choice set against a justifiable trust in the
medical system that offers screening.
Requirements for a worthwhile public
health screening programme
See Box 2.12.2.
Screening is a public health activity that should meet certain re-
quirements that arise from a professional responsibility to achieve
a collective health benefit. It is not the provision of a consumer
commodity. Its purpose is to improve the health of individuals and
thereby the health of the community.
Once the requirements for a worthwhile screening test shown
in Box 2.12.1 are met, there are four additional requirements for a
worthwhile screening programme implemented as a public health
service. These are summarized in Box 2.12.2.
In public health terms, interventions that reduce exposure to the
causes of disease should have priority over screening to detect early
disease and offer treatment, but they are not mutually exclusive.
A population approach correcting or reversing adverse risk factors is
often more effective. For example, the human papillomavirus (HPV)
vaccine is expected to steadily replace screening for cervical cancer
after the next 40 years.
Screening for specific disorders
Antenatal screening
See Table 2.12.3.
Much of antenatal care is screening—​looking for problems be-
fore they arise clinically. Detecting rises in blood pressure to warn
of the risk of pre-​eclampsia (which may cause perinatal death and
serious illness in the mother), and detecting maternal anti-​D anti-
bodies to warn of rhesus haemolytic disease of the newborn, are two
examples. The purpose of such screening is usually the welfare of
the mother and fetus, but in antenatal screening there is the unusual
situation in which some fetal disorders are so severe or potentially
disabling to justify screening and diagnosis, and the offer of a ter-
mination of pregnancy. Antenatal screening for open neural tube
defects, Down’s syndrome, severe congenital heart malformations,
and severe, incurable single-​gene disorders are examples.
Screening for four infections is worthwhile (syphilis, HIV, hepa-
titis B, and bacteriuria) because they may not be clinically apparent
in the mother but can cause serious preventable illness in the neo-
nate (either immediately or in later life). Prognosis is substantially
improved if the infection can be detected in the mother and
Box 2.12.2  Requirements for a screening programme
implemented as a public health service
1	 Equitable: equal access to screening services
2	 Organized: individuals are offered screening in an organized manner
according to a specified protocol and with relevant information pro-
vided to permit an informed choice
3	 Comprehensive: screening is the first step in a programme of service
and care that includes counselling screen positives, diagnosis, support,
and treatment
4	 Monitored and auditable:  key aspects of the programme should
be monitored so that remedial steps can be taken if they are below
standard
144
section 2  Background to medicine
Table 2.12.3  Summary of antenatal screening tests of proven value
Disorder
Approximate
natural birth
prevalence
(per 10 000) in UK
Primary screening
test
Secondary
screening test(s)
(if available)
Detection
rate (%)
False-
positive
rate (%)
Odds of being affected given a
positive result
Diagnostic test
Intervention
1°ry screening
test
2°ry screening
test
Autosomal or sex-linked recessive disorders
Cystic fibrosis
4
Test for CF mutation
in both parents
(‘couple screening’)
72
0.09
1:3
–
CVS or amniocentesis
a
Sickle cell disease
3
Ethnic origin enquiry
(Afro-Caribbean)
Sickling test; Hb
electrophoresis in
mother, and in father
if positive in mother
99
3
1:100
1:3
CVS or amniocentesis
a
β-Thalassaemia
6
Red cell MCV or
MCH in mother
Hb A2 assay in
mother, and in father
if positive in mother
89
7
1:125
1:3
CVS or amniocentesis
a
Tay–Sachs disease
0.04
Ethnic origin enquiry
(Ashkenazi Jew)
Hexoseaminidase
assays in father, and
mother if positive in
father
50
1
1:3600
1:3
CVS or amniocentesis
a
Haemolytic disease
of the newborn
(D-antigen of Rh
system)
40
Rh grouping and
test for antibody in
mother
Rh grouping of
father; quantitation of
maternal antibody
100
16
1:31
1:26
CVS or amniocentesis
Intrauterine transfusion,
early delivery with
exchange transfusion
Haemophilia
0.5
Recognition of
affected male relative
(carrier detection)
Test for mutation in
mother
55
<0.01
1:35
1:3
CVS or amniocentesis
a
Chromosomal disorders
Down’s syndrome
(Trisomy 21)
18
Integrated 1st and
2nd trimester
89
1.0
1:5
CVS or amniocentesis
a
1st trimester alone
77
1.0
1:6
CVS or amniocentesis
a
2nd trimester alone
70
1.0
1:6
CVS or amniocentesis
a
Reflex DNA screening
91
0.05
4:1
Amniocentesis
Trisomy 18
2.3
Integrated 1st and
2nd trimester
92
0.2
1:3
CVS or amniocentesis
1st trimester alone
89
0.2
1:3
CVS or amniocentesis
2nd trimester alone
58
0.2
1:5
CVS or amniocentesis
Reflex DNA screening
89
0.05
1:1
Amniocentesis
Trisomy 13
1.4
Integrated 1st and
2nd trimester
72
0.2
1:12
CVS or amniocentesis
1st trimester alone
73
0.2
1:11
CVS or amniocentesis
2nd trimester alone
20
0.2
1:42
CVS or amniocentesis
Reflex DNA screening
79
0.05
1:3
Amniocentesis
2.12  Medical screening
145
Other congenital malformations
Spina bifida (open)
8.5
Maternal serum AFP
assay
Ultrasound
87
0.5
1:4
Amniotic fluid
acetylcholinesterase +
repeat ultrasound
a
Anencephaly
10
Ultrasound
100
0
1:0
–
Independent
conf irmation
a
Severe cardiac
malformations
20
Ultrasound
46
≤0.6
≥1:6
–
Independent
conf irmation
a
Infections transmitted from mother to fetus
Congenital rubella
syndromeb
0.12
Absent antibodies in
mother
90
1.6
<1:1300
–
None
Vaccinate mother after
delivery to protect
subsequent pregnancies
Congenital syphilis
0.2
VDRL test or
flocculation test in
mother
Specific treponemal
test in mother
90
0.2
1:100
1:50
None
Penicillin
AIDS
1
ELISA test for IgG
antibody in mother
(repeated on same
sample if positive)
ELISA test on repeat
sample
99.9
0.13
1:13
1:<5
None
Antiretroviral drugs to
mother and infant
Hepatitis B causing
hepatoma and
chronic liver disease
1.4
ELISA test for HBsAg
in mother (repeated if
positive)
≥98
0.14
1:10
None
Recombinant vaccine
to neonate, hepatitis B
immunoglobin at birth
except when mother has
antibodies to e antigen
Maternal
bacteruria causing
pyelonephritisc
200
Urine culture
76
4
1:4
None
Antibiotics to mother
Noninfectious maternal disease affecting fetus
Maternal high
blood pressure/
pre-eclampsia
causing perinatal
death
93 (rate of all
perinatal deaths)
Maternal
blood pressure
measurement
Test for proteinuria
38 (of all
perinatal
deaths)
30
1:77
1:41
None
Blood pressure lowering
drugs
a Information on disorder and its prognosis, counselling, termination of pregnancy, or preparation for birth of affected child, advice on risk of recurrence.
b Worthwhile only with low uptake of childhood rubella vaccination in community.
c May cause low birthweight or fetal death.
AFP, α-fetoprotein; CVS, chorionic villus sampling.
Adapted from Wald NJ, Leck I, eds. Antenatal and Neonatal Screening (2nd ed). (2000) Oxford University Press, Oxford.
146
section 2  Background to medicine
appropriate treatment given to the mother before birth, the neonate
at birth, or both. Routine screening for rubella syndrome in preg-
nancy is generally not worthwhile because it cannot prevent the dis-
order in the pregnancy screened; it can only lead to vaccination after
birth in women without rubella antibodies. The preferred method of
prevention is childhood vaccination.
In recent decades antenatal screening has taken on a scientific
methodology and rigour that has permitted the development of
screening programmes that are now standard throughout the world.
The first such initiative arose with antenatal screening for open
neural tube defects, first by measurement of maternal serum AFP
and later by ultrasonography, which is used with AFP in many places
and has replaced it in some. Screening now detects virtually all cases
of anencephaly with scarcely any false positives, and 87% of cases of
open spina bifida with a false-​positive rate of less than 1%. The birth
prevalence of neural tube defects in Britain has declined by over 90%
from 1 in 250 births in the early 1970s to less than 1 in 2500 now, due
in part to screening, in part to an increase in folate intake through
food and vitamin supplements.
Until the 1980s, antenatal screening for Down’s syndrome (trisomy
21) was based on maternal age. In 1988 the triple test was described,
based on combining second trimester serum markers with maternal
age. Fig. 2.12.3 shows the subsequent improvement in screening per-
formance as the number of available markers increased over time.
The integrated test can detect about 85% of affected pregnancies for a
false-​positive rate of only 0.9%; the low false-​positive rate is important
because women with positive results usually have an amniocentesis,
which may induce the miscarriage of a healthy fetus. Combining
markers to obtain a single test result for an individual involves the
multiplying of the likelihood ratios for each marker in that individual
(as in Fig. 2.12.4a), allowing for any correlation between them (con-
sidered separately among affected and unaffected individuals). So,
for example, in the simple situation of three independent screening
markers that correspond to likelihood ratios of 3, 4, and 5, the com-
bined likelihood ratio is 60 (3 × 4 × 5). To determine the screening
performance of tests based on multiple markers, a hypothetical popu-
lation of screened individuals is generated and the combined like-
lihood ratio for each individual calculated and converted to risk by
multiplying it by prevalence expressed as an odds. The overlapping
distributions of risk in affected and unaffected individuals are plotted,
determining detection rates for specified false-​positive rates in the
same way as for a single screening marker. Then risk itself becomes
the screening variable—​which is convenient, because it is exactly what
is needed in reporting results to screened individuals.
Most screening markers associated with Down’s syndrome vary
with gestational age, so a high level at one gestational age could be low
at another. A widely used advance in screening is to express all values
as ‘multiples of the median’ (MoM) for unaffected (or all) screened in-
dividuals at a specified gestational age, so that 1.0 MoM represents the
median value (‘normal’), 2.0 MoM is twice ‘normal’, and 0.5 MoM is
half ‘normal’. The MoM has the advantages that as a ratio it is unitless
and so avoids the need to specify the original units of measurement
(which vary from centre to centre), that it automatically adjusts for
gestation, and that it indicates how high or low a particular value is.
In pregnancy, fragments of placental DNA (which reflect the DNA of
the fetus) are shed into the maternal circulation and mix with fragments
of maternal DNA shed from maternal cells in an approximate ratio of
1:10. DNA analysis, such as counting the number of cell-​free DNA
fragments that map to chromosome 21, has been used as an antenatal
screening test for Down’s syndrome, with a high screening perform-
ance (detection rate 98–​99%, false-​positive rate 0.2%), but there is a test
failure rate of a few per cent, partly due to a lack of placental DNA in
the maternal plasma. At present the test is costly and labour intensive
which has tended to preclude its use as a universal screening test. DNA
analysis has become part of routine antenatal screening for Down’s
syndrome and other chromosome disorders (trisomy 18 and 13). A
cost-effective screening strategy that has been introduced is antenatal
reflex DNA screening in which women who accept screening have a
conventional first trimester test. Women are screen-negative unless the
first trimester markers yield a risk of having a pregnancy with trisomy
21, 18 or 13 above a certain level (eg ≥ 1 in 800 comprising about 10%
of the women) and then automatically have a DNA sequencing ana-
lysis using stored plasma from the original blood sample (i.e. a reflex
analysis), thereby avoiding the need to recall them for counselling and
another blood collection. The conventional test markers and the DNA
analysis are considered together as a single test.
Neonatal screening
Neonatal screening for phenylketonuria, one of the first population-​
wide screening programmes to be introduced, has proved to be ef-
fective and worthwhile in spite of the rarity of the disorder (about 1 per
10 000 births). A low-​phenylalanine diet prevents severe mental retard-
ation in affected infants. Neonatal screening has prevented cretinism,
which is now extremely rare. Additional screening tests could be added
to the blood already collected for phenylketonuria and hypothyroidism
screening (e.g. MCADD; see Table 2.12.4), and may be justified for
other inborn errors of metabolism, given that much of the cost and ef-
fort is already spent. However, a line needs to be drawn; tandem mass
spectrometry can identify over 40 disorders, but only a handful jus-
tify screening as defined. Neonatal screening for congenital deafness is
worthwhile and has recently been introduced in the United Kingdom
using technology that does not rely on voluntary subject response to
noise, thus making it possible to test for hearing deficit in infancy.
Screening for congenital dislocation of the hip has been widely prac-
tised for many years, without good evidence of efficacy. Galactosaemia
(an autosomal recessive inborn error of metabolism) may cause ser-
ious illness in the neonate, including septicaemia and encephalop-
athy, and cognitive impairment in later life, but it has not been shown
that neonatal screening prevents these effects. Neonatal screening for
cystic fibrosis has been introduced in some places without evidence
that screening reduces the incidence or severity of the associated lung
disease, the main cause of disability and death from cystic fibrosis.
Screening in childhood
Children are examined routinely to see if they are gaining weight and
height as expected and to assess their hearing and vision. There is
no evidence that systematic examination of children achieves greater
health benefits than encouraging parents to take their child to a doctor
if they are concerned, but nonetheless much such activity has taken
place. In spite of the lack of formal evidence, it is probably sensible
to check the visual acuity of children on starting school, as is current
practice in many places. Unfortunately, the lack of evidence to sup-
port screening in childhood is often camouflaged in the term ‘child-
hood surveillance’. As with all screening, evidence of benefit should
be sought before acceptance, even if this requires large-​scale studies.
One disorder that merits screening, is screening for familial hyper-
cholesterolaemia, an inherited disorder with a prevalence of about
4 per 1000 that leads to early cardiovascular disease. Parents can be
2.12  Medical screening
147
Table 2.12.4  Summary of neonatal screening tests of proven value
Disorder
Approximate
natural
prevalence
(per 10 000
births) in UK
Primary screening
test
Secondary
screening test(s)
Detection
rate (%)
False-​positive
rate (%)
Odds of being affected given a
positive result
Diagnostic test
Intervention
1°ry screening
test
1°ry and 2°ry
screening tests
Congenital
hypothyroidism
3
T4 or TSH assay
before hospital
discharge
TSH and T4 at
5–​7 days
100
20
1:668
1:19
Clinical examination,
T4 , free T4 , TSH, thyroid
scan
Thyroxine
Phenylketonuria
1
Serum
phenylalanine assay
Repeated serum
phenylalanine
assay
100
0.2
1:22
1:0.05
High plasma
phenylalanine (>16.5
mg/​dl) using quantitative
technique; exclusion of
biopterin defects
Diet low in
phenylalanine
Medium chain
acyl CoA
dehydrogenase
deficiency
(MCADD)
1
Tandem mass
spectrometry
(together with PKU)
100
0
1:0
Repeat test
Avoidance of
fasting, prompt
treatment of minor
illnesses
Deafness
14
Transient evoked
otoacoustic
emissions (TEORE)
Automated
auditory brainstem
response (AABR)
80
0.6%
1:5
Repeat test
Hearing aid or
cochlear implant
Adapted from Wald NJ, Leck I, eds. Antenatal and Neonatal Screening (2nd ed). (2000) Oxford University Press, Oxford.
148
section 2  Background to medicine
screened at the same time in child-parent screening for familial hyper-
cholesterolaemia. The method uses the timing of childhood vaccin-
ation as a convenient turnstile when a cholesterol measurement is
most discriminatory for familial hypercholesterolaemia (1–2 years
of age) and a cholesterol measurement used in combination with a
DNA mutation analysis. The parents of affected children are offered
testing because as the disorder is inherited as an autosomal dominant
disorder one parent of an affected child will also be affected. The af-
fected parent can be offered preventive statin therapy immediately
and the child after the age of ten. The method leads naturally to cas-
cade testing in which close relatives are tested: half of all first degree
relatives being affected. This protocol has been shown to be feasible
and acceptable in a national demonstration project. This method of
screening is being considered by various public health agencies.
Adult screening
Perhaps surprisingly, only a few disorders justify medical screening
in adults. These are summarized in Table 2.12.5.
Cancers
Three cancers meet the screening requirements: breast, cervical, and
colorectal. Cervical cancer screening illustrates the principle that
effective adult screening programmes require a population age–​sex
register. Everyone in the appropriate age–​sex group for screening
can then be identified and sent written invitations at appropriate
intervals. Formerly, cervical screening was carried out ‘oppor-
tunistically’ when women happened to consult doctors, and such
screening failed because younger women, at lower risk, see doc-
tors more frequently than older women, at higher risk, so cervical
smears were carried out on the low-​risk group, and at more frequent
intervals than necessary for effective screening. It was only with the
introduction of a systematic screening programme based on age–​sex
registers that most older women were screened and cervical cancer
mortality fell appreciably in the United Kingdom and other Western
countries. Now women are invited 3-​yearly between the ages of 25
and 49, and 5-​yearly between the ages of 50 and 64. Screening in
the United Kingdom is based on testing a cervical brush sample for
human papillomavirus (HPV) followed by cytological examination
on the same sample if the HPV test is positive (reflex testing). The
evidence on efficacy comes from non-​randomized studies: screening
reduces mortality from cervical cancer by about 80%.
Mammographic breast cancer screening is offered in the United
Kingdom at 3-​yearly intervals to women aged 50 to 70 (though the age
range may soon extend down to age 47 and up to 73). Randomized
trials have shown that it reduces breast cancer mortality, by about a
fifth in a population offered screening or a third in women who accept
screening. Manual breast self-​examination by women to screen for
breast cancer has been shown in randomized trials not to significantly
reduce mortality, an observation which illustrates that screening for
cancer and other diseases should not be practised simply because it
seems intuitively useful: rigorous evidence on efficacy is needed.
A colorectal cancer screening programme based on 2-​yearly
faecal occult blood testing in men and women aged 60 to 70 has been
shown in randomized trials to reduce colorectal cancer mortality by
about 15% in a population offered screening. Population screening
has been introduced in the United Kingdom.
A second screening procedure further reduces colorectal cancer
mortality. Two randomized trials of once-​only flexible sigmoid-
oscopy versus no intervention, with identification and resection of
colonic polyps, showed a reduction in colorectal cancer mortality of
about 30% in a population offered screening, or about 40% in people
who attended for screening. There are no published randomized
trials of colonoscopy but case–​control (observational) studies indi-
cate that the mortality reduction is little or no greater than with flex-
ible sigmoidoscopy (as fewer cancers occur in the ascending colon
and few of the ones that do seem to be prevented). A meta-​analysis
of cross-​sectional studies in people who had both CT colonography
and colonoscopy showed that CT detected 48%, 70%, and 84% of all
polyps less than 6, 6–​9, and more than 9 mm, with a false-​positive
rate of 7% (i.e. 7% of unaffected people require colonoscopy). This
suggests that CT may be an acceptable surrogate for flexible sigmoid-
oscopy or colonoscopy: it may be a little less effective in preventing
colorectal cancer, but it is less invasive.
Chest radiography to screen for lung cancer has been shown in
randomized trials not to significantly reduce mortality. However,
low-​dose spiral computed tomographic screening has been shown
in a randomized trial to reduce mortality from lung cancer, by
about 20%. The screened group had yearly CT over three years with
6.5 years follow-​up; the control group had single view chest X-​rays,
but since these have been shown to be ineffective the trial can be
interpreted as CT versus no screening. However, the false-​positive
rate was high; about 22% of people without lung cancer had positive
screening tests. Much of the diagnostic uncertainty was resolved with
further imaging examinations, but about 4% of participants without
lung cancer had bronchoscopy, thoracoscopy, or thoracotomy. There
is also uncertainty as to whether screening is cost-​effective, and con-
cern that smokers may derive false reassurance from negative screens
and continue to smoke. Launching lung cancer population screening
programmes by identifying smokers and offering them a spiral CT
examination, while effective, remains an issue for discussion.
In systematic population-​based cancer screening programmes,
only people within a relatively narrow age range are invited for tests
(effectively, age is used as the initial screening enquiry). Cancer
screening tends to be most effective around the age of 60 in terms of
cost per year of life saved; the lower incidence of cancer in younger
people, and the shorter life expectancy in older people, mean that
fewer years of life will be gained for the same number of people
screened. The justification for a narrow age range is economic. Older
people are not turned away, however, and the age range over which
women are invited for mammographic screening, for example,
widened over time (it was originally 50–​64). Usually it is not ap-
propriate to stop inviting people for screening examinations above a
certain age; if they are fit enough and willing to attend for screening
examinations, they are suitable candidates for screening. Cancer
screening is generally conducted at 2-​ to 3-​yearly intervals; in prin-
ciple, more frequent screening would detect more cancers but the
yield per 1000 screening examinations would be lower.
Screening for prostate cancer, mainly through measurement of
serum prostate-​specific antigen (PSA) was introduced into medical
practice with no evidence of reduction in mortality. PSA can distin-
guish between individuals who will and will not die of prostate cancer.
However, discrimination weakens as the interval between the PSA test
and clinical presentation or death from the cancer lengthens. By the
time the PSA test is highly discriminatory, the disease may be too far
advanced for treatment to be effective. The usual cut-​off levels pro-
posed for PSA screening (c.4 ng/​ml) lead to a high proportion of older
men being positive. A prostate biopsy in these individuals is often
positive, because 25% of prostates in men aged 70 have histological
2.12  Medical screening
149
Table 2.12.5  Summary of adult screening for selected disorders
Disorder
Prevalence
Screening
procedure
Age range
Subsequent
investigation
Detection
rate
Positive rate
Odds of
disorder
in screen
positives
Uptake of
screening
Treatment
Reduction in
disease
Breast cancer
4% of all deaths
(women)
Mammography
2–​3-​yearly
50+
Further imaging;
fine needle biopsy
Not
applicable
8% first screen,
4% subsequent;
biopsy rate
0.8%
1:6 (2:1 among
women
biopsied)
70–​80%
Surgery (±
chemotherapy,
radiotherapy)
24% reduction in
mortality at age 50–​74;
16% at age 40–​49
(from meta-​analyses
of randomized trials)
Colorectal
cancer
3% of all deaths (men
and women)
Faecal occult
blood testing
2-​yearly
60+
Colonoscopy ±
barium enema
Not
applicable
2–​3%
1:10
50–​60%
Surgery
15–​18% reduction in
mortality (from two
randomized trials)
Cervical cancer
0.5% of all deaths
(women)
Cervical smear
± HPV testing
3–​5 yearly
25+
Repeat smear in
6 months (mild
dyskaryosis);
colposcopy
(moderate/​severe
dyskaryosis)
Not
applicable
5–​10% (higher
in younger than
older women),
lower with HPV
test and smear
–​
80%
Local ablation or
excision (rarely
hysterectomy)
90% reduction in
mortality (from
case–​control studies)
Diabetic
retinopathy
Proliferative retinopathy
50% IDDM
50% NIDDM
Macular oedema
15% IDDM
10% NIDDM
Retinal
photography
with mydriatic
yearly
All
Assessment by
ophthalmologist
78%
0
50% if done
in hospital
clinics
Photocoagulation
Reduction in
blindness >90%
(proliferative
retinopathy) 65%
(macular oedema)
Abdominal
aortic aneurysm
rupture
Men aged 65+
2% of all deaths
7% have aortic diameter
≥3.0 cm
Ultrasound scan
65 (men)
CT or MRI
86%
0.6%
75%
Open surgery
Chlamydia
trachomatis
genital infection
(subsequently
causing PID)
Chlamydia: 5% among
women under 25
PID 2%
Nucleic acid
amplification
test on urine
sample
<25
(sexually
active)
–​
90–​95%
<1%
64%
Doxycycline or
azithromycin
56% reduction in PID
HPV, human papillomavirus; IDDM, type 1 diabetes; NIDDM, type 2 diabetes; PID, pelvic inflammatory disease (Ashton et al., 2002).
150
section 2  Background to medicine
evidence of cancer even though only a small minority of these men
will suffer from or die of the disease. Such ‘overdiagnosis’ (the diag-
nosis of cancers that would otherwise never have come to clinical at-
tention) is a potentially serious problem in cancer screening. These
cancers are best never diagnosed; once diagnosed, anxiety and un-
necessary hazardous investigation and treatment will ensue.
In 2009, two randomized trials of PSA screening for prostate
cancer were reported, one showing a significant (p = 0.01) 20% re-
duction in prostate cancer mortality in men invited for screening
(27% in those who were screened) and the other one showing a
non​significant increase, but consistent with a 15% reduction.
Both trials showed a high rate of overdiagnosis; in the larger of
the two trials, for every one prostate cancer death prevented, 1410
men were screened, of whom 16% (230) had a biopsy, identifying
49 prostate cancers of which 48 were treated unnecessarily. Taking
the two trials together, screening for prostate cancer by PSA testing
probably does reduce prostate cancer mortality. Subsequent smaller
trials have shown similar results. The reduction, however, may not
be judged sufficient to warrant the level of overdiagnosis which
leads to many men receiving unnecessary hazardous treatment.
A randomized trial of ovarian cancer screening using ultrasound
examination of the ovaries and measurement of a serum protein
marker (CA125), found no statistically significant difference be-
tween women randomized to screening and those randomized to
‘no screening’. While the efficacy of such screening cannot be com-
pletely excluded the evidence suggests that, if there is an effect, it is
small. A cancer screening programme that is currently under inves-
tigation is screening for future stomach cancer by identifying people
with Helicobacter pylori infection of the stomach.
Non​malignant diseases
Screening for abdominal aortic aneurysms that, in the absence of
surgery, are likely to rupture, by the ultrasound measurement of the
aortic diameter, is worthwhile. The test is very discriminatory (see
Fig. 2.12.6). Ruptured abdominal aortic aneurysms account for 2%
of all deaths in men over 65, but are rare when the maximal aortic
diameter is less than 5 cm. In the United Kingdom a screening pro-
gramme based on abdominal aortic diameter using ultrasound is in
progress for men aged 65 (rupture is rare in younger men). Over all
ages ruptured abdominal aortic aneurysm is about twice as common
in men as in women. Mortality rates for women for women are
similar to those in men about 10 years younger. Most men will need
only a single scan in the year in which they reach 65.
Screening people with diabetes for retinopathy using retinal pho-
tography is very effective; it has been shown in randomized trials
to reduce blindness by 90% with proliferative retinopathy and 65%
with macular oedema. A national screening programme operates in
the United Kingdom, based on inviting people from diabetic regis-
ters held in general practice.
Chlamydia infection in young women causes pelvic inflammatory
disease (which may be complicated by chronic pelvic pain, ectopic
pregnancy, and tubal infertility and, when giving birth, causes neo-
natal eye and lung damage). Screening for chlamydia infection based
on urine samples is followed by short term antibiotic treatment and
is effective. Screening women under 25 has been recommended but
no systematic screening programme has been introduced in the
United Kingdom.
Much screening activity falls under the category of ‘risk factor
screening’ and such screening tends to be ineffective (e.g. cholesterol
testing in screening for future ischaemic heart disease events; see
Fig. 2.12.7), blood pressure measurement in screening for future
stroke, and bone density measurement as a screening test for future
hip fractures. The problem arises because, for the reasons given here,
risk factors that may be important causes of disease are usually poor
screening tests. Most adults have high serum cholesterol and high
blood pressure relative to levels in young adults (say at age 20), and
all postmenopausal women have low bone density relative to pre-
menopausal women, so nearly all older adults are ‘exposed’.
Fig. 2.12.8 shows the effect of combining different markers on
the detection rate and false-​positive rate where several markers that
each have a detection rate for a 5% false-​positive rate (DR5) of 10%,
15%, or 20% and the standard deviation is the same in affected and
unaffected individuals. Only when tests individually have a DR5 of
about 20% or greater will multiple marker screening become a real-
istic proposition. For example, combining five relatively weak inde-
pendent markers, each with a DR5 of 15%, yields only a 40% overall
detection rate for a 5% false-​positive rate, and combining ten yields
a 60% detection rate for the same 5% false-​positive rate. At present,
screening for future coronary disease and most other diseases using
causal risk factors is not effective because even in combination they
are not sufficiently discriminatory.
Hypothyroidism in adults is widely regarded as a preventable
cause of lethargy and depression. This has prompted attempts at
screening for this disorder by measuring levels of thyroxine (T4)
or thyroid-​stimulating hormone (TSH) and classifying individ-
uals as positive if TSH is above or T4 below the relevant reference
range (which is usually the 95th centile range in the population).
This is an example of the ‘tautological screening’ that arises from
defining a disorder in terms of the test used to screen for it (see
earlier). The solution to this circularity is to identify individuals
from a population with TSH or T4 outside specified TSH or T4
limits and then offer each, in random order, thyroxine or placebo
to determine whether thyroxine treatment relieves the symptoms
more often than can be explained by chance. Each person is
therefore their own control, and the response to treatment defines
the clinical disorder. Such a cross-over randomized trial has been
100%
80%
Number of screening markers
0
5
Detection rate for multiple markers
10
15
20
25
30
35
40
60%
40%
20%
0%
Fig. 2.12.8  Overall screening performance from combining
individual screening markers: detection rate for a 5% false-​positive rate
(DR5) according to the number of screening markers combined that
individually have a DR5 of 10% or 15% or 20%.
Reproduced from Wald N, Morris J and Rish S, ‘The efficacy of combining several risk
factors as a screening test’, J Med Screen 2005; 12: 197–​201. London: Royal Society of
Medicine Press, 2005.
2.12  Medical screening
151
done and showed that screening for hypothyroidism is worthwhile.
In screening the same approach should be used to identify which
individuals will benefit from treatment.
Clarity of terminology and purpose
Certain terms used in screening are probably best avoided because
they lack clarity. The term ‘carrier screening’ implies that carriers of
autosomal recessive disorders (e.g. cystic fibrosis) themselves have a
disease; they do not. The goal of such screening is to identify couples
who are both carriers. ‘Couple screening’ involves collecting samples
from both parents and reporting a positive result only when both are
carriers. The term ‘genetic screening’ lacks clarity and tends to imply
screening for inherited disorders even though some genetic disorders
that are screened for (e.g. Down’s syndrome) are usually not inherited.
The term creates a false impression that something special is being
offered that other forms of screening lack. For many people genes and
consequently all things genetic are seen as highly determinant, even
inevitable, influences, which is usually not the case. Genetic markers of
a disease are in most instances too insensitive and too non​specific for
screening purposes. The term ‘case finding’ often implies the identifica-
tion of cases of the disorder being screened for, while in fact it identifies
individuals with a positive screening test for that disorder. For example,
a case of ‘hypertension’ relates to the test result (high blood pressure),
not the diseases it causes. The term ‘opportunistic screening’ is a eu-
phemism for non​systematic and non​organized screening.
The purpose of medical screening is clear—​to avoid disability and
premature death at an acceptable level of safety. Determining efficacy
is essential. Many screening tests are effective and should be part of
public health practice. But particular care is needed in evaluating tests
that arise out of technological development in the absence of a clear
case of medical need. Whole-​body scanning using MRI and fetal ultra-
sound examination are examples. Such screening, without defining
the specific disorders being screened for, detects ‘incidentalomas’ (so-​
called ‘abnormal’ findings with little or no knowledge of their med-
ical significance). There is no place for such screening in responsible
medical practice. For example, total body MRI scanning is now ad-
vertised to the public as a screening test with little attention paid to
whether it prevents serious disability or death, or meets the criteria set
out in Box 2.12.1. A routine fetal anomalies scan at about 18 weeks of
pregnancy has some proven specific applications (e.g. the detection of
anencephaly, severe congenital heart disease, and placenta praevia ex-
tending to cover the internal cervical os), but the term ‘fetal anomaly
screening’ lacks specificity. The challenge in performing a scan is to seek
these specific anomalies but not to report other ‘incidentalomas’ which
will undoubtedly lead to parental anxiety and further investigation but
for which early detection has not been shown to be worthwhile. Under
the ambiguous heading of genetic screening, so-​called ‘gene chips’ have
been developed, that can detect in one test many hundreds of genetic
mutations with little or no evidence that knowledge of these will lead to
useful medical intervention that will improve the health and quality of
lives of the people being so tested. Medical screening needs to be driven
by the medical need, not the technological capacity.
Doctors have a professional responsibility to discourage techno-
logically driven screening and to ensure that all screening meets
the requirements set out in Box 2.12.1. Screening promoted only
in terms of the application of a particular technology should not be
part of medical practice.
FURTHER READING
Abu-Helalah M, Law MR, Bestwick JP, Monson JP, Wald NJ (2011). A
randomized double blind cross-over trial to investigate the efficacy of
screening for adult hypothyroidism. J Med Screen, 17, 164–9.
Ashton HA, et al. (2002). The Multicentre Aneurysm Screening Study
(MASS) into the effect of abdominal aortic aneurysm screening
on mortality in men: a randomised controlled trial. Lancet, 360,
1531–​9.
Atkins WS, et al. (2010). Once-​only flexible sigmoidoscopy screening
in prevention of colorectal cancer: a multicentre randomised con-
trolled trial. Lancet, 375, 1624–​33.
Breslow N, et al. (1977). Latent carcinoma of prostate at autopsy in
seven areas. Int J Cancer, 20, 680–​88.
Holland WW, Stewart S (2005). Screening in disease prevention.
Nuffield Trust, Radcliffe Publishing, Oxford.
Journal of Medical Screening:  screening briefs (1994a, 1994b, 1995,
1996, 1997). J Med Screen, 1, 73; 1, 255; 2, 126; 3, 110; 4, 54.
Law MR, Morris J, Wald NJ (1994). Screening for abdominal aortic
aneurysms. J Med Screen, 1, 110–​16.
McKeown T (1968). Validation of screening procedures. In: Screening
in medical care. Reviewing the evidence. Nuffield Provincial Hospital
Trust, Oxford University Press, Oxford.
Mulhall BP, Veerappan GR, Jackson JL (2005). Meta-​analysis: com-
puted tomographic colonography. Ann Intern Med, 142, 635–​50.
National Lung Screening Trial Research Team (2011). Reduced lung-​
cancer mortality with low-​dose computed tomographic screening.
N Engl J Med, 365, 395–​409.
Palomaki GE, et  al. (2012). DNA sequencing of maternal plasma
reliably identifies trisomy 18 and trisomy 13 as well as Down
syndrome:  an international collaborative study. Genet Med, 14,
296–​305.
Scoen RE, et  al. (2012). Colorectal-​cancer incidence and mortality
with screening flexible sigmoidoscopy. N Engl J Med, 366, 2345–​57.
Thorner RM, Remein QR (1961). Principles and procedures in the eval-
uation of screening for disease. Public Health Monograph No 67.
Public Health Service Publication No 846. US Department of Health
Education and Welfare, Washington, DC.
Wald DS, Bestwick JP, Morris JK, Whyte K, Jenkins L, Wald NJ (2016).
Child-parent familial hypercholesterolemia screening in primary
care. N Engl J Med, 375, 1628–37.
Wald NJ (1994). Guidance on terminology. J Med Screen, 1, 76.
Wald NJ (2004). The epidemiological approach: an introduction to epi-
demiology in medicine, 4th edition. Wolfson Institute of Preventive
Medicine/​Royal Society of Medicine Press, London.
Wald NJ, Cuckle H (1989). Reporting the assessment of screening and
diagnostic tests. Br J Obstet Gynecol, 96, 389–​96.
Wald NJ, Hackshaw AK, Frost CD (1999). When can a risk factor be
used as a worthwhile screening test? BMJ, 319, 1562–​5.
Wald NJ, Leck I (eds) (2000). Antenatal and neonatal screening, 2nd
edition. Oxford University Press, Oxford.
Wald NJ, Law MR (2003). A strategy to reduce cardiovascular disease
by more than 80%. BMJ, 326, 1419–​23.
Wald NJ, et al. (2004). SURUSS in perspective. Br J Obstet Gynaecol,
111, 521–​31.
Wald NJ, Morris JK, Rish S (2005). The efficacy of combining several
risk factors as a screening test. J Med Screen, 12, 197–​201.
Wald NJ,  et al.  (2018). Prenatal reflex DNA screening for trisomies 21,
18, and 13. Genet Med, 20, 825–30.
Wilson JMS, Jungner G (1968). Principles and practice of screening
for disease. WHO Public Health Paper No. 34, World Health
Organization, Geneva.

2.13 Health promotion 152
2.13 Health promotion 152
ESSENTIALS
The ‘Ottawa Charter for Health Promotion’ (1986) remains a bench-
mark for the global health promotion community, but the context
for health promotion has changed with increasing recognition of
the significance of inequities in health. Health promotion is a key
strategy to deal with the social determinants of health that create
these inequities. Attention has shifted from the mere recognition
that all public policies may impact on health to active strategies
and actions to move health concerns into all policies. Clinicians are
key actors in shaping social and cultural priorities and beliefs: they
should be committed to the reduction of health inequity, with
health promotion as a core commitment and responsibility. ‘Why
treat people, and then send them back to the conditions that made
them sick?’ (Sir Michael Marmot).
Introduction
Investing in better health and greater resilience of patients, families,
and communities is a key responsibility of the health professional.
This is a challenge that is quite different from providing services and
interventions to return sick individuals to health, or engage in pri-
mary prevention efforts (see Chapter 2.11) for those at immediate
risk of developing disease.
Building the health of individuals and populations has tradition-
ally been the remit of public health services, the evolution of which
has been described as having happened in waves, but rather than
a new development superseding and replacing an earlier one, in
public health the different ideas have built on, and complemented,
each other. Fig. 2.13.1 shows the five public health developments
since the early 1800s.
Through the structural stage (e.g. installation of sanitation hard-
ware) and biomedical (e.g. surveillance and infection control) as well
as clinical (e.g. vaccination, screening) stages, public health—​by the
end of the 20th century—​embraced health as a social value and a
social and economic resource, rather than just an individual attribute.
Some parts of the global community have not wholly advanced
along all these five waves. In remote areas, even in wealthy countries,
the full social and cultural dimensions of health often have not at-
tained the same stature as they have in urban environments. Cities
generally, even in middle-​ and low-​income countries, offer oppor-
tunities for health that are better than elsewhere.
With the growth of the health industry (in many OECD countries
the most important contributor to local economies) it seems we have
moved even beyond this ‘social model of health’ and are embracing
a ‘health society’. In this society the pursuit of health is a lofty goal
and one that yields significant commercial and political interest.
Health has become culture. Yet, at the same time not everyone bene-
fits equitably from this health society. In fact, the health inequity
gap is growing, and remains persistent even in the most egalitarian
societies.
This chapter considers the components of modern health promo-
tion that respond to and address the health society. Health promo-
tion, as a concept, emerged from a more behaviourist approach to
health development called health education. In many countries and
cultures, the two concepts are still considered synonymous.
The Ottawa Charter for Health Promotion
The evolution of public health as described earlier, along with secular
social change, created impetus for the World Health Organization,
the Canadian Public Health Association, and Health Canada (the
Canadian federal ministry of health) to convene an international
conference in 1986. This conference on health promotion, with the
subtitle ‘the move towards a new public health’ (Fig. 2.13.2) de-
scribed priorities for health promotion in an ‘Ottawa Charter for
Health Promotion’. This Charter has remained a benchmark for the
global health promotion community. Follow-​up global conferences
(https://www.who.int/healthpromotion/conferences/en/) fine-​tuned
the components of the Charter and updated them to new contexts.
Historical developments
The development of the idea of health promotion was the result of
changes in society, and our thinking about society, since the 1960s.
Social movements emerged that claimed greater control over the
plight and destinies of certain populations. The women’s health
movement, for instance, put gender on social and clinical agendas
2.13
Health promotion
Evelyne de Leeuw
2.13  Health promotion
153
and expected full involvement and control of women in their own
health management. Since the early 1980s the scourge of HIV/​AIDS
created new forms of health activism. Greater attention was paid
to critical analyses of the workings of society and its core insti-
tutions, including clinical medicine, for which terms like the
‘medical-​industrial complex’ were coined. This was reflected in
philosophy and social science, and applying such gazes to the world
of health led, for instance, to the ‘discovery’ of iatrogenesis (chal-
lenges to health, or exacerbated disease conditions, created by the
actions of the disease care system).
In medical sociology a breakthrough development was Aaron
Antonovsky’s work on ‘salutogenesis’, the phenomenon in which
health is built (rather than disease prevented) by shaping condi-
tions for greater ‘sense of coherence’, which has three components.
First, comprehensibility, the sense of understanding events in life
and feeling able to predict what is likely to happen in the future.
Secondly, manageability, a belief that you will be able to take care
of things that might happen. Thirdly, meaningfulness, a belief that
things are worthwhile and that there is good reason to care about
what happens. Antonovsky found, in his research with Holocaust
survivors, that individuals with a greater sense of coherence suc-
cumb less easily to disease and chronic health conditions. In the
health promotion field, this is seen as ‘positive health’. It suggests
that clinical intervention to ‘fix’ disease or infirmity and restore
health is only the first stage on a continuum towards positive health.
Investing in better social, economic, and natural capital would create
individuals and communities with greater resilience.
In some realms of the prevention community, there was also dis-
satisfaction with the degree to which behavioural interventions
yielded limited health gain. In fact, some saw that misguided be-
havioural interventions in fact amplified adverse health conditions
through victim-​blaming and social exclusion. It became increas-
ingly recognized that different kinds of interventions (communica-
tive, facilitative, and regulatory) should complement each other and
in fact create synergy. For instance, mere knowledge provision on
seatbelt use would only work to advance health if cars are equipped
with such devices, and legislation mandates and sanctions their use.
This logic applies throughout the health spectrum: the account of
smallpox eradication shows that it was not just the clinical evidence
that surveillance and control made eradication possible. This feat
was only accomplished by a fortuitous alignment between clinical
competence, technological advance (the bifurcated needle), strict
rules, as well as social, political, and financial resourcing of the
effort.
The authority and sovereignty of governments and professionals
were also challenged. Globalization started to be seen as a threat to
the livelihood of local communities. Views on the role of government
changed to what some have called ‘laissez-​faire’ governance (and
others Thatcherism, Reaganomics, or New Public Management)
since the late 1970s. Many scholars and activists felt that things like
universal health coverage was eroded, sustainability and ecological
balance suffered, and social justice was tested. It was time for a new
vision on public health, and the Ottawa Charter provided it.
Ottawa Charter action areas
The Charter was developed over four days in the spirit that led to
the need for the conference: ‘experts’ from all walks of life, including
clinicians, scholars, and community representatives (albeit more
Structural
1800
1850
1920
1970
2000
Biomedical
Clinical
Social
Cultural
Fig. 2.13.1  Five stages of modern public health evolution since the 1800s.
154
section 2  Background to medicine
dominantly male and from richer countries than female and from
lower income states) worked to continually draft a simple, two-​page
message on priorities in action for the new public health.
Enable, mediate, advocate
The Charter recognizes that health is a resource for everyday life,
and that its conditions are created in everyday life. ‘Settings’ such
as school, work, sports clubs, the home, neighborhoods and cities,
may all challenge and enhance health. There is a role for those
who wish to promote health to enable individuals, groups, and
communities to control the determinants of their health; to me-
diate between groups and organizations for health; and to advocate
for health at all levels of policy and practice. This is a broader social
objective, not just a role for one discipline.
Reorient health services
The conference considered that most health services are in fact disease
services, and do very little to invest in positive health. Maybe worse,
in the 1980s most health services were not too proficient at patient
education and counselling and some had primitive ideas about how
Fig. 2.13.2  The cover of the Ottawa Charter for Health Promotion, including its logo describing action areas.
Reprinted from The Ottawa Charter for Health Promotion, First International Conference on Health Promotion, Ottawa, 21
November 1986.
2.13  Health promotion
155
to secure best patient outcomes. The dominant model adopted the
‘Trust me, I’m a doctor’ position. The Ottawa Charter identified that
health services, be they primary care posts, specialist hospitals, or ter-
tiary care facilities, are pivotal social institutions in the communities
they serve. Not just the health of their patients (or clients), but also
of staff, visitors, and their ecological footprint could be considered
important. Rather than disease cure institutions, they could become
health-​promoting health services including better, health enhancing
architectural design, community hubs, and healthy workplaces.
Supportive environments
The emerging discipline of social epidemiology had provided evidence
since the 1970s that the social, physical, natural, and economic envir-
onment impacts significantly on individual and community health.
Like Sir Michael Marmot, many years later (2008) at the launch of the
report of the WHO Commission on Social Determinants of Health
said: ‘Why treat people, and then send them back to the conditions
that made them sick?’, the Ottawa Charter urges individuals, groups,
communities, practitioners, and politicians to see how these envir-
onments contribute to making healthier choices the easier choices.
For instance, in most countries fat meat is cheaper than lean meat
and changing the commercial environment to facilitate the healthier
choices (e.g. through different slaughter procedures, subsidies, but
also advocacy) would yield easier access to healthy food. The suc-
cessful Heartbeat Wales programme inspired various aspects of the
health promotion vision, including the effective partnerships be-
tween diverse partners in the food chain to accomplish exactly this.
Heartbeat Wales significantly inspired the Charter from the early
1980s onward.
Personal skills and community action
There is no denying that health is mostly an individual attribute (al-
though René Dubos already in 1959 explicitly included society in
his definition: ‘Health is the expression of the extent to which the
individual and the social body maintain in readiness the resources
required to meet the exigencies of the future’—​a view that aligns
well with positive health). But health behaviour is not necessarily
individual, it is shaped through social norms and community op-
portunity. The Charter identifies skills (such as a prescient view on
health literacy) to deal with health challenges and opportunities, in-
dividually and collectively, as key to the success of health promotion.
Build healthy public policy
Many of the options and opportunities outlined here depend on
collective choice, or (as Nancy Milio analysed) on public policy out-
side the health sector. This idea had started to take hold in Canada
in the early 1970s with the Lalonde Report that advocated for ac-
tion on elements of the ‘health field’ outside medicine. This idea
was adopted by the Swedish government in the 1980s (Fig. 2.13.3)
when they identified how action in one government sector could
well have added effect across several health conditions: it has taken
decades of deliberate health promotion action to allow the country
to effectively implement its comprehensive health strategy. Milio
systematically reviewed the policy literature and showed health
impacts from all sectors under the influence of public policy (e.g.
including agriculture, food, and nutrition; education and literacy;
transportation and mobility; defence; economic development, and
so on). The Ottawa Charter was straightforward in its call (Box
2.13.1) that making the healthier choice should also be the easier
choice for all politicians.
A health promotion for the 21st century
The Ottawa Charter remains an important benchmark for global
health promotion. It has influenced policies and practices around
the world. On its 25th and 30th anniversaries the legacy of the state-
ment was reviewed. Progress has been made in virtually all action
areas, either in practice or in scholarship. There is, however, also evi-
dence that health services are still facing problems to redefine them-
selves as ‘health-​promoting environments’.
Strong correlation
Social upbringing environment
Mental illness
Tumours
Injuries
Respiratory diseases
Cardiovascular diseases
skeletomuscular disease
Some correlation
Social work environment and unemployment
Diet
Tobacco
Alcohol and drugs
Traffic
Air/water pollutants
Physical living environment
Social living environment
Physical work environment
Fig. 2.13.3  The Swedish government’s map of public sector impacts
on health and disease (HS90—​the Swedish Health Services for the
1990s, 1984).
Box 2.13.1  Ottawa Charter for Health Promotion view
of Healthy Public Policy
Building healthy public policy
Health promotion goes beyond healthcare. It puts health on the agenda
of policymakers in all sectors and at all levels, directing them to be aware
of the health consequences of their decisions and to accept their respon-
sibilities for health.
Health promotion policy combines diverse but complementary ap-
proaches including legislation, fiscal measures, taxation, and organiza-
tional change. It is coordinated action that leads to health, income, and
social policies that foster greater equity. Joint action contributes to en-
suring safer and healthier goods and services, healthier public services,
and cleaner, more enjoyable environments.
Health promotion policy requires the identification of obstacles to the
adoption of healthy public policies in non​health sectors, and ways of
­removing them. The aim must be to make the healthier choice the easier
choice for policymakers as well.
Reprinted from The Ottawa Charter for Health Promotion, First International
Conference on Health Promotion, Ottawa, 21 November 1986.
156
section 2  Background to medicine
The context for health promotion also has changed. The debate
around health equity has intensified, with many countries around
the world (led by WHO) having inequalities in health on national
and local agendas. Health promotion is still seen as a key strategy
to deal with the social determinants of health that create these in-
equities. Attention has shifted from the mere recognition that all
public policies may impact on health to active strategies and actions
to move health concerns into all policies. This is called ‘Health in All
Policies’ and cunning leadership of the health sector and its profes-
sionals is identified as critical to the future success of health promo-
tion and global health.
A clinical commitment and challenge
For public health doctors it will be clear what their roles and commit-
ments in health promotion are. They mediate, advocate, and enable
diverse groups of stakeholders that shape environments for health to
be explicit about the health impacts and choices available to the in-
dividuals, groups, and communities they serve. In this way they will
contribute to the development of Health in All Policies, and support
people in individual and community action for health.
For the clinician dealing with individual patients, the responsibility
may be less clear. The concepts discussed in this chapter may be con-
sidered abstract and divorced from the individual complaint voiced
in the surgery, clinic, or ward. However, clinicians are key actors in
shaping social and cultural priorities and beliefs, and their commit-
ment to the reduction of health inequity should be in no way less than
those of public health specialists.
The British Medical Association responded to this challenge
and presented its members, and the rest of the world, with a grid
that sees levels of clinical action, both geographically (local and
international) and in terms of performance (treating patients—​
community leadership—​advocacy—​research)—​see Table 2.13.1.
To contribute to better health of individuals and communities,
and to reduce inequities in health between groups, there are im-
portant roles for doctors:
•	Practice holistic medicine from a patient-​centred perspective
•	 Embrace and exploit the social and community stature of the medical
profession in the community to bring together partners for health
•	Advocate for, do not just treat, marginalized and socially excluded
individuals and their communities
Health promotion is a core commitment and responsibility of the
clinician. As the Declaration of Geneva states: I will not permit con-
siderations of age, disease or disability, creed, ethnic origin, gender, na-
tionality, political affiliation, race, sexual orientation, social standing,
or any other factor to intervene between my duty and my patient.
FURTHER READING
Baum F, Lawless A, Williams C (2013). Health in All Policies from
international ideas to local implementation: policies, systems and
organizations. Health promotion and the policy process: practical and
critical theories, 188–217.
British Medical Association (2011). Social determinants of health—​
what doctors can do. British Medical Association, London.
de Leeuw E (2011). The boulder in the stream. Editorial, ii157–​ii160.
In:  de Leeuw E (ed) The Ottawa Charter 25 Years On. Health
Promotion International, 26(suppl 2), ii157–​ii272.
de Leeuw, E (2017). Engagement of sectors other than health in in-
tegrated health governance, policy, and action. Annual Review of
Public Health, 38, 329–49.
de Leeuw E, Ståhl T, Tang KC (eds) (2014). The eighth global confer-
ence on health promotion. Helsinki, June 2013. Health Promotion
International, 29(suppl 1), i1–​i151.
Dubos R (1959). Mirage of health:  utopia, progress, and biological
change. Doubleday, Garden City, NY.
Eriksson M, Lindström B (2006). Antonovsky’s sense of coherence
scale and the relation with health: a systematic review. J Epidemiol
Community Health, 60(5), 376–​81.
Huber M, et al. (2011). How should we define health? BMJ, 343, d4163.
Milio N (1981). Promoting health through public policy. FA Davis Co,
Philadelphia, PA.
Newby L, Denison N (2014). ‘If you could do one thing.’ Nine local
actions to reduce health inequalities. The British Academy, London.
Pelikan JM, et al. (2005). 18 core strategies for health promoting hos-
pitals (HPH). In: Groene O, Garcia-​Barbero M (eds) Health promo-
tion in hospitals: evidence and quality management. World Health
Organization, Copenhagen.
Ståhl T, et al. (2006). Health in all policies. Prospects and potentials.
Finnish Ministry of Social Affairs and Health, Helsinki.
Ziglio E, Simpson S, Tsouros A (2011). Health promotion and health
systems: some unfinished business. Health Promot Int, 26(suppl 2),
ii216–​ii25.
Table 2.13.1  What doctors can do in health promotion to address
social determinants of health and health equity (BMA, 2011)
Treating
patients
Community
leadership
Advocacy
Research
Local
A
B
C
D
Regional
E
F
G
H
National
I
J
K
L
International
M
N
O
P

2.14 Deprivation and health 157
2.14 Deprivation and health 157
ESSENTIALS
The health of an individual is influenced by the circumstances in
which he or she lives. Individuals who live in poverty are more
likely to be unhealthy and die younger than individuals who are
wealthy.
The mechanism by which complex social circumstances cause
health inequalities might be a failure to create capacity to manage
life’s challenges. Most agree that well-​being is created where indi-
viduals have an optimistic outlook, a sense that they are in control of
their own lives, a sense of purpose and meaning in life, confidence
in their ability to deal with problems, a supportive network of friends
and a nurturing family.
If society is serious about tackling health inequalities, action
needs to happen across the life course. Children from families who
have no expectation of success need encouragement and, when
they fall into addiction or criminality, society needs to think about
rehabilitation rather than punishment. Inclusion into society and
building empathy is the basis for narrowing inequality.
Early observations
The industrial revolution was marked by increasing disparity in
wealth across society. As the health of the rich improved, the health
of the working classes was often damaged by the conditions in which
they worked. In 1842, the social reformer Edwin Chadwick pub-
lished a study in which he noted that labourers, on average, died
around 20 years younger than members of the professional classes.
Three years later, Friedrich Engels published a description of life in
cities in Northern England. He reported that the annual death rate
in mill towns such as Manchester and Liverpool was significantly
higher than the national average (around 1 in 32 compared with an
average of 1 in 45).
In Glasgow, a city at the heart of the industrial revolution, city
officials published in 1861 data linking infant mortality with levels
of affluence. In the poorest areas of the city, infant mortality was 18
times higher than that seen in the wealthiest families.
Over the last 150  years, it has become apparent that the rela-
tionship between poverty and health is complex and involves
many aspects of society. It has now become customary to talk of
‘socioeconomic determinants of health’ and, where these determin-
ants are unequally distributed, to see the resultant health inequality
as a matter requiring action.
The World Health Organization defines the social determinants
of health as: ‘the conditions in which people are born, grow, work,
live, and age, and the wider set of forces and systems shaping the
conditions of daily life. These forces and systems include economic
policies and systems, development agendas, social norms, social
policies and political systems.’
Health inequalities in the 21st century
There are inequalities in life expectancy between countries, within
countries, and within individual towns and cities. Internationally,
the gap between industrialized countries and the low-​income
countries narrowed between 1960 and 1990 from around
23  years to approximately 12  years. The gap between the two
groups in infant mortality was 123 (per 1000 live births) in 1960
and this narrowed to 61 by 1990. The exception to this improving
picture was in some African countries where mortality from AIDS
caused a fall in life expectancy
By 2010, the OECD countries had attained an average life expect-
ancy at birth for the whole population of 79.7 years, a gain of almost
10 years since 1970. Japan had the highest life expectancy and most
OECD countries had attained a life expectancy at birth of at least
80 years.
Within the United Kingdom, life expectancy in England, Scotland,
and Wales continues to improve, although there are differences
between the three countries with Scotland having the lowest life
expectancy for both men and women (Table 2.14.1).
Despite increasing life expectancy across the United Kingdom,
significant variations between districts are still apparent. In
Kensington and Chelsea, the wealthiest part of London, a man can
expect to live to 88 years, while a few kilometres away in one of
London’s poorer areas, Tottenham Green, male life expectancy is 71.
2.14
Deprivation and health
Harry Burns
158
section 2  Background to medicine
The causes of health inequality
Health in any society is determined by the complex interaction of
many factors, hence attempts to attribute inequality to any one factor
are likely to be an oversimplification. However, several explanations
have received attention over the years.
Access to healthcare
One commonly held belief is that reduced life expectancy among the
poor is due to inequality in access to healthcare. In 1980, the UK gov-
ernment published a report by Sir Douglas Black. He had been asked
to answer the question of why, after 30 years of a National Health
Service, providing universal, free access to healthcare, there were
persistent inequalities in health in the UK population. Black was
able to discount this theory, saying healthcare was primarily to treat
illness when it occurred, not to prevent its occurrence. However, it
remains possible that, in countries without universal, free health
services, this explanation could contribute to inequalities.
Other, more plausible explanations for inequalities in the United
Kingdom were considered in Black’s Report. These included the
following.
Natural or social selection
Could it be that, rather than social position affecting health, health
actually determines social position? People with poor health might
be less effective in the workplace and gradually drift down the occu-
pational hierarchy. However, it was generally accepted that this was
not a credible explanation for the pattern of inequality observed in
the United Kingdom at that time.
Cultural or behavioural factors
This explanation focuses on the choices made by individuals about
their health. Use of alcohol, tobacco, lack of exercise, and consump-
tion of unhealthy foods are more common at the lower end of the
social scale. The resulting poor health is often seen as the consequence
of a lack of education or simply a lack of interest in remaining healthy.
However, a cohort study that followed British civil servants over a long
period of time, collecting data on risk factors such as body weight,
cholesterol, smoking, and blood pressure, found that risk factors
could only explain one-​third of the observed variation in health.
Material factors
Poor people are more likely to live in overcrowded homes of poor
quality. They may work in noisy, polluted environments with inad-
equate safety regulations. They will be less financially secure and
may be stressed by debt and difficulty paying bills. All of these fac-
tors can adversely affect health.
Psychosocial factors
If we feel overwhelmed by events around us, we will be stressed, and
stress has behavioural and biological consequences. This makes it
difficult to separate behavioural, structural, and psychosocial factors
as causes of inequality.
One way of unravelling the complexity of the relationship is to
understand the biological mechanisms through which social cir-
cumstances influence health.
The biology of deprivation
Many studies report high levels of stress hormones in people
at the lower end of the socioeconomic scale. A normal stress re-
sponse allows us to respond to challenge. Sitting an exam, moving
house, or an accident will produce a short-​lived elevation in
stress hormones which will return to normal once the challenge
is overcome. Many stressors are less easy to resolve. Unexpected
bills, work conflicts, family crises are persistent and often difficult
to manage. In poorer areas, problems may include noise, damp
housing, discrimination, and living in a violent and dangerous
neighbourhood.
Greater resilience allows us to manage successfully everyday
problems. This is particularly true for those who are more affluent
and can afford to live in a safe environment. Chronically activated
stress responses appear commoner among those in a poorer social
position and this chronic stress burden may interfere with health
in several ways which ultimately impair well-​being and long-​term
survival (Fig. 2.14.1).
Low socioeconomic status
Adverse environments
Poverty
Lack of social support
Chronic stress
Biological damage
Lack of control
Poor health choices
Poor health
Fig. 2.14.1  Mechanisms by which low socioeconomic status may lead
to poor health.
Table 2.14.1  Life expectancy at birth by year of birth
Year
Men
Women
England
Scotland
Wales
England
Scotland
Wales
07/​09
78.2
75.3
77.2
82.3
80.1
81.5
08/​10
78.5
75.8
77.6
82.5
80.3
81.8
09/​11
78.9
76.2
78.0
82.9
80.6
82.2
10/​12
79.2
76.5
78.2
83.0
80.8
82.2
11/​13
79.4
76.8
78.3
83.1
80.9
82.3
2.14  Deprivation and health
159
The mechanism by which complex social circumstances cause
health inequalities might be a failure to create capacity to manage
life’s challenges.
Salutogenesis—​the creation of health
The American sociologist, Aaron Antonovsky, studied the question
‘How do people manage stress and stay well?’. Some people main-
tain health despite exposure to potentially overwhelming stressors.
‘Salutogenesis’ was the term he used to describe those factors that
support human health and well-​being, rather than the factors that
cause disease. Salutogenesis represents a counterbalance to patho-
genesis with which medicine is more concerned.
Antonovsky’s theory is only one of many concepts that might
underpin the processes of health creation. Emotional intelligence,
internal locus of control, empowerment are a few of many theories
that have been advanced to explain why some individuals have an
increased capacity for well-​being. Some attributes are common to
several of these theories. Most agree that well-​being is created where
individuals have:
•	An optimistic outlook
•	A sense that they are in control of their own lives
•	A sense of purpose and meaning in life
•	Confidence in their ability to deal with problems
•	A supportive network of friends
•	A nurturing family
Antonovsky’s work is of special interest since he offers a link between
the psychosocial drivers of inequality and their biological conse-
quences. Central to Antonovsky’s theory is the concept that, early
in life, we acquire a mental outlook which he described as ‘having
a sense of coherence’. This he described as a set of psychological at-
tributes which he identified as being associated with positive health
outcomes. They include seeing the world as:
•	Comprehensible: having a belief that things happen in an orderly
and predictable fashion, a sense that you can understand events in
your life and predict what will happen in the future.
•	Manageable: confidence that you have the skills, ability, support,
and resources necessary to manage events in your life.
•	Meaningful:  believing that things in life are interesting and a
source of satisfaction, and that there is good reason or purpose to
care about what happens.
Failure to develop a sense of coherence in early childhood, Antonovsky
suggested, would result in the individual being chronically stressed.
His thinking hinted at an explanation for the physical consequences
of psychosocial adversity.
The early origins of well-​being
The basis for a salutogenic outlook is laid down in early childhood
and relates to the way the child learns to manage stress. A positive
stress response might be caused by exposure to a new caregiver or an
immunization but, so long as the child attaches to an adult as part of
a safe, consistent, nurturing relationship, they learn that the world is
a comprehensible and manageable place.
However, when a child experiences frequent, or prolonged
adversity—​such as physical or emotional abuse, or chronic neglect
due to parental substance misuse or mental illness—​such prolonged
activation of the stress response can be associated with altered brain
development. Specifically, abnormalities of the prefrontal cortex,
hippocampus, and amygdala make it harder to suppress inappro-
priate behaviour, impair learning, and increase the risks of mental
and physical health problems throughout the life course. Changes
to the hippocampus also impair the individual’s capacity to manage
stress. Across the life course, adverse childhood events increase the
risk of poor health, failure in education, increased risk of offending,
and a criminal record as well as poor health.
Antonovsky did not claim that sense of coherence alone could
produce health. He also argued that individuals needed access to ex-
ternal supports that allowed them to be resilient in the face of severe
challenge. These resources include assets such as money and social
support. These are precisely the factors which those living in poverty
and chaos are lacking. Fig. 2.14.2 outlines Antonovsky’s concept of
health creation.
Resources which
enhance resilience
Sense of coherence
Events
Stress
Tension
Resolution
Well-being
Seeing the world as:
Structured
Predictable
Feeling that it is:
Manageable
Meaningful
Wanting to engage
Family
Nurture
Intelligence
Work
Material resource
Identity
Cultural stability
Stable set of answers
Optimism
Fig. 2.14.2  Antonovsky’s concept of health creation.
160
section 2  Background to medicine
How might health inequalities be improved?
Healthy brain architecture depends on a foundation built by ap-
propriate attachment formed between a child and stable, caring
adults. If an adult’s responses to a child are unreliable, inappro-
priate, or simply absent, the lifetime risk of failure is increased.
Adverse childhood events can set in motion a lifetime of failure.
Mental health problems in childhood can lead to school failure,
poor educational record, and unemployment. One cohort study
suggested that childhood abuse and neglect could have a signifi-
cant impact on the likelihood of arrest for delinquency and vio-
lence. By the age of 32, almost half of the victims of abuse and
neglect were arrested for a non-​traffic offence. Criminal conviction
enhances the likelihood of unemployment and poverty, and the
individual becomes increasingly alienated from society. Domestic
violence damages children and the cycle of alienation continues
(Fig. 2.14.3).
If society is serious about tackling health inequalities, action
needs to happen across the life course. It needs to begin by ensuring
that babies have the best possible start in life. Supporting parents
who have not themselves experienced a nurturing childhood to
nurture their children is a necessary first step. Children from
families who have no expectation of success need encouragement
and, when they fall into addiction or criminality, society needs
to think about rehabilitation and repair rather than punishment.
Inclusion into society and building empathy is the basis for nar-
rowing inequality.
FURTHER READING
Center on the Developing Child. Key Concepts. http://​developingchild.
harvard.edu/​
Kawachi I, Subramanian V, Almeida-​Filho N (2002). A glossary for
health inequalities. J Epidemiol Community Health, 56, 647–​52.
Lindstrom B, Eriksson M (2005). Salutogenesis. J Epidemiol Community
Health, 59, 440–​2.
Marmot M, et  al. (2010). Fair Society, Healthy Lives (The Marmot
Review). http://​www.instituteofhealthequity.org/​resources-​reports/​
fair-​society-​healthy-​lives-​the-​marmot-​review
Marmot MG (2015). The health gap: the challenge of an unequal world.
Bloomsbury Press. ISBN 1632860783.
McEwen BS, Morrison JH (2013). The brain on stress: vulnerability
and plasticity of the prefrontal cortex over the life course. Neuron,
79, 16–​29.
Initial event
Unequal outcomes
Alienation
Consequences
Loss of self efficacy,
self esteem, sense of
control
Worklessness
Poverty
Failure in
education
offending
health
Mental
health
problems
Chaotic early
years
Fig. 2.14.3  The cycle of alienation.

2.15 How much should rich countries’ governments s
2.15 How much should rich countries’ governments spend on healthcare? 161
ESSENTIALS
The argument commonly made by politicians in richer countries is
that a universal health system is unaffordable because of the enor-
mous gains in life expectancy leading to an ageing population, and
that market-​driven competition and choice is preferred because it in-
creases efficiency. There is no evidence to support these claims, and
evidence to the contrary on both counts. The ageing of the popula-
tion contributes little to the overall growth in expenditure, which is
largely attributable to increasing marketization of health systems and
costs of medicines and technologies.
Since 1948, the United Kingdom has had a universal integrated
public health system free at the point of delivery and funded through
central taxation. The UK NHS became the model for many countries’
health systems across the world, as having the lowest cost, most effi-
cient and fairest system, and guaranteeing healthcare to all its citizens
without fear of charges or denial of care. In 2012, after two decades
of market incrementalism, the universal public model was abolished
in England (although Scotland and Wales still retain it) in favour of a
model which increasingly resembles the non​universal market form
of the United States where risk selection and denial of care prevails.
The United States is one of the richest countries in the world and
has the most expensive healthcare, but in spite of that it denies more
than one in five of its population access to healthcare. Overtreatment
and denial of care, catastrophic costs, and spiralling health expend-
iture, go hand in hand in the United States. Those countries that
have adopted the US model of mixed funding and private provision
have more marketization, the greatest inequalities in access, lack
of coverage, and highest out-​of-​pocket payments because market
models operate on the basis of risk selection and risk avoidance.
We conclude that the decision to have a universal public
healthcare system is political. Many countries have decided that
universal healthcare is the hallmark of a civilized society and that it
is both necessary and affordable for governments to legislate for its
citizens to that end. The question of how much any country should
spend is inextricably linked to the chosen model of funding and
provision, the degree of marketization, and how much risk selec-
tion and denial of care a government is prepared to tolerate in its
health system.
Introduction
Among many high-​income countries, healthcare spending has been
in relative decline since 2007, when the latest economic recession
began. According to the Organisation for Economic Cooperation
and Development (OECD), a grouping of the richest countries in
the world,1 a third of developed world governments reported a ‘real
term cut in overall health spending in 2013’. In Italy, Portugal, and
Greece real reductions have been taking place for several years. At
the same time, checks on future spending levels are being discussed
throughout the European Union. A common claim is that advanced
economies can no longer afford universal health systems, that is, sys-
tems which guarantee a right of access to healthcare for everyone in
need of it. What should higher income countries spend?
The relative decline in health spending has been accompanied by
a growing tendency to treat this question as a technical and non-​
political issue. According to the International Monetary Fund
(IMF), reform is needed because public health spending is ‘an im-
portant macro-​fiscal issue’: ‘Public spending on health care has been
a key driver of aggregate increases in public spending over the past
40 years. [ . . . ] [S]‌pending is projected to continue rising as a share
of gross domestic product (GDP) unless reforms are undertaken
to help break these trends’. Both the European Commission (EC)
and the IMF have advised or forced some of the countries worst af-
fected by the recession and the aftershock of the banking crisis in
2007–​2008 to cut health spending, and Greece to abandon universal
healthcare, in return for financial assistance.
Throughout the developed world this technical rationale for
what has become known as ‘austerity’ is bolstered by the hypothesis
that rich countries are victims of their own success in improving
the life expectancy of their citizens. A longer-​lived population will
2.15
How much should rich countries’
governments spend on healthcare?
Allyson M. Pollock and David Price
1  The OECD consists of the following high-​income countries:
Australia, Austria, Belgium, Canada, Chile, Czech Republic, Denmark,
Estonia, Finland, France, Germany, Greece, Iceland, Ireland, Israel,
Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Poland,
Portugal, Republic of Korea, Slovakia, Slovenia, Spain, Sweden,
Switzerland, United Kingdom of Great Britain and Northern Ireland,
United States of America.
162
section 2  Background to medicine
contain a higher proportion of elderly people than a shorter-​lived
one and this, it is argued, has created public budgetary pressures that
threaten economic well-​being. The proffered solution to an ‘ageing
society’ is to cap or reduce public health budgets so that they take up
a declining (or no greater than at present) share of national wealth.
However, decisions about future levels of spending on healthcare
are fundamentally political, or normative, just as are decisions on
public expenditure on defence or international development. Today,
most governments of the wealthiest countries spend between 7 and
10% of their national wealth on healthcare: the United States is an out-
lier at almost 18% (Fig. 2.15.1), and judging how much they should
spend in the future is now a pressing political question. To address it
we examine, first, the political basis of austerity policy. Secondly, since
all normative claims have testable elements or can be reframed so that
they are testable, we examine the concepts and empirical evidence that
can be used in debates about future levels of spending. To simplify mat-
ters we look at the role of politics and austerity through four research
lenses, namely, public health, law, finance, and economics and trade.
The role of politics and austerity in determining
healthcare expenditure
Austerity policy is based on claims that are no less political than
those that inform alternative spending policies. This is because
the econometric models that underpin the policy of reducing
public expenditure and public services involve implicit and con-
testable value judgements about the benefits of free markets and
of a diminishing role for government or relatively low public ex-
penditure. Standard economic theory, according to Paul Krugman,
prescribes increased public spending during a recession in order to
stimulate demand, even at the expense of growing public deficits
(in which governments spend more than they collect). However,
the dominant international political policy position known as the
‘Washington consensus’ has been to support the adoption of pri-
vate markets with relatively low government involvement in place
of tax funded public services.
Following the banking crisis in 2008, many governments pursued
the policy of reducing health expenditure. Those European Union
(EU) countries worst affected by austerity—​Italy, Greece, Spain,
Portugal, and Ireland—​all experienced severe reductions in total
and public and private per capita spending on healthcare. The EC
and the IMF have required some countries to reduce healthcare
spending substantially in exchange for financial assistance. Since
2008, the UK government has opted for substantial cuts in public ex-
penditure, provoking criticism from some leading economists that
this was less conventional economics than ideological commitment
to the private sector and markets.
Since 2000, alternative hypotheses about economic develop-
ment have been used to generate testable questions about or rebut
Health expenditure 2014 for selected OECD countries plus Argentina, Brazil, Costa Rica, Cuba, Dominican Republic,
Peru and Venezuela
Total health expenditure as a percentage of gross domestic product (public and private) by country
18.0
16.0
14.0
12.0
10.0
8.0
6.0
4.0
2.0
0.0
USA
Sweden
Switzerland
France
Germany
Austria
Cuba
New Zealand
Netherlands
Denmark
Belgium
Canada
Japan
Norway
Finland
Portugal
Australia
Costa Rica
Italy
United Kingdom
Spain
Brazil
Greece
Chile
South Korea
Peru
Venezuela
Argentina
Dominican Rep
Private
Public
Fig. 2.15.1  Health expenditure for OECD countries plus some Latin American countries.
Data from World Health Organization Global Health Expenditure Database.
2.15  How much should rich countries’ governments spend on healthcare?
163
claims that public spending on healthcare is an economic burden
that impairs future rates of economic development. There is now a
large body of research that shows that growing inequalities in in-
come and health status themselves create economic burdens that
retard growth, and that investing in healthcare can stimulate the
economy.
Several studies have examined the costs of high levels of inequality
for developed economies. Wilkinson and Pickett’s The Spirit Level
(2009) and Thomas Piketty’s Capital in the Twenty-​first Century
(2014) challenge the hypothesis that health spending is an eco-
nomic burden. Adding weight to the analysis is the World Health
Organization’s (WHO) Macroeconomics and Health Commission
final report (2003) and the EU’s Investing in Health (2013), the au-
thors of which argued that investing in health is a desirable eco-
nomic goal because it ‘contributes to the Europe 2020 objective of
smart, sustainable and inclusive growth’. Contrary to the IMF, the
EU maintains that health spending is now ‘recognised as growth-​
friendly expenditure’.
In a further sign of a changing policy arena, the United Nations
Conference on Trade and Development suggested in 2014 that a
more flexible approach was needed to public spending and govern-
ment institutions than allowed under the Washington consensus
(UNCTAD, 2014).
A second argument for the political basis of austerity policy arises
from findings that economic forecasts have proved to be badly
wrong. In April 2014, IMF research staff published an acknowledge-
ment that their econometric models had proved to be wide of the
mark after it was found that those countries that had implemented
the deepest public spending cuts subsequently experienced the
lowest levels of economic growth.
Thirdly, while the ageing society hypothesis has taken on a
mythical status among politicians, it has been rebutted in a series
of studies around the world. Ageing, it turns out, is too gradual to
explain anything more than quite marginal changes in total health
spending. This is because most people’s healthcare costs are incurred
in the last six months of life at whatever age they die. Additionally,
there is a body of research which shows that not only do older people
contribute more financially and in kind than they take out of the
system, but they are living longer and healthier and productive lives
(the compression of morbidity). In any event, the ageing hypothesis
lacks plausibility when used to suggest that a finite population has
infinite healthcare needs.
In the United Kingdom, for example, older people have accounted
for a relatively small proportion of the increase in spending on
healthcare, and while overall spending on health services between
1965 and 1999 grew by 3.8% a year in real terms, demographic
change (ageing) accounted for a real increase of just 0.5% a year in
sharp contrast to the cost of pharmaceuticals and technology, which
rise well above inflation.
Background—​trends in health spending in richer
and some middle-​income countries (OECD area
plus some Latin American countries)
Fig. 2.15.1 and Table 2.15.1 show total spending on healthcare as a
proportion of GDP for 2013/2014.
Poor-​quality data on private spending
It is important to note that the quality of data on private spending
in both rich and poor countries is limited by the fact that they are
based on household surveys, are not comprehensive, and often lack
comparability. Moreover, most data reflect spending by or on those
who use services, not those who are excluded, and data on the poor
and workers in informal economies might not be collected. The re-
liability of data about unmet need and inequality remains a critical
issue in all countries.
Allowing for these limitations, the spending data show:
1.	 That there are political differences among all countries in the
total health expenditure as a proportion of GDP, and in the rela-
tive spending on public and private healthcare. The United States
is an outlier among all countries, with the highest public and
private expenditure.
2.	 That national (non-​insurance-​based) health systems (United
Kingdom, Norway, Denmark, Italy, New Zealand, Sweden) have
Table 2.15.1  Health expenditure 2014 for selected OECD countries
plus Argentina, Brazil, Costa Rica, Cuba, Dominican Republic, Peru,
and Venezuela as a percentage of gross domestic product
Country
Public
Private
Total
United States
8.3
8.9
17.1
Sweden
10.0
1.9
11.9
Switzerland
7.7
4.0
11.7
France
9.0
2.5
11.5
Germany
8.7
2.6
11.3
Austria
8.7
2.5
11.2
Cuba
10.6
0.5
11.1
New Zealand
9.1
1.9
11.0
Netherlands
9.5
1.4
10.9
Denmark
9.2
1.6
10.8
Belgium
8.2
2.3
10.6
Canada
7.4
3.0
10.4
Japan
8.6
1.7
10.2
Norway
8.3
1.4
9.7
Finland
7.3
2.4
9.7
Portugal
6.2
3.3
9.5
Australia
6.3
3.1
9.4
Costa Rica
6.8
2.5
9.3
Italy
7.0
2.3
9.2
United Kingdom
7.6
1.5
9.1
Spain
6.4
2.6
9.0
Brazil
3.8
4.5
8.3
Greece
5.0
3.1
8.1
Chile
3.9
3.9
7.8
South Korea
4.0
3.4
7.4
Peru
3.3
2.2
5.5
Venezuela
1.5
3.7
5.3
Argentina
2.7
2.1
4.8
Dominican Rep
2.9
1.4
4.4
164
section 2  Background to medicine
similar levels of total spending to insurance-​based countries, but
relatively higher levels of public spending, expressed as a propor-
tion of GDP, while private spending on health is relatively low in
these countries.
3.	 Countries with high levels of private spending may have mul-
tiple public and private insurance systems.
4.	 Latin American countries also show major differences in public
and private expenditure which are related to the design of their
health systems. For example, Costa Rica and Cuba both have
high levels of public expenditure and universal tax-​funded
public health systems, whereas Brazil, the Dominican Republic,
and Chile have mixed funding and private insurance systems
which are not universal.
Spending levels
The perspective of public health
Public health sciences seek to understand, among other things, the
best ways in which to maximize population health status and meet
public health needs through fairness of funding and services. Within
the international public health community, universal access to
healthcare has been adopted as the preferred goal of rational health
systems for the last 45 years (World Health Organization, 1978).
This is the case because it is now accepted that although healthcare
is only one of several factors contributing to population health, uni-
versal access to it makes a measurably greater contribution than
more limited access.
In 2015, the two major international health agencies, the WHO and
the World Bank Group, jointly announced that universal healthcare
‘is a critical component of the new Sustainable Development Goals
(SDGs)’, which have replaced the Millennium Development Goals
as the standard against which to measure policy progress in popula-
tion health. Included among SDGs is a target to ‘Achieve [ . . . ] access
to quality essential healthcare services and access to safe, effective,
quality and affordable essential medicines and vaccines for all’
(World Health Organization and World Bank, 2015). The agencies
declare: ‘Supporting the right to health and ending extreme poverty
can both be pursued through universal health coverage.’
Goals such as the SDGs therefore provide a framework based on
public health science for normative debate about health spending.
From them it follows that governments committed to the public
health goal of maximizing population health status should spend
sufficient to ensure universal access to healthcare.
The perspective of laws and conventions
National laws, constitutions, and international conventions on
human rights provide a second framework for debate. Legal analysis
can be used to determine to what extent domestic laws may influ-
ence governments’ healthcare budgets. In most developed coun-
tries, governments are required by primary legislation to fund public
healthcare of some description, and within reason to vote through
sufficient money to discharge that duty. For example, the parliamen-
tary act of 1946 which brought in the NHS throughout the United
Kingdom, committed the UK government to funding and providing
key listed services for everyone throughout the United Kingdom,
a duty that was abolished by the Health and Social Care Act 2012.
Similar laws can be found in other countries with universal sys-
tems. The Danish Health Act (amended June 2010) requires the gov-
ernment to meet, among other things, ‘the need for [ . . . ] free and
equal access to health care’. The objective of Sweden’s health act (the
Health and Medical Services Act 1982:763) is ‘care on equal terms
for the entire population’. Italy’s national health service law requires
that health benefits must be guaranteed to all citizens. The EU,
which shares with member states responsibility for healthcare, has
adopted ‘principles of equality of access and solidarity in funding ar-
rangements, whether that is primarily through taxation or through
regulated social insurance’. These principles are applied through the
European Court of Justice, the judgements of which are superior to
national courts.
National laws can be amended and revoked, however, and simply
describing them may provide insufficient guidance about duties.
Spain abolished its national health service by royal decree in 2010.
England’s NHS legislation was changed in 2012 so as to absolve the
government of the day from the duty to meet specific service re-
quirements for everyone throughout England. Legal analysis was
crucial to show the scale and nature of the proposed changes and
the ways in which the government can now reduce public expend-
iture and withdraw entitlements to healthcare and restrict access
to public services without fear of legal challenge. However, the
implications of the repeal are still not widely appreciated, and for
political reasons the government claims still to be committed to
universality.
Apart from duties and responsibilities set out in national legis-
lation, high-​income countries are also subject to provisions of
international conventions on human rights to which they are signa-
tories. These, though not always enforceable in domestic courts, are
relevant to normative debate. Among the most important human
rights instruments are the Convention on the Rights of the Child,
the Convention on the Elimination of All Forms of Discrimination
Against Women, and the Covenant on Economic, Social, and
Cultural Rights. All, according to the WHO, provide ‘a legal and nor-
mative framework for the respect, protection and fulfilment of the
right to health and other related rights of women, children and ado-
lescents’. It is generally accepted that the instruments place national
governments under a duty (although not necessarily enforceable) to
‘ensure that health facilities, goods, and services are of good quality,
are available in sufficient quantity, and are physically accessible and
affordable on the basis of non-​discrimination’.
These various laws and conventions constitute a range of legal
or quasi-​legal provisions that can guide decisions about future
spending.
The perspective of financing
Financial analysis derives from both public health and legal ana-
lyses, for if there is an obligation to fund universal healthcare,
how should money be raised and distributed so that the duty can
be discharged most successfully and at least cost? In general, this
question bears on the extent to which there is equity in resource
distribution across society, and particularly on the role of publicly
financed health spending in redistributing resources from the rich
to the poor and from the well to the sick. It resolves itself into three
testable questions. The first is the extent to which low incomes are
barriers to access, that is, the extent to which patients are required
to pay at the point of use (so-​called out-​of-​pocket payments or user
2.15  How much should rich countries’ governments spend on healthcare?
165
charges) for the services that they need instead of paying in advance
through taxes or insurance. The second is the extent to which ad-
vanced payments (instead of user charges) are equitably calibrated
and collected, either through insurance premiums or taxes (i.e. are
they in line with earnings or income?). The third question concerns
the extent to which healthcare funds are allocated according to
healthcare needs.
Solutions to the first two questions depend largely on the way in
which governments raise money to pay for healthcare, that is, on
which type of financing maximizes universality. Evidence shows
that universality is more likely to be achieved when out-​of-​pocket
payments are minimized, and personal incomes cease to be a bar-
rier to access. In many countries, user charges, or private spending,
entail catastrophic costs for the household and the individual be-
cause of the way they redirect resources from other necessities of life
and basic household expenses (absorbing more than 20% of mean
household income). In the United Kingdom, most people who re-
quire medicines are exempt prescription charges, while in Scotland
and Wales prescription charges have been abolished. However, in
most countries the cost of medicines is borne by individuals. This
is often very expensive for the patient and household expenditure.
For example, in poorer countries a single course of medicines for
common conditions such as malaria or diabetes may cost several
days’ wages. Medical costs including medicine costs are a common
cause of personal bankruptcy in all countries, and in the United
States they account for three-​fifths of all bankruptcies.
Since poverty and income inequality are the main causes of ill
health it would be paradoxical to raise health funds by means that
deepened or failed to address either. Hence, if money is raised in
advance rather than at point of use, it needs to be raised by methods
that impinge the least (or at least proportionately) on the incomes
of the poor. The most equitable funding mechanism is progressive
direct tax. The main alternatives, social or private insurance, both
involve user charges, sometimes substantial, at point of use and have
more limited capacity to address inequality by spreading costs across
the whole of the population (a practice known as ‘risk pooling’). In
a public healthcare system, the optimal form of financing is progres-
sive taxation with a maximum separation between the user and the
caregiver to reduce any conflicts of interest.
How should money be distributed around the system?
Selection versus inclusion
Raising money equitably is only half the battle; resources must be al-
located according to need if universality is to be achieved. Clearly, if
healthcare money is not allocated according to need, or is siphoned
off for non-​healthcare purposes, healthcare budgets are likely to be
underestimated for some parts of the population and overestimated
for others.
Whether markets or market-​like bodies are capable of needs-​
based allocation is a testable question. One key measure concerns
the extent to which the institutions within a health system that are
responsible for securing healthcare and the institutions that provide
it are inclusive, that is, are responsible for everyone in a geographical
area and are not able to exclude more expensive population groups
such as older people and the chronically sick. Were selection of this
type to occur the universal goal would be more expensive to achieve
or might not be achieved at all. Systems based on private insurance
or privately run provision are inherently selective. In these systems
corporations manage their financial risks by excluding high cost pa-
tients or by limiting the range and nature of services available.
Risk selection and out-​of-​pocket costs: Integrated universal
systems and market models
Universal public systems must be designed so that they cannot ex-
clude patients and citizens from care and are built on redistribution
and solidarity and universal risk pooling. In contrast, commercial
health insurers and providers select risks (cherry picking). They
seek to exclude unprofitable patients such as the chronically ill and
elderly, either directly by refusing coverage, or indirectly by char-
ging very high premiums, often restrict coverage for pre-​existing
medical conditions, and frequently leave patients with unaffordable
out-​of-​pocket costs. These insurance practices explain why, in the
United States, mainly unprofitable patients’ groups are covered by
public programmes namely—​the poor by Medicaid and the elders
by Medicare.
Solidarity mechanisms which are integral to the design of public
integrated systems cannot operate in commercial health insurance
or among private providers because their objective is profitability
and solidarity among the wealthy. Private insurers and private pro-
viders tailor coverage to purchasing power, not to need, often leaving
patients with the most expensive conditions uncovered. Premiums
generally rise with age, and high copayments often restrict the use of
medical services. Bureaucratic administrative procedures are neces-
sary to select out patient who require high-​cost care and those who
are at high risk. This can lead to patient and system delays. These
problems have been observed not only in the United States, but also
in middle-​income countries that have adopted market-​oriented
health policies.
A second key measure of the wisdom of utilizing market bodies in
the pursuit of a population’s health provision is the extent to which
commercial pressures to consume more may inflate total (public plus
private) healthcare costs so that more is spent than required from a
public health perspective. This brings us to economic analysis.
The perspective of economics and trade
All governments want their nation’s economies to grow faster, and
health spending is often conceived as a factor in economic growth.
But both increases and reductions in spending can be supported
by different economic analyses. In 2014, for example, a measurable
drop of 1.2% in the United States’ economic growth forecasts was at-
tributed to steps taken by the Obama government to increase access
to healthcare and reduce costs and waste, reductions that were hotly
contested by the healthcare industry.
Similarly, attempts by the UK government in 1957 to control
pharmaceutical profits were successfully opposed by the com-
panies affected. The industry body, the Association of the British
Pharmaceutical Industry, argued—​and the government ultimately
accepted—​that generous profits from proprietary medicines sold
to the NHS subsidized standard drugs, funded export marketing,
and supported research and development. That principle survived
until 2008.
Health spending can thus be conceived, economically, as a bene-
ficial component of growth or as aid to industry. Alternatively, as
we show earlier, under the austerity project it can be seen as an eco-
nomic burden, at least so far as public spending is concerned, but as
we show next, public and private spending are linked.
166
section 2  Background to medicine
The cost of markets
Economists frequently consider the determinants of spending
levels in the form of the structural arrangements that lie behind
prices and how some structures lead to higher prices than others.
The evidence shows that publicly run systems based on geographic
areas of administration are cheaper to run. By contrast, markets
are relatively expensive. For example, before market elements were
introduced, the United Kingdom’s NHS had administration costs
of around 6% of total expenditure. These costs now exceed 15% of
total expenditure, while in the United States, since 1974 one of the
most heavily marketized health systems in the developed world,
they exceed 30%.
Critiques of the role that markets play in inflating costs have been
a staple of American economists for some years. In 2012 the USA’s
Institute of Medicine published a report in which it reproduced find-
ings showing that in the United States, wasted expenditure (approxi-
mately $765 billion) made up a large part its uniquely high level of
spending on healthcare:
‘a substantial proportion of health care expenditures is wasted, leading
to little improvement in health or in the quality of care. [Table 2.15.2]
contains estimates of excess costs in six domains: unnecessary services,
services inefficiently delivered, prices that are too high, excess admin-
istrative costs, missed prevention opportunities, and medical fraud.’
(Institute of Medicine, 2012)
From a public health perspective these excess costs are inefficient,
for not only do they represent a huge loss of productive health cap-
acity, but to the extent that the wastage is publicly funded they also
waste scarce tax resources. In the United States, high levels of pri-
vate spending on healthcare have generated public spending levels
broadly comparable to those of European countries with universal
health systems. However, the United States does not achieve uni-
versality, or anything like it, because much of the public spending
is devoted to the sort of market-​driven costs and wastage listed in
Table 2.15.2 and because commercial considerations dominate in
resource allocation.
The United States has the most expensive system in the world, with
around 60 million people uninsured and almost 120 million people
underinsured at any one time. Overtreatment and overdiagnosis go
hand in hand with the denial of care. Catastrophic health expend-
iture due to out-​of-​pocket health charges is very high.
Case study of cost of markets: Medicines
One of the biggest drivers of catastrophic costs and healthcare costs
in most countries is pharmaceuticals. On average 25% of a country’s
total health expenditure is spent on medicines, with a wide range
from 7.7% to 67.6%, but the source of spending varies enormously.
In many low-​income countries medicines can absorb as much as
50% of the public healthcare budget annually, compared with 12%
in the UK NHS. However, most expenditure on medicines is due
to out-​of-​pocket costs and charges, with pharmaceutical costs ac-
counting for 64% of all out-​of-​pocket payments on healthcare in
some countries.
Since 1995, the private share of total pharmaceutical expenditure
has increased across the world. According to the WHO, across all
income groups, private spending by households in 2000 was the
principal source of pharmaceutical expenditure, at 57.8% in high in-
come, 70.9% in middle-​income, and 71.6% in low-​income countries.
Most of the money that the pharmaceutical industry makes is not
spent on producing medicines. It is estimated that 60% of all pharma-
ceutical revenues are diverted from healthcare to profits, administra-
tion, and marketing costs. According to one analysis of the Forbes
500 drug companies in 2003, of all revenues, 30.8% was spent on
marketing and administration, 17% on profits and 14% on research
and development (R&D). In the United Kingdom, the Association of
the British Pharmaceutical Industry estimated that R&D was 33% of
sales compared with 3.7% in the manufacturing industry, and com-
pany profits as a percentage of sales ranged from 20 to 43% in 2014.
Economists will argue that ‘waste’ such as the diversion of funds to
market bureaucracy, including marketing and billing and invoicing,
is nevertheless beneficial in a broader sense because, being eco-
nomic activity, it contributes to employment and economic growth.
Table 2.15.2  Estimated sources of excess costs in healthcare in the United States, 2009
Category
Sources
Estimate of excess costs
Unnecessary services
• Overuse—​beyond evidence-​established levels
• Discretionary use beyond benchmarks
• Unnecessary choice of higher-​cost services
$210 billion
Inefficiently delivered services
• Mistakes—​errors, preventable complications
• Care fragmentation
• Unnecessary use of higher-​cost providers
• Operational inefficiencies at care delivery sites
$130 billion
Excess administrative costs
• Insurance paperwork costs beyond benchmarks
• Insurers’ administrative inefficiencies
• Inefficiencies due to care documentation requirements
$190 billion
Prices that are too high
• Service prices beyond competitive benchmarks
• Produce prices beyond competitive benchmarks
$105 billion
Missed prevention opportunities
• Primary prevention
• Secondary prevention
• Tertiary prevention
$55 billion
Fraud
• All sources—​payers, clinicians, patients
$75 billion
Adapted from Institute of Medicine (2010). The Healthcare Imperative: Lowering Costs and Improving Outcomes: Workshop Series Summary.
Washington, DC: The National Academies Press.
2.15  How much should rich countries’ governments spend on healthcare?
167
The cost of free trade agreements
Free trade is another economic perspective which has a bearing
on spending. Economists frequently point to the importance
of international trade for economic growth, and to the role of
healthcare, health insurance, and related goods, principally
pharmaceuticals, in generating export-​led growth. However, the
promotion of free trade (the international equivalent of free mar-
kets) can have major implications for spending levels when it is
associated with extending healthcare markets and diluting gov-
ernment controls.
For example, free trade treaties can prevent governments applying
price controls, the logical approach to containing inflationary pres-
sures on public healthcare budgets. Today, more rigorous pharma-
ceutical price controls continue to be opposed by the industry and
are under assault by governments of pharmaceutical producing
countries through trade treaty negotiations, such as those involving
the Trans-​Pacific Partnership (TPP) and the Transatlantic Trade
and Investment Partnership (TTIP). Both these draft treaties could
outlaw or restrict the use of pharmaceutical price controls in cen-
trally administered health systems. Such measures threaten the use
of global budgets where governments set annual limits to health
spending. Rates of expenditure growth are lowest in countries that
cap health budgets under central oversight rules, such as England,
Italy, Japan, and Sweden, and do not allow the market, consumer de-
mand, and provider price-​setting to determine global expenditure,
such as in the United States.
The TPP and TTIP are only the most recent of a series of trade
treaties with implications for health spending. In the 1990s, a
World Trade Organization (WTO) treaty known as the General
Agreement on Tariffs and Trade was exploited by the US insur-
ance industry and the US government as a means of opening up
new markets for the then struggling, multibillion-​dollar busi-
ness of health-​maintenance organizations. Profits had fallen after
1997 because of market saturation, government, and employer
strategies to contain healthcare costs and high-​profile scandals.
To restore profitability, the industry sought to capture new mar-
kets abroad by acquiring publicly run facilities, converting public
funds into profits, and repatriating the profits to the United States.
In this, it received influential backing for its foreign acquisitions
policy in Latin America from the US government, the World Bank,
and multilateral financial institutions such as the Inter-​American
Development Bank. The universal health systems of Europe are the
latest target of healthcare corporations with international trade
strategies.
As can be seen, the aforementioned analyses lead to contradictory
findings for health spending using a market model. The branch of
economics that deals with supply and demand, microeconomics, is
not relevant to our question if universal healthcare is adopted as a
goal. This is because microeconomics is built on a principle of effi-
ciency in which maximizing individual utility—​not access—​is the
desideratum, and in which healthcare is defined as a private com-
modity not an essential public service. The United States offers prac-
tical lessons into the effects of the market model which relies on
commercial insurers and providers. The inhumanity of that system
is a warning to other rich nations and yet it is rapidly becoming the
alternative model to the integrated public health system that the
United Kingdom once had.
Conclusion
We conclude that the decision to have a universal public healthcare
system is political. There is strong evidence that universal healthcare
is sustainable. In contrast, the United States shows that the more
marketized and privatized the systems, the higher the costs, the
waste, and the greater the risks of being denied care. The question
of how much any country should spend is inextricably linked to the
chosen model of funding and provision, the degree of marketiza-
tion, and how much risk selection and denial of care a government is
prepared to tolerate in its health system.
FURTHER READING
Anderson GF, et al. (2003). It’s the prices, stupid: why the United States
is so different from other countries. Health Affairs, 22(3), 89–​105.
Cameron A, et  al. (2011). The World Medicines Situation 2011—​
Medicines Prices, Availability and Affordability. World Health
Organization, Geneva. http://​www.who.int/​medicines/​areas/​policy/​
world_​medicines_​situation/​WMS_​ch6_​wPricing_​v6.pdf
Clements B, Coady D, Gupta S (eds) (2012). The economics of public
health care reform in advanced and emerging economies. International
Monetary Fund, Washington DC.
Institute of Medicine (2012). Best care at lower cost: the path to continu-
ously learning health care in America. The National Academies Press,
Washington DC.
Krugman P (2012). End this depression now. Norton, New York.
Marmot M (2016). The health gap: the challenge of an unequal world.
Bloomsbury, London.
Mossialos EP, Govin BR, Hervey TK (2010). Health systems govern-
ance in Europe: the role of EU law and policy. In: Mossialos Elias P,
Govin BR, Hervey TK (eds), Health systems governance in Europe,
pp. 1–​83. Cambridge University Press, Cambridge.
Piketty T (2014). Capital in the twenty-​first century. Harvard University
Press, Harvard.
Pollock A (2005). NHS plc: the privatisation of our health care. Verso,
London.
Price, D, Pollock AM, Shaoul J (1999). How the World Trade
Organisation is shaping domestic policies in health care. Lancet,
354(9193), 1889–​92.
Rosenbaum SJ (2009). Insurance discrimination on the basis of health
status: an overview of discrimination practices, federal law, and fed-
eral reform options. J Law Med Ethics, 37(2), 101–​20.
Starfield B (2008). The future of primary care: refocusing the system.
N Engl J Med, 359 (20), 2087–​91.
UNCTAD (2014). Trade and development report. United Nations,
New York.
Webster C (1988). The health services since the war. Vol 1: Problems of
health care: the National Health Service before 1957. HMSO, London.
Wilkinson R, Pickett K (2010). The spirit level: why more equal societies
almost always do better. Penguin, Harmondsworth.
World Health Organization (1978). Declaration of Alma-​Ata, International
Conference on Primary Health Care, Alma-​Ata, USSR, 6–​12 September
1978. http://​www.who.int/​publications/​almaata_​declaration_​en.pdf
World Health Organization and World Bank (2015). Tracking uni-
versal health coverage. World Health Organization, Geneva.
World Health Organisation (2018). World Health Statistics 2018: moni-
toring health for the SDGs. https://www.who.int/gho/publications/
world_health_statistics/2018/en/

2.16 Financing healthcare in low- income developin
2.16 Financing healthcare in low- income developing countries: A challenge for equity in health 168
ESSENTIALS
Low-​income developing countries remain challenged by the
increasing demand for essential healthcare combined with gov-
ernments’ inability to mobilize and allocate adequate financing to
health systems. According to their country specific contexts, they
need to improve the equity and efficiency of their health systems, in
particular their public sector management, and increase their gov-
ernmental health spending through enhanced domestic resource
mobilization and more efficiency in public spending. They need to
sustain external assistance and implement social protection policies
that decrease out-​of-​pocket payments, thereby preventing people
with catastrophic illness from becoming impoverished.
Economic growth and increased health spending alone are not
enough to scale-​up healthcare coverage and achieve better health
outcomes. These must be combined with accountability of results,
transparent management of public funds, and multisectoral efforts
with community involvement at implementation level. We should
appreciate the progress made in all low-​income developing coun-
tries in reducing both under-​five mortality rate (U5MR) and ma-
ternal mortality rate (MMR), even if most did not achieve the relevant
Millennium Development Goals (MDGs). The successful cases of
Bangladesh, Bhutan, Mongolia, and Nepal, who did achieve MDGs 4
(U5MR) and 5 (MMR) over the last 10 years, illustrate good practices
that deserve more careful study.
Public spending on health in low-​income developing countries
has not had a strong effect on reducing U5MR and MMR. Much
more evident is the effect of female education. In designing public
policy to reduce excess mortality in low-​income developing coun-
tries, beyond increasing gross national income per capita, education
of women seems to be a crucial issue.
There is no magic blueprint for health financing: each country
should design its own strategy based on evidence provided by
national health accounts. The central issues to be addressed are:
Where does the money come from? Where does the money go
to? What kind of services are performed? And what types of goods
are purchased? These accounts should be monitored and updated
annually on the basis of epidemiological and other relevant evi-
dence, and the outcome of this process should generate the support
for updates of national health financing strategy.
Introduction
People have aspirations to fulfil their health needs. Health systems
of some nature have existed as long as communities have tried to
treat diseases and protect their health, but healthcare has a cost and
people cannot access it without creating the ability to pay for it. This
chapter provides an overview of the essential health financing issues.
It includes policies, concepts, approaches, practices, and experiences
that could enhance the understanding of comprehensive health sys-
tems reforms that would both provide sufficient financial resources
for equitable access to quality health interventions and protect
people from catastrophic health expenditures. Health financing is
analysed from the perspective of governments, but the roles of other
important stakeholders such as the wider society and private sector
are also taken into account to assist in the design, implementation,
and evaluation of effective health financing reforms.
The focus of this chapter is on the group of low-​income developing
countries1 considered according to 2014 International Monetary
Fund classification as ‘countries that were designated Poverty
Reduction and Growth Trust (PRGT) eligible and had a level of per
2.16
Financing healthcare in low-​income
developing countries: A challenge
for equity in health
Luis G. Sambo, Jorge Simões, and Maria do Rosario O. Martins
1  Afghanistan, Bangladesh, Benin, Bhutan, Bolivia, Burkina Faso, Burundi,
Cambodia, Cameroon, Central Africa Republic, Chad, Comoros, Côte d’Ivoire,
Congo, Congo Democratic Republic, Djibouti, Eritrea, Ethiopia, Ghana, Guinea,
Guinea-​Bissau, Gambia, Haiti, Honduras, Kenya, Kiribati, Kyrgyzstan, Lao
People’s Democratic Republic, Lesotho, Liberia, Madagascar, Malawi, Mali,
Mauritania, Moldova, Mozambique, Nicaragua, Nigeria, Myanmar, Mongolia,
Nepal, Niger, Papua New Guinea, Rwanda, São Tomé and Príncipe, Senegal, Sierra
Leone, Sudan, Solomon Islands, Somalia, South Sudan, Togo, United republic of
Tanzania, Tajikistan, Uganda, Uzbekistan Yemen, Vietnam, Zambia, Zimbabwe.
2.16  Financing healthcare in low-income developing countries
169
capita GNI less than (US) $2390’ (IMF, 2014). Qualitative and quan-
titative analysis is made for both controllable and uncontrolled vari-
ables, beginning with the health situation in low-​income developing
countries when the Millennium Declaration was adopted in the year
2000 by the United Nations General Assembly. Particular attention
is paid to the way incidence and prevalence of diseases has evolved
since then, and the way this has impacted on mortality rate of chil-
dren under five years old (U5MR), maternal mortality ratio (MMR),
and life expectancy at birth. The cases of the low-​income developing
countries recording the best progress in reducing the U5MR and
MMR—​Bangladesh, Bhutan, Mongolia, and Nepal—​are analysed
in terms of the relationship between their health financing patterns
and the attainment of Millennium Development Goal (MDG) 4 (re-
duction of child mortality) and Goal 5 (improved maternal health),
with particular emphasis on cases in which health targets have been
reached with limited financial resources.
After discussion of health systems, the chapter then analyses health
spending, the financing pattern, and the relationship between the
latter and health outcomes in terms of achievement of the MDGs.
We caution that there are variances that decision-​makers can con-
trol, such as funding, financial access, social protection, quality of
healthcare, and efficiency, and aspects that cannot be controlled, such
as demographic trends, new norms, and standards resulting from sci-
entific breakthroughs, epidemics/​pandemics, and other public health
threats, which can all affect health outcomes. In light of current sci-
entific knowledge and best practices on health financing, the chapter
generates possible outcomes, depending on choices made, that are
relevant to the context of low-​income developing countries.
Subsequently, the chapter discusses the levels and sources of
health financing in low-​income developing countries and the way
these could have influenced the scaling-​up of cost-​effective public
health interventions towards health MDGs. It makes an analysis of
health financing patterns and the way health economic indicators
evolved from 1995 to 2013, comparing with what happened in high-​
income countries during the same period.
Our last analysis refers to the health production function, which
is calculated on the basis of a multivariate regression model using
a mathematical equation that relates health outcomes to enabling
health system functions. We are not prescriptive and do not generate
a blueprint model or template for health financing in low-​income
developing countries. Available evidence shows that low-​income
developing countries that made some strides in improving key health
status indicators applied different models that combined various
funding sources such as taxation, social insurance, and direct pay-
ments (among other variables) in different macroeconomic envir-
onments. We pursue scientific evidence for trustful health financing
approaches in the context of low-​income developing countries. We
provide ideas on macroeconomic policy and fiscal space measures
that could create budget room, and thereby allow governments to
provide additional resources to the health sector.
Epidemiology of diseases in low-​income
developing countries by the year 2000
The overall level of development influences the quality of life and
health of a population. Political, economic, social, cultural, and en-
vironmental determinants play a role in both the access to care and
the health status. We realize that health improvements in the devel-
oped world over the last century were mainly due to enhancement
of living and working conditions, notwithstanding that scientific
breakthroughs and technology innovation played an important role.
In most low-​income developing countries health profiles are still
characterized by high birth rates, high mortality rates, low life ex-
pectancy, and high disease burden, with particularly adverse impact
on vulnerable populations such as children and pregnant women.
Poverty and hunger, which span over all low-​ and lower-​middle-​
income countries, are central determinants of heath and were
considered by the United Nations as the number one Millennium
Development Goal (MDG1—​eradicate extreme poverty and
hunger). Poverty is the root of malnutrition, drives the spread of
disease, increases vulnerability to outbreaks, epidemics, and pan-
demics, and deters indigents from seeking healthcare. Iniquities in
access to health determinants generate uneven distribution of health
outcomes between continents, countries, and within countries (be-
tween people according to their socioeconomic condition, age, sex,
and geographic area). Human rights movements aim to influence
governments to provide the political, economic, and social environ-
ments that enable the fulfilment of health as a right of each citizen.
The incidence and prevalence of diseases
In 2000 the biggest seven killers of the poor in Africa and Southeast
Asia were HIV/​AIDS, diarrheal diseases, acute respiratory infec-
tions, maternal and perinatal conditions, malaria, measles, and
tuberculosis. Among children and young adults (0–​44 years old),
more than 36 million people lived with HIV/​AIDS worldwide, 95%
in developing countries and especially in sub-​Saharan Africa. The
same year, about 8.8 million people developed active tuberculosis
and 1.7 million died of the disease, with 99% of all tuberculosis (TB)
cases living in developing countries, and malaria killed more than
1 million people (3000 deaths a day), mostly children under 5 years
old and pregnant women in sub-​Saharan Africa.
Maternal death
The main causes of maternal death are severe haemorrhages, infec-
tion, hypertension, prolonged obstetric labour, and complications
of unsafe abortions. In low-​income developing countries in 2000,
the MMR was 628.1 per 100 000 live births compared with 18 per
100 000 in high-​income countries.
The under-​five mortality rate
The U5MR was 132.2 per 1000 live births in low-​income developing
countries, compared with 9.7 in high-​income countries, yet with the
existing knowledge and available technologies it would have been
possible to prevent and treat most of the diseases responsible and
reduce suffering of many millions of people over the world.
Life expectancy at birth
In 2000, life expectancy at birth in low-​income developing countries
was 55 years, compared with 76 in high-​income countries. This was
strongly correlated with very high infant and child mortality and
premature death.
There were, however, some examples of success in the midst of
poverty such as: (1) Cambodia, with the highest HIV/​AIDS preva-
lence in Asia (4%), succeeded in slowing the spread of HIV infec-
tion among sex workers under 20 years old, from 40 to 23% between
170
section 2  Background to medicine
1998 and 2000; (2); the Vietnam malaria programme that reduced
the malaria death toll by 97% between 1992 and 1997; and (3) the
successful tuberculosis programme in Peru that increased the direct
observed treatment short-​course coverage from 70% in 1990 to
100% in 1998, with a cure rate of 90%. Examples such as this pro-
vided evidence of the feasibility of reducing disease burden in low,
lower-​middle, and upper-​middle-​income countries. Best practices
and success stories encouraged world leaders to strive towards a
global response to fight these three killer diseases, giving direct help
to the most vulnerable people, removing significant obstacles that
keep them in poverty, preventing their families from falling into
poverty, and strengthening health services’ capacity and improving
health status in general.
Striving towards health Millennium Development
Goals in low-​income developing countries
All United Nations member states and about 23 international or-
ganizations committed to achieve the following Millennium
Development Goals (MDGs), established following the Millennium
Summit of the United Nations in 2000, by 2015:
•	to eradicate extreme poverty and hunger;
•	to achieve universal primary education;
•	to promote gender equality;
•	to reduce child mortality;
•	to improve maternal health;
•	to combat HIV/​AIDS, tuberculosis, malaria, and other diseases;
•	to ensure environmental sustainability;
•	to develop a global partnership for development.
In 2005, the G8 finance ministers agreed to provide enough funds
to the World Bank, International Monetary Fund, and the African
Development Bank to cancel $40–​55 billion in debt owed by
members of the heavily indebted poor countries to allow them
to redirect resources to programmes for improving health and
education and for alleviation of poverty. Each economically ad-
vanced country was expected to progressively increase its offi-
cial development assistance to reach a minimum net amount of
0.7% of its gross national product at market prices by the middle
of the decade. In 2009, the High-​Level Taskforce on Innovative
International Financing for Health Systems estimated that on
average $44 per capita—​rising to $60 in 2015—​would be needed
to ensure coverage with a set of essential health services in 49 low-​
income countries.
The following indicators were selected to assess the progress of
health status in low-​income developing countries from 1990 up to
2014: to reduce by two-​thirds, between 1990 and 2015, the under-​
five mortality rate (MDG target 4); to reduce by three-​quarters the
maternal mortality ratio (MDG5 target 5); to reduce the burden of
HIV/​AIDS, tuberculosis, malaria, and non​communicable diseases
(MDG target 6); and to increase life expectancy at birth.
The following paragraphs assess the progress made in reaching
these important indicators in the 60 countries under review. For
analysis we computed, for each country and based on the World
Health Organization (WHO) estimates, the way each indicator
evolved from 1990 to 2013.
A critical analysis of MDGs 4 (U5MR) and 5 (MMR) reveals four
categories of countries (Table 2.16.1 and Fig. 2.16.1):
1.	 Those who are clearly outliers: Zimbabwe and Lesotho are inside
the red ellipse in Fig. 2.16.1 because their very high burden of
HIV/​AIDS has negatively impacted on child and maternal health.
Table 2.16.1  Progress towards MDG4 (U5MR) and MDG5 (MMR)
in low-​income developing countries
Country
% reduction in
MDG4
% reduction in
MDG5
Bhutan
–​72.9%
–​86.7%
Malawi
–​72.3%
–​53.6%
Nepal
–​72.1%
–​75.9%
Bangladesh
–​71.4%
–​69.1%
Liberia
–​71.3%
–​46.7%
Mongolia
–​70.5%
–​65.6%
Tanzania, United Republic of
–​69.0%
–​54.9%
Ethiopia
–​68.6%
–​70.0%
Niger
–​68.2%
–​37.0%
Bolivia
–​68.1%
–​60.8%
Cambodia
–​67.7%
–​85.8%
Eritrea
–​66.9%
–​77.6%
Rwanda
–​65.7%
–​77.1%
Madagascar
–​65.2%
–​40.5%
Nicaragua
–​64.8%
–​41.2%
Mozambique
–​63.2%
–​63.1%
Kyrgyzstan
–​63.2%
–​11.8%
Uganda
–​63.0%
–​53.8%
Honduras
–​62.4%
–​58.6%
Senegal
–​60.8%
–​39.6%
South Sudan
–​60.8%
–​59.4%
Yemen, Republic of
–​58.9%
–​41.3%
Guinea
–​57.6%
–​40.9%
Gambia, The
–​56.5%
–​39.4%
Lao People’s Democratic Republic
–​55.9%
–​80.0%
Tajikistan
–​55.9%
–​35.3%
Zambia
–​54.6%
–​51.7%
São Tomé and Príncipe
–​53.8%
–​48.8%
Myanmar
–​53.5%
–​65.5%
Vietnam
–​53.0%
–​65.0%
Benin
–​52.5%
–​43.3%
Moldova
–​52.3%
–​32.0%
Burkina Faso
–​51.7%
–​48.1%
Mali
–​51.7%
–​50.0%
Burundi
–​51.5%
–​43.1%
Haiti
–​49.7%
–​43.3%
Congo, Republic of
–​46.7%
–​38.8%
Afghanistan
–​45.7%
–​66.7%
Nigeria
–​44.9%
–​53.3%
(continued)
2.16  Financing healthcare in low-income developing countries
171
2.	 A second group of countries (Somalia, Central African Republic,
Kenya, and Cameroon) that made very small improvements (in-
side the orange ellipse in Fig. 2.16.1) are countries that faced
frequent humanitarian crises or/​and present high inequalities.
3.	 A third group of 14 countries (inside the blue rectangle in Fig.
2.16.1) with an achievement above the median (Bolivia, Cambodia,
Eritrea, Ethiopia, Honduras, Lao, Malawi, Mozambique, Rwanda,
São Tomé and Príncipe, South Sudan, Tanzania, Uganda, Zambia).
4.	 Four countries with very good progress towards these two
goals: Mongolia, Bangladesh, Nepal, and Bhutan (inside the
green ellipse in Fig. 2.16.1). In Mongolia and Bhutan primary
healthcare services are fully state funded, ensuring free access
to everyone, and social health insurance population coverage
is high. This, however, is not so in Bangladesh and Nepal.
Bangladesh invested significantly in strengthening the health
sector. In Nepal specific programmes provided free delivery
of care and financial incentives to pregnant women to access
a variety of maternity services, including antenatal care, de-
livery in a facility and postnatal care. Programmes targeting
child health have been intentionally community-​based. In this
group of countries, wealth inequalities (as measured by Gini
Index) are low.
Regarding MDG6, the HIV prevalence for adults dropped on
average from 3.6% to 2.5% (from 2000 to 2013), while the average
number of deaths due to tuberculosis (per 100 000 population) de-
creased from 50 to 30. For the same period, the reported number of
Country
% reduction in
MDG4
% reduction in
MDG5
Guinea-​Bissau
–​44.9%
–​39.8%
Togo
–​42.1%
–​31.8%
Djibouti
–​41.3%
–​42.5%
Uzbekistan
–​40.5%
–​45.5%
Sudan
–​40.2%
–​50.0%
Sierra Leone
–​40.0%
–​52.2%
Kiribati
–​39.0%
–​48.0%
Ghana
–​38.8%
–​50.0%
Comoros
–​37.9%
–​44.4%
Côte d’Ivoire
–​34.0%
–​2.7%
Congo, Democratic Republic of
–​32.7%
–​27.0%
Chad
–​31.3%
–​42.4%
Papua New Guinea
–​31.1%
–​53.2%
Cameroon
–​30.6%
–​18.1%
Kenya
–​28.4%
–​18.4%
Mauritania
–​23.5%
–​49.2%
Solomon Islands
–​22.2%
–​59.4%
Central African Republic
–​21.3%
–​26.7%
Somalia
–​19.0%
–​34.6%
Lesotho
13.6%
–​31.9%
Zimbabwe
18.6%
–​9.6%
–0.8
–0.7
–0.6
–0.5
–0.4
–0.3
–0.2
–0.1
0
0
–0.1
–0.2
–0.3
–0.4
–0.5
–0.6
–0.7
–0.8
–0.9
–1
% Reduction in MDG4
Weak
Good progress
% reduction in MDG5
0.1
0.2
Fig. 2.16.1  Achievement of MDG4 (U5MR) and MDG5 (MMR) (% of reduction between 1990 and 2013).
Source: WHO, Millennium Development Goals Indicators. 2014.
172
section 2  Background to medicine
malaria deaths decreased on average 65%. However, the pattern of
progress was different across countries: for example, in Malawi and
Zimbabwe, both with very high HIV prevalence, considerable pro-
gress has been made in the last 15 years, but this has not happened
in other countries, such as Mozambique. TB rates remain high in
countries like Nigeria, Somalia, and Cameroon.
Mean life expectancy at birth in low-​income developing countries
has improved from 55.08 years (in 2000) to 61.57 (in 2012), com-
pared with 76 (in 2000) to 79 (in 2012) in high-​income countries.
Age-​standardized mortality rate due to communicable diseases,
non​communicable diseases and injuries has dropped 35%, 7%, and
20%, respectively, from 2000 to 2012.
In summary, the progress towards health MDGs in low-​income
developing countries is heterogeneous, with some champions, a
large group of good performers, a group of countries with slow im-
provements, and two countries with no progress. In 2018 the WHO
published a progress report on the health-related MDGs, which re-
corded impressive progress in some countries. Bangladesh, Bhutan,
Mongolia and Nepal did achieve MDGs 4 (U5MR) and 5 (MMR)
over the last ten years (Tables 2.16.1 and 2.16.2), but most low-​
income developing countries did not and will need to address them
in their national health development policies and related health
system designs.
Apart from HIV/​AIDS, tuberculosis, and malaria with serious
impact on MDGs 4 (U5MR) and 5 (MMR), many other infectious
diseases—​neglected tropical diseases and epidemic prone diseases—​
mostly hit poor people. These diseases place a substantial economic
burden on health systems and contribute to lifelong disadvantage of
already deprived people. For example, malaria hits the subtropical
areas around the world—​the poverty belt—​and in high endemicity
areas of sub-​Saharan Africa it accounts for 40% of public health ex-
penditure, 30–​50% of inpatients admissions, and up to 50% of out-
patient visits.
There is currently a shift in global focus from communicable dis-
eases to non​communicable diseases as the major cause of morbidity
and mortality in low-​income developing countries, but most of these
countries are facing the burden of both. More research is required
to understand the associations between communicable diseases and
non​communicable diseases.
The 2002 emergence of severe acute respiratory syndrome (SARS)
in Asia; the 2007 bird flu outbreak caused by highly pathogenic H5N1
avian influenza in Asia, Europe, and Africa; the 2009 swine flu pan-
demic produced by H1N1 influenza in South-​America; the 2014–16
and 2018–19 (ongoing) haemorrhagic fever epidemics caused by
Ebola virus in West Africa; and the Zika virus epidemic of 2015–16,
which began in the Americas; have all vibrantly demonstrated the vul-
nerability of communities to emerging and re-​emerging diseases, and
the need for more resilient health systems at local and global levels.
By identifying ways to strengthen health systems and improve
access to essential healthcare, it is possible to reach people in poor
resource settings, even in remote areas, and rescue them from the
trap of disease–​poverty–​disease. Tackling diseases requires invest-
ments in knowledge, technology, manpower, and other required
capacities to strengthen the overall health infrastructure. However,
bearing in mind that interrelationships between health–​disease and
its determinants are complex and dynamic, and often taking place in
turbulent and unforeseen environments, leaders, and health man-
agers must apply a critical approach to designing and developing
health systems. Governments, partners, communities, and other im-
portant stakeholders require reliable and appropriate evidence at the
right time to develop health policies, strategic plans, and operational
budgets that respond to epidemiological patterns in terms of pro-
vision of care, and reflect equity principles in terms of access to it.
The challenge of financing healthcare in 
low-​income developing countries
General approach of health systems financing
A health system can be defined as all organizations, people, and
actions whose primary intent is to promote, restore, or maintain
health. According to WHO a national health system has six core
components: health workforce, service delivery, health technologies
(medical products, vaccines, and technologies), financing, informa-
tion, and leadership and governance.
Health systems play a key role in improving the health status of
populations and contribute to the improvement of living standards
and economic and social development. In health terms, ‘ideally
Table 2.16.2  Likely timescale for achieving MDGs 4 and 5 in low-​income developing countries
MDG 4 –​ Reduce under 5 mortality
MDG 5 –​ Improve maternal health
Achieved by 2019
Bangladesh; Bhutan; Mongolia; Nepal
Bangladesh; Bhutan; Mongolia; Nepal
Achieve 2020–​2029
Benin; Bolivia; Cambodia; Comoros; Eritrea; Ethiopia; Ghana; Haiti;
Honduras; Kiribati; Lao People’s Democratic Republic; Liberia;
Madagascar; Malawi; Myanmar; Nicaragua; Niger; Rwanda; São
Tomé and Príncipe; Senegal; Sierra Leone; Somalia; Tajikistan;
Tanzania; Uganda; Yemen; Vietnam
Central African Republic; Kiribati; Rwanda; Tajikistan; Vietnam
Achieve 2030–​2039
Burundi; Djibouti; Gambia; Guinea; Mali; Mozambique
Myanmar; Sudan; Uganda
Achieve after 2040
Afghanistan; Burkina Faso; Cameroon; Central African Republic;
Chad; Congo (DR); Congo; Côte d’Ivoire; Guinea-​Bissau; Kenya;
Kyrgyzstan; Lesotho; Mauritania; Nigeria; Papua New Guinea;
Solomon Islands; Sudan; Togo; Uzbekistan; Zambia; Zimbabwe
Afghanistan; Benin; Bolivia; Burkina Faso; Burundi; Cambodia;
Cameroon; Chad; Comoros; Congo (DR); Congo; Côte d’Ivoire;
Djibouti; Eritrea; Ethiopia; Gambia; Ghana; Guinea; Guinea-​Bissau;
Haiti; Honduras; Kenya; Kyrgyzstan; Lao People’s Democratic
Republic; Lesotho; Liberia; Madagascar; Malawi; Mali; Mauritania;
Mozambique; Nicaragua; Niger; Nigeria; Papua New Guinea;
São Tomé and Príncipe; Senegal; Sierra Leone; Solomon Islands;
Somalia; Tanzania; Togo; Uzbekistan; Yemen; Zambia; Zimbabwe
2.16  Financing healthcare in low-income developing countries
173
everyone should have a fair opportunity to attain their full health
potential and, more pragmatically, none should be disadvantaged
from achieving this potential if it can be avoided’ (Whitehead, 1990).
In healthcare the principle of equity should lead to universal access
of everyone to available care for equal need, equal utilization, and
equal quality. Around the world, health systems face serious chal-
lenges in terms of their financial sustainability and providing equity
in health and universal access to quality healthcare. Although public
funding—​through taxes or social insurance—​is dominant, out-​of-​
pocket payments from citizens represent an important share of total
spending.
In general, the financial flows in a health system run between
three types of agents:  population, financing organization(s), and
healthcare providers, as described in Fig. 2.16.2.
The payment for healthcare provided to the population can
come from two major sources: payments made by the population
at the time of use and/​or payments made by third-​party payers.
The latter source of payment is based on previous fund-​raising
among the population, which finances the health system. Within
this general description fall the various existing systems, such as
financing by taxes, social insurance, or private insurance premiums.
Consequently, the financing organization may be the state, sickness
funds, or insurance companies. Adjusting the financial model to the
socioeconomic context is the central problem for health financing in
any country in the world.
The diversity and complexity of outcomes and determinants of
health leads to a difficult question: how can we evaluate the perform-
ance of a health system? In this chapter we consider three out of the
five indicators proposed in the World Health Report of 2000: the
overall health status, the distribution of health in the population, and
the distribution of the financial contribution. With the epidemio-
logical transition from infectious diseases to non​communicable
­diseases, the measure of non​fatal consequences of diseases, espe-
cially chronic diseases and external causes, proves to be relevant in
both low-​ and high-​income countries. The distribution of health and
health determinants, in turn, allows for the assessment of equity in
the distribution of health outcomes in terms of MMR and U5MR, life
expectancy, and disease burden in general. The application of these
indicators to WHO member countries and the resulting classification
of countries has given rise to an intense debate, either directed to the
results or to methodological and scientific aspects of the assessment.
The context in low-​income developing countries
In low-​income developing countries, the first aspect to consider is
that the state plays a central role, not only in funding but also in
provision of services, through a model of national health service
supported by social insurance, usually associated to formal em-
ployers. A second issue is the amount of out-​of-​pocket payments by
families to support healthcare. In the African region, where about
60% of low-​income developing countries are situated, the general
government expenditure on health constituted 48.7% of total ex-
penditure on health, and by the end of 2011 only 6 of the 45 coun-
tries reporting on this factor had met the Abuja target of allocating
at least 15% of total government expenditure to the health sector.
Moreover, out-​of-​pocket payments form over 50% of total health
spending, which exposes families to the risk of catastrophic health
expenditure. Advance payment schemes are a possible way to over-
come this problem. These can be arranged through general taxation
and/​or compulsory contributions to health insurance. A third aspect
to consider is the need to promote efficiency in national health
services to optimize resource allocation and maximize results. The
2010 WHO report on health systems financing estimated that 20–​
40% of resources dedicated to health are wasted.
In recent years there has been an increase in the number of
African countries seeking universal health coverage based on man-
datory prepayment schemes. The literature identifies a wide range
of factors that would allow these countries to progress towards uni-
versal health coverage: revenue growth, which increases the finan-
cial capacity of companies and citizens; improving tax collection
systems, which increases the volume of contributions; the devel-
opment of communication infrastructures, improving geographic
access to healthcare; increase in the health system workforce; pro-
moting social solidarity, facilitating financial transfers from the rich
to the poor and from the healthy to the sick; and strong leadership by
governments in respect to all these issues. However, it is unlikely that
universal health coverage can be achieved solely through tax funding
in the present circumstances. The ability for low-​income countries to
generate more tax revenue is limited by structural factors: per capita
income is relatively low and gross domestic product (GDP) growth
is modest; there are large informal economy sectors (including agri-
culture) that make it difficult to collect taxes; there are low volumes
of taxable imports; and due to institutional factors related to political
instability there is weak institutional capacity, weak accountability,
and inefficient tax administration systems. All these are exacerbated
by the unpredictability and non​alignment of myriad partners with
national health priorities. For example, an important share of health
financing, especially in Africa, comes from international funding,
with a particular increase in funding for specific diseases such as that
provided by the Global Alliance for Vaccines and Immunization and
the Global Fund to Fight AIDS, Tuberculosis and Malaria.
Payments to
providers
Collection of funds
for health care
Out-of-pocket
payments
Financing
organizations
Population
Healthcare providers
Delivery of healthcare
Fig. 2.16.2  Financial flows in health systems.
174
section 2  Background to medicine
While universal health coverage is an ideal that no country can
fully achieve, many countries are now getting closer to universal
coverage and others are making good progress using different ap-
proaches. Interpreting universal health coverage in the context of
each country involves disaggregating the concept in terms of ob-
jectives and emphasizing the progress in improving equity in access
to health services, in improving service quality, and in providing fi-
nancial protection, while considering the access to other key health
determinants.
What do countries need to do to make progress towards health
MDGs? Research has shown, for example, that in countries with
good governance, additional government health spending does re-
duce child mortality. World Bank studies also concluded, however,
that development assistance has a stronger effect in countries with
solid policies and institutions than in countries with only average-​
quality policies and institutions, and an insignificant effect in coun-
tries where policies and institutions are weak.
The crucial question remains how to finance the extra spending
that is required for most countries to attain the MDGs. Health
spending can be divided into three categories:  private (out-​of-​
pocket expenditures and private insurance); public (financing from
general revenues and social insurance contributions); and external
sources (development assistance). There is evidence that where out-​
of-​pocket spending is less than 20% of total health expenditure, the
risk of catastrophic health expenditure is negligible. Two other im-
portant indicators in this context are the share of general revenues
allocated to the health sector (the 2001 Abuja target, only for African
countries, pledged to increase government funding for health to at
least 15%) and the per capita total expenditure on health. The High-​
Level Taskforce on Innovative International Financing for Health
Systems estimated that on average $60 per capita in 2015 would be
needed to ensure coverage with a set of essential health services in
low-​income countries.
Health spending
As illustrated in Table 2.16.3, the median expenditure on health as
a percentage of GDP has grown from 4.5% in 1995 to 5.9% in 2013,
largely attributable to increase in median government expenditure
from 1.9% to 3% of GDP over this period. Government expenditure
on health as a percentage of total government expenditure rose from
a median value of 8.5% in 1995 to 10.5% in 2013. However, this pat-
tern is very heterogeneous, and only 8 of the 57 countries reporting
on this factor had met the Abuja target of 15%. In 2013, median per
capita total expenditure on health was $137.6 compared with $55.2
in 1995 (at constant 2011 prices). It was less than $44 in only four
countries, $44 to $60 in two countries and over $60 in 51 (90%) of
the countries.
Health financing pattern
Concerning health-​spending categories, the median out-​of-​pocket
payments as a share of total health expenditure decreased from
48% in 1995 to 40.5% in 2013. This indicator is still above 20% in
49 countries, meaning that many of the population in most of the
countries (86%) remain exposed to the risk of financial catastrophe
and impoverishment. Median government expenditure on health
constituted 40.7% of total expenditure on health in 1995 and 49.2%
in 2013. The external resources for health made up less than 20%
of total health expenditure in half of the countries, 20–​50% in 21
countries, and over 50% in six countries. These figures contrast with
those obtained by high-​level income countries, as can be seen in
Table 2.16.3.
The relationship between the pattern of health financing
and the attainment of the MDG
Key health spending and health financing indicators for coun-
tries performing best in terms of both MDG4 (U5MR) and MDG5
(MMR) are shown in Figs. 2.16.3–​2.16.6. The pattern of the four
countries is substantially different: whereas Bhutan and Mongolia
achieved good performance with relatively high and considerable
increasing per capita expenditures on health, Bangladesh attained
MDGs achievement with relatively low per capita health expend-
itures. The pattern of Nepal health economic indicators is similar to
the median of the countries (Fig. 2.16.3).
There is also a surprising amount of variation across countries in
the share of GDP allocated to government health programmes and
its trend over time. Countries achieving the two MDGs spend quite
different amounts: Bangladesh with relatively low and unchanged
share over time; Mongolia had relatively high and increasing trend
(Fig. 2.16.4). Bangladesh, Bhutan, and Mongolia decreased gov-
ernmental health expenditures as a percentage of total government
expenditures between 1995 and 2013, while in Nepal this indicator
increased over time, presenting the same pattern as the median of
the 57 countries (Fig. 2.16.5).
The share of the three health-​spending categories composition is
also different across countries. In Bangladesh and Nepal, the share
of out-​of-​pocket expenditure in total health spending is persistently
Table 2.16.3  Health economic indicators for low-​income developing countries (LIDC) and high-​level income countries (HIC)
LIDC
HIC
Indicator/​Year
1995
2013
2013
Per capita total expenditure on health (PPP, Int US$, and constant prices 2011)
55.2
137.6
4516
Total expenditure on health as a percentage of GDP
4.5
5.9
11.6
Government expenditure on health as a percentage of GDP
1.9
3.0
7.3
General government expenditure on health as a percentage of total government expenditure
8.5
10.5
16.8
Out-​of-​pocket expenditure as a percentage of total expenditure on health
47.9
40.5
15
Government expenditure on health as % of total health expenditure
40.7
49.2
60.6
Source: World Bank Statistics.
2.16  Financing healthcare in low-income developing countries
175
higher than in the other two countries (Fig. 2.16.6). This clearly
shows that distinct health financing patterns can lead to the same
result, namely substantial improvement in MDG4 (U5MR) and
MDG5 (MMR).
The health production function
Health economists define the health production function as a math-
ematical equation that relates health outcome to a set of factors,
namely per capita income, share of government health spending
in GDP, education attainment, and access to health services. Using
women years of education as a proxy for educational attainment and
the percentage of urban population as a proxy for access, we have
estimated the impact of each variable on achievement of MDGs,
adjusting for possible confounders. The multivariate regression for
2009 data shows that there are two relevant indicators that affect
U5MR (MDG4): GNI per capita and female education. Adjusting
for other factors, the under-​five mortality rate decreased by 0.327%
when GNI per capita increased by 1%. Relating to female education,
the U5MR was roughly 7% lower per additional year of female edu-
cation. MDG5 (MMR), after adjusting for other factors, decreased
by 0.58% when GNI per capita increased by 1%, and there was a 15%
reduction in maternal mortality rate per additional year of female
education.
Conclusions
Low-​income developing countries remain challenged by the
increasing demand for essential healthcare combined with gov-
ernments’ inability to mobilize and allocate adequate financing to
health systems. In spite of people’s motivation to pay for better access
to quality healthcare, the poor health infrastructure, low salaries
in the health sector, and limited access to essential health techno­
logies all frustrate implementation in the field of equitable health
policies and the aspiration towards universal health coverage.
Economic growth and increased health spending alone are not
enough to scale-​up healthcare coverage and achieve better health
outcomes. These must be combined with accountability of results,
transparent management of public funds, and multisectoral ef-
forts with community involvement at implementation level. We
should appreciate the progress made in all low-​income developing
countries in reducing both U5MR and MMR, even if most did not
achieve the relevant MDGs. The successful cases of Bangladesh,
Bhutan, Mongolia, and Nepal, who did achieve MDGs 4 (U5MR)
and 5 (MMR) over the last ten years, illustrate good practices that
deserve more careful study.
While we appreciate the rise in median government expend-
iture for health in low-​income developing countries from 40.7%
in 1995 to 49.2% in 2013, it is very important to recognize that the
median out-​of-​pocket payments as a share of total health expend-
itures is still above 20%, which means that most people in these
countries are seriously exposed to catastrophic illness and im-
poverishment, and this situation has not significantly improved.
Poverty increases the need for healthcare while dwindling the
capacity to pay for it.
0.00
100.00
200.00
300.00
400.00
500.00
600.00
Bangladesh
Bhutan
Mongolia
Nepal
Median
1995
2013
Fig. 2.16.3  Per capita total expenditure on health in 1995 and 2013
(PPP, Int$, constant prices 2011).
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
Bangladesh
Bhutan
Mongolia
Nepal
Median
1995
2013
Fig. 2.16.4  Government expenditure on health as a percentage of GDP
in 1995 and in 2013.
Source: World Bank statistics.
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
Bangladesh
Bhutan
Mongolia
Nepal
Median
1995
2013
Fig. 2.16.5  General government expenditure on health as a percentage
of total government expenditure in 1995 and in 2013.
Source: World Bank statistics.
176
section 2  Background to medicine
Finally, it is important to highlight that health financing should be
fit for its purpose as a key input to national health systems. It should
be consistent with national health policies and budgeting models that
enable the health system to perform its functions and ensure social
protection. There is no magic formula, no blueprint model to be pro-
scribed. Communities, governments, and international development
partners should keep striving for improved local, national, and global
health, which presupposes improved health financing at all levels.
FURTHER READING
IMF (2014). Poverty reduction and growth trust. International
Monetary Fund, Washington DC.
Sambo LG, Kirigia JM (2014). Investing in health systems for universal
health coverage in Africa. BMC Int Health Hum Rights, 14, 28.
Wagstaff A, et  al. (2006). Millennium development goals for
health: what will it take to accelerate progress?. In: Jamison DT, et al.
(eds) Diseases control priorities in developing countries, 2nd edition,
Chapter 9. World Bank, Washington DC.
Whitehead M (1990). Concepts and principles of equity and health.
Copenhagen, World Health Organization Regional Office for
Europe.
WHO (2000). World health report. World Health Organization,
Geneva.
WHO (2012). Malaria in Africa. World Health Organization,
Geneva.
WHO, et al. (2002). Scaling up the response to infectious diseases. World
Health Organization, Geneva.
WHO (2018). Millenium Development Goals (MDGs). World
Health Organization, Geneva. https://www.who.int/news-room/
fact-sheets/detail/millennium-development-goals-(mdgs)
61.30
30.79
11.60
69.56
47.90
36.24
69.21
82.50
26.48
40.70
2.45
5.90
3.96
11.40
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
60.23
25.41
37.01
46.15
40.50
35.26
73.84
60.25
43.29
49.20
4.51
2.74
10.56
10.30
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
Bangladesh
Bhutan
Mongolia
Nepal
Median
Bangladesh
Bhutan
Mongolia
Nepal
Median
Fig. 2.16.6  Share of out-​of-​pocket (in blue), government (red), and other health expenses (green) (as a % of the total health
expenditure) in 1995 (left) and 2013 (right).
Source: World Bank statistics.

2.17 Research in the developed world 177
2.17 Research in the developed world 177
ESSENTIALS
The entire health system seeks to prevent disease, or diagnose and
intervene effectively to limit its impact. The goal of most funders of
medically related research in the developed world is to support re-
search that is upstream of that which is relevant and attractive to
health providers or pharmaceutical companies, and to work syner-
gistically in the health research ecosystem with these and other com-
plementary organizations when appropriate. It is often impossible to
predict which advances in basic science will lead to breakthroughs,
but if this upstream work is not done, then in the mid and long term,
the flow of medical advances will dry up. The central problem for the
funder of medical research is how to choose the best use for the next
Dollar, Pound, or Euro. This is essentially a problem in prediction, but
as Yogi Berra said: ‘It’s tough to make predictions, especially about
the future.’
Introduction
‘And over them triumphant Death his dart Shook, but delay’d to strike,
though oft invok’d.’ (Milton, J., Paradise Lost)
A recent estimate is that current annual global spending on research
and development is in the region of (US) $1.5 trillion (c.2012), of
which roughly $270 billion (or about 18%) goes to medically re-
lated research. Thus, in the 19 years since satirical online publication
The Onion reported that the death rate remained at 100% despite
huge efforts from health workers and doctors worldwide, roughly
$5 trillion has been spent research work with the objective of health
benefit. Yet the death rate has not been affected. Furthermore, over
the next five years, the Wellcome Trust intends to spend another $7.5
billion with the same goal!
So why do we do it? How do we attempt to do it well? Because,
while Death cannot be denied, life can be improved through better
health and Death itself can at least be delayed and premature Death
prevented. Health economists might state this as an effort to maxi-
mize the global number of Disability Adjusted Life Years. While
the entire health system seeks to prevent, diagnose, and intervene
effectively, the goal of health research is upstream of that, to deepen
our understanding of the biological and social determinants that
underlie health and the processes that undermine it to develop better
interventions for prevention, cures, and palliative interventions.
Time scales
The goal at the Wellcome Trust, alongside most other funders of
medically related research in the developed world, is to support
research that might be upstream of that which is relevant and at-
tractive to health providers (such as the National Health Service
in the United Kingdom) or to pharmaceutical companies, and
also to work synergistically in the health research ecosystem with
these and other complementary organizations when appropriate.
We seek to support the best work throughout the research-​to-​
implementation cycle mostly in academic institutions (universities
and institutes), some of which will be too far from application or
of too great a risk of failure to be justifiable in the budget of a pro-
vider or a for-​profit company. However, we are confident that if this
upstream work is not done, then in the mid and long term the flow
of medical advances will dry up. The Wellcome Trust also offers
extensive funding to encourage public engagement and involve-
ment with science and to research work in the humanities, ethics,
and public policy. We believe that this approach ensures that the
best science remains integrated within the context of the societies
in which it operates and has the greatest opportunity to impact on
human health whether that impact is to be in the short, medium,
or long term.
At the Wellcome Trust, we choose to target our funds to a range
of time scales. Some investments are made in work that we estimate
to be very far upstream, such as our funding of parts of the human
genome-​sequencing project (with the US National Institutes of
Health (NIH) and others) in the 1990s. That work was clearly then
of little immediate utility, but now accelerates much more targeted
work in genetic testing, gene therapy, and the identification of gene
products that may be useful targets for drug development. Genomics
is now becoming widely useful, with larger scale genome projects in
both humans and pathogens, and the beginnings of attempts to link
sequence data to medical records.
2.17
Research in the developed world
(a view from the Wellcome Trust)
Jeremy Farrar
178
section 2  Background to medicine
Some of our funds are targeted even further upstream. A signifi-
cant example is our funding (with the Gatsby Foundation) of the
Sainsbury-​Wellcome Centre for Neural Circuits and Behaviour.
This Institute is focused on the study of basic neuroscience, by the
study of the behaviour and activity of neurons and their circuits in
the brains of living animals (rodents and fish) while they engage in
solving mazes and similar tasks. A dozen years ago techniques were
developed to target genes to neural cells that express proteins which
can reveal the activity of those cells by emitting specific colours of
light: almost literally putting light-​bulbs in the head of an animal as
it thinks about its work. At roughly the same time light-​activated
channels were adapted from bacteria for expression in animals, so
that now specific neurons can be switched on, or off, by the appli-
cation of light. Taken together these new technologies are broadly
known as ‘optogenetics’. And at the same time people have devel-
oped better methods for mapping the synaptic connections of neural
cells using light and electron microscopy, to elucidate the precise
wiring diagrams of neural circuits.
So that now, scientists can map the neural networks, see the ac-
tivity and control specific cells, all while the animal is behaving. In
an analogy to the London Underground: suddenly we can see the
tracks, the trains as they move along them and even through some
switches to see what happens to the paths of the trains. The clock-
work of the mouse or fly mind is being revealed. One hopes that in
a few decades a similar understanding of the human brain could be
developed (a much bigger task).
But now the science is jumping ahead again, with the discovery
and engineering of proteins to control nerve cells not by light, but
by small molecules (designer receptors exclusively activated by de-
signer drugs, DREADDs) or by ultrasound. Work that seemed to be
decades away from the clinic is now possibly only a few years away
and we stand at the brink of major progress in the understanding
and treatment of diseases of the brain. Perhaps conditions like epi-
lepsy will be the first that are amendable to this sort of intervention,
but we are increasingly confident that these new technologies will be
applicable to highly complex conditions, including mental illnesses
and dementia in the medium to long term.
It is often impossible to predict which advances in basic science
will lead to such breakthroughs. For example, study of channel rhod-
opsin in bacteria and fluorescent proteins in jellyfish leading to the
understanding of and then control of the brain, or the work in fun-
damental discovery science of over 50 years ago which had led to
today’s opportunities for treatment with stem cells and monoclonal
antibodies.
Such cases are the justification for funding research all along the
path from the very fundamental to the very applied. But which or-
ganizations should work to support which stages along the path?
Health services and pharmaceutical and medical engineering com-
panies are motivated towards research that is close to clinical utility
or medical products. But there are gaps at this end of the spectrum
for diseases where there is ‘market failure’. Bluntly put; situations for
which there are no financial incentives to develop interventions, dis-
eases which predominantly affect the poor or diseases of poor coun-
tries. Funders in wealthy countries often support work on neglected
tropical diseases, for example, through genuine humanitarianism.
This is sometimes augmented by enlightened self-​interest: most ob-
viously in the case of emerging infections that become epidemic in
poor countries (such as Ebola virus). Developing vaccines or cures
for these illnesses is clearly in the interest of all, as such epidemics
can have a devastating and destabilizing impact on the whole of a
society, country, or region and many outbreaks are only the length
of a jetliner’s flight from the rest of the world.
Another reason why organizations in high-​income countries
might fund research in low-​ and middle-​income ones is simply the
power of studying a disease where it is common. One clear example
of this is HIV disease and tuberculosis in Southern Africa. With the
prevalence of HIV in parts of South Africa at 60%, there are imme-
diate benefits to all for research to be conducted where the need is
greatest.
Governments fund research all along the path from basic science
to applied work, but there is increasing pressure to spend tax-​payers’
money on research that will bring health or economic benefit in the
short term. We can see this in the pressure for demonstrable ‘im-
pact’ in the United Kingdom and in the United States, in the reduced
budget for basic research in Australia, and the level-​funding of the
NIH in the United States, while the National Science Foundation
(NSF) is actually in decline.
Types of funders
Government funding mechanisms in developed countries are di-
verse. The largest single funder in the world is the NIH’s USA
extramural programme. In total the NIH funds over $30 billion a
year, mostly in the United States and mostly through universities,
although a significant fraction (about 17%) is spent at the intra-
mural programmes (mostly at the various Institutes of the National
Institutes of Health in Bethesda Maryland). The US National Science
Foundation covers a broader range of science, but spends only about
a quarter of the sum that the NIH does. The NSF runs no directly
funded institutes. Other US federal agencies also contribute to the
broad biomedical research agenda including the Department of
Energy, Federal Drug Administration, and many others. As a result,
the total government funding available in the United States for bio-
medical research amounts to over $50 billion a year.
The European Union’s European Research Council (ERC) funds
roughly €13 billion per year. In addition, individual European coun-
tries fund science themselves. Germany funds biomedical research
through both the Deutsche Forshungsgemeinschaft (DFG, or German
Research Foundation), which funds mostly through universities,
and the Max Planck Society, which works by funding its own insti-
tutes. In the United Kingdom the Medical Research Council (MRC)
runs a mixed model, with funding both through grants to academic
research leaders in universities as well as its own units and institutes.
The largest MRC sole-​funded institute now is the Laboratory of
Molecular Biology (the ‘LMB’) in Cambridge, although the MRC is
also a founding partner in the Francis Crick Institute, which opened
in 2016 and will host 1200 researchers.
Some non​government funders have been set up as charities
that specifically fund work on specific diseases or areas, such as
the American Cancer Society, the British Heart Foundation, the
Fragile-​X Association, and many others. These organizations gen-
erally raise funds from donors in the community and with specific
disease interests, and many do not accumulate large endowments.
Traditionally it has not been attractive for disease-​motivated donors
to fund fundamental research with a long-​term perspective because
2.17  Research in the developed world
179
they are mostly focused on short term progress in the disease areas
of interest. These pressures often turn these charities away from
basic research. However, this is changing and increasingly some of
these philanthropic organizations including, for instance, Cancer
Research UK and Paul Allen Foundation, are supporting extensive
programmes in discovery science and taking a much longer-​term
perspective.
Other private charities are funded by one (or a very few) living
donors. The largest of these now is the Bill and Melinda Gates
Foundation, which is focused on the interests of the donors:  for
impact on global health during their lifetimes. Another example is
the Gatsby Foundation, in the United Kingdom, which is funded by
Lord Sainsbury of Turville and is focused on research into food sus-
tainability and neuroscience areas of interest to the donor.
Some of these personal foundations are directed to spend out their
funds during the lives of the donors, whereas others are organized on
large endowments left by the original donors or corporations to be-
come perpetual foundations or charities. These organizations invest
the endowment and spend a defined fraction of the value towards
the charitable activities, mostly research. The two largest of these are
the Wellcome Trust, in the United Kingdom, with an endowment of
over £18 billion at the last financial report, and the Howard Hughes
Medical Institute (HHMI), with an endowment of roughly $18 bil-
lion. Neither the Wellcome Trust nor the HHMI focus on specific
diseases, but fund research broadly across the biomedical science
and health area.
What is the ‘unit of funding’
Most non​commercial biomedical research funding in developed
countries goes towards work done by academics in universities. This
is for historical reasons (that is where most of the basic and clinical
scientists are), but also because universities have evolved to provide
a comprehensive academic infrastructure, based on scholarship and
teaching, that includes key resources (buildings, libraries, vivaria,
and more). Most important is the support of the academics them-
selves, through salaries and a career structure. In serious research
intense universities, academic freedom is assured, at least at the level
of the independent group leader (often but not always some variant
of ‘professor’). Thus independent research group leaders are free to
pursue their interests as far as they can find funding to do so. While
they may (or usually these days do not) have ‘tenure’ (permanent
employment), today they very rarely have long-​term institutional re-
search funding and must therefore secure grants from government
or private funders. These academics are overwhelmingly dedicated,
imaginative, creative, hard-​working, and competitive. Because they
must continuously run a permanent marathon cycle of research,
publication, and grant-​winning, the benefits of this dedication
come to us at far too cheap a price (academic salaries do not impress
bankers).
While there are very great advantages to funding into univer-
sities, there are also drawbacks. The academic competitive structure
is based (usually) on the ‘brand identity’ of the individual group
leader/​professor. This is most often scored as senior-​authored
papers. This culture can be a major inhibitor of collaboration
between research groups. As science moves forwards it is often
crucial for fields to meet, for new technologies to be developed
and adopted and for ideas and theories to cross boundaries. The
very reward-​culture system of the academic university thus incen-
tivizes individual achievement and success, but at the same time
dis-​incentivizes interdisciplinary collaboration. While there are
some university structures that are also barriers (departments and
schools), the real barriers are inherent to the system of credit-​by-​
authorship, the competitive cycle of short-​term funding and the
increasing pressure for ‘impact’.
In addition to the independent group leader/​professor, it is also
critical to support the training and career-​progression of more
junior researchers, through doctoral programmes and various forms
of fellowships designed for basic scientists as well as clinicians. In
most disruptive and creative industries, it is these individuals who
have provided most of the breakthrough innovations that have led
to major changes in the way we live and work. We, and our com-
munity are too conservative in the freedom and support we give to
this group. In our current system, at least in the United Kingdom
and the United States, there is a very great excess of the number of
junior people entering training over the number of professorial jobs
that might await them. In the United Kingdom a study by the Royal
Society in 2010 showed that only 0.45% of those entering PhD stu-
dents are likely to end up as professors. While there are many other
useful and productive paths for many people other than research
independence, this ratio is clearly too great and is disheartening to
young people. Many undergraduates sample research through pro-
jects of vacation scholarships and are driven away to other fields
when they see the odds for themselves. We are losing a great many
talented people.
The main alternative to funding research in universities is to sup-
port science in institutes. These range from small ‘centres’ or ‘units’
within universities, sometimes in their own buildings, to larger or
free-​standing ‘institutes’, usually on their own land, such as the NIH
campus, HHMI’s Janelia Farm, or the Wellcome Trust’s Genome
Campus at Hinxton. In such places it is possible to rewrite some or
all of the normal ‘rules’ of academic research in an attempt to over-
come some of the barriers. Commonly there is no tenure. Research
focus can be required. Often, resources are controlled centrally by
a director. Thus, the diffuse power of the university can be concen-
trated to some specific goal or purpose. Some have even attempted to
abolish publication as an outcome metric. The risk is that academic
freedom maybe impinged upon and creative energy suppressed. At
their best, core-​funded institutes bring together dedicated commu-
nities that work synergistically around particular problems or tech-
nologies, are liberated from the constraints of the short-​term grant
cycle to take on big long-​term questions and take risks. Historical
examples include, in their best years, the MRC’s LMB, the Institut
Pasteur in Paris, or the Institute for Advanced Studies at Princeton.
At their worst, core-​funded institutes can become ossified, bureau-
cratic, and stultifying.
How to choose
Whether funding is to be devoted to training, to individual fellows
or professors, or larger centres or institutes, a central problem for
the funder is how to choose the best use for the next Dollar, Pound,
or Euro. This is essentially a problem in prediction. As Yogi Berra
said: ‘It’s tough to make predictions, especially about the future.’
180
section 2  Background to medicine
Another way to consider this is to distinguish orthogonal dimen-
sions. One dimension is the fit to the strategic goals of the funder.
For example, for the Wellcome Trust a grant proposal on astronomy
is outside the remit and is a non​starter. On the other hand, spe-
cific goals are often developed (such as the President’s BRAIN ini-
tiative in the United States) and then proposals that align to these
are favoured. The strategy may be politically driven, by the focus
areas of the funder or by objective realities such as the burden of
disease. Getting the balance between supporting the top-​down de-
fined priority areas of the funder and the bottom-​up ideas from the
community is a constant challenge and a point of discussion among
all funders. In the end, no single funder can support everything it is
asked to support and a mixed economy of top-​down and bottom-​up
funding approaches are how most organizations work. A common
problem is that just wishing to make progress in a disease area and
defining a goal does not make the research tractable. One example
would be Nixon’s National Cancer Act 1971—​the technologies re-
quired were just not yet available when he made the statement and
the Act was signed into law with a view to eradicate cancer Likewise,
despite many years of effort, a vaccine for HIV has not yet been de-
veloped despite being critically needed and despite the very best ef-
forts of many people.
The overriding dimension is clearly scientific quality. If one could
somehow perfectly rank all available proposals across all the areas
of interest at any one time from best to worst, then the funder could
just support down the list until the money ran out. Of course, there
are problems with this as well. Aside from Yogi Berra’s problem, it
can be very difficult to quantify quality. As Ottoline Leyser recently
pointed out ‘Quality is qualitative’. This does not stop many, many
attempts to derive numerical metrics of research quality and im-
pact. Some approach objective truth, such as citation analysis: at
least it is apparent that some people are interested in a paper. But
unfortunately it usually takes many years before the relative import-
ance of a discovery becomes clear; much too late for any funding
decision, and citation analysis can just reflect the inherent conser-
vatism of the modern science environment. People work, are cited,
get funded, and promoted by working on projects which others also
work on.
At present, almost all funders rely heavily on the opinions of sci-
entific advisors. This can be by written peer-​review, discussion in
committees, or by direct face-​to-​face interviews of applicants before
committees. All of these are attempts to rank applications or appli-
cants to allow for funding decisions to be made. All of these tech-
niques contain possible artefacts and are rarely truly quantitative. It
is also difficult to show that these processes are actually accurate in
predicting success.
The deepest problem for funders is to judge quality and oppor-
tunity across different fields. Most avoid the problem by preallocating
funds by area. Some (HHMI and the Wellcome Trust) do rank appli-
cations and applicants across widely different fields. This is done by
a series of screens: first of many applications for technical feasibility
and broad promise. Thus field-​experts can remove proposals that
are technically totally flawed, but these are increasingly rare. But the
later stages converge on a smaller number of (mostly by this stage
exceptional) candidates interviewed by a group of scientific advisors
drawn from across many disciplines. Inevitably it can be difficult to
ensure all disciplines are equally represented and all equal in how
they judge their own discipline and other disciplines. There can also
be the challenge of decision by committee which can lead to conser-
vative judgements, or a single negative view ending the prospects of
an application. In the end, judgements must be made.
Lastly, an important principle has become acknowledged by most
funders of research: that a grant is not a contract. In other words, if a
scientist proposes a project, that project should be considered as an
intellectual exercise: can the candidate state a clear and important
question, can they think through the problems, can they work out
how many people and years and dollars will be needed. But once
they have succeeded and been awarded the funds, then they must be
free to follow their path freely towards the best results. In addition,
these may have very little to do with the original proposal. If the
scientist discovers something important, or solves a big problem, or
makes a breakthrough, no one will ask if it was in the proposal from
five years before. If they are wise, neither will they care too much
which journal it was published in.
FURTHER READING
Milton J (1667). Paradise lost. Book xi.
The Onion (1997). World Death Rate Holding Steady at 100 Percent.
https://​www.theonion.com/​world-​death-​rate-​holding-​steady-
​at-​100-​percent-​1819564171

2.18 Fostering medical and health research in reso
2.18 Fostering medical and health research in resource- constrained countries 181
ESSENTIALS
Access to quality healthcare and education for all are essential elem-
ents underpinning national development as well as prerequisites for
personal well-​being. This makes it important to grasp the inequalities
that still prevail in human and institutional capacity, including access
to and use of research resources. The goals of fostering medical re-
search in resource-​constrained countries are: to bring outstanding
science to bear on their medical and health problems; invest in
building the requisite medical and health research systems; develop
an evidence base to inform policies and programmes, resource al-
location, and health practice; enhance scientific processes, findings,
explanation, and ‘discovery’ by drawing on local know-​how; and
gain sustained engagement with policymakers, senior managers, and
community leaders.
Context is all-​important: every country, no matter how small or
resource-​constrained, should have the capacity to conduct essential
national health research to identify, prioritize, and respond to its own
health problems. Critical to fostering medical research1 in resource-​
constrained countries is the need for balanced, equitable collabor-
ations that emphasize longer-​term building of human and institutional
capacity. This speaks to the true opportunity offered by the promise
and practice of research in resource-​constrained countries: to con-
tribute widely to both national and global health development and,
in so doing, to major and equitable progress in the human condition.
Introduction
Medical and health research today take place within a new era of
‘global health’, south-​north-​south collaboration, and 21st century
interdisciplinary knowledge convergence. These provide promise,
challenges, and opportunities, particularly for fostering research in
resource-​constrained countries.
Globalization today is premised on speed of communication
and interconnectedness which serve to ‘level’ peoples’ aspirations and
expectations across all settings; similarly, the questions, focus, and
methods of science are converging more than ever before despite the
variation in socioeconomic contexts. Anticipating this, a multipolar
Global HIV Vaccine Enterprise was proposed in 2003. This cata-
lysed an upsurge in research and institutional development towards
a still-​elusive vaccine against HIV/​AIDS.
Most visible of 21st century innovations is the internet, which is re-
shaping all forms of communication from the social to the scientific.
Less visible, though in time as pervasive, is the biological revolution
which has already ensured a new scientific ‘universe’ reflected in un-
ravelling of the human genome along with the tools to more deeply
understand and manipulate our genomic complement. The full ram-
ifications and impact of the ‘biological revolution’ on science, society,
and well-​being are yet to be fully understood and appreciated.
Despite the sometimes intolerable hardship experienced by mi-
grant or refugee populations—​ever-​present in today’s globalized ‘in-
stant media’—​the moral standards the world uses to judge its actions
are today more apparent and widely held. This is reflected in global
debates on climate and the environment, sustainable energy use and
acceptable standards of political governance that have given rise to
equity-​oriented research initiatives such as Future Earth and Urban
Health and Wellbeing, which seek to harness science across discip-
lines, sectors, and geographies. Contemporary UN Declarations as
well as national constitutions—​that of South Africa, our own country,
for example—​enshrine human dignity, the achievement of equality,
and advancement of human rights and freedoms as supreme qualities.
Access to quality healthcare and education (for all) are recognized
as essential elements underpinning national development as well as
prerequisites for personal well-​being. This makes it important to grasp
the inequalities that still prevail in human and institutional capacity,
including access to and use of research resources. For example, national
research and development (R&D) expenditure among OECD countries
was estimated at 2.4% of gross domestic product (GDP) in 2012, with
the R&D expenditure in the United States comprising 2.79% of GDP,
and Korea 4.36%. While China’s R&D expenditure is growing (cur-
rently estimated at 1.98% of GDP), R&D expenditure in South Africa,
India, and the Russian Federation fall below this average (most recent
estimates amounting to 0.76%, 0.81%, and 1.12% of GDP, respectively).
2.18
Fostering medical and health research
in resource-​constrained countries
Malegapuru W. Makgoba and Stephen M. Tollman
1  The term ‘health research’ encompasses a breadth of disciplines to investigate
the determinants of health and illness and the interventions, system changes, and
policies needed to address these.
182
section 2  Background to medicine
A further indicator of this disparity is the level of researchers per
thousand employed people in these different regions. The OECD
average is 7.7 people per thousand employed in 2011, with Finland
reporting the highest ratio of 16.06. South Africa and China report
relatively low estimates of 1.43 and 1.83 people per thousand em-
ployed in research. In the health sector, OECD countries average
3.16 physicians to 1000 inhabitants with the United Kingdom re-
porting a ratio of 2.8 and the United States a ratio of 2.5 in 2011. In
contrast, India and South Africa most recently estimate 0.7 and 0.8
physicians to 1000 people, respectively.
It follows that across low-​ and middle-​income countries (LMICs)
there are different constraints with little uniformity in the infrastructure
available for research, reflecting variable investment and the uneven
status of research in national development. Pockets of excellence exist
in middle-​income settings that range from Brazil to China to India and
South Africa—​but widely inadequate research and infrastructure devel-
opment characterize most lower-​income countries thereby denying the
fruits of research and scientific advance to many of the world’s poorest
and marginalized communities and, indeed, to future generations.
Critical, unresolved research questions confront the scientific,
policy and citizen leadership of LMICs—​on the rapidly evolving
burden of disease and risk and their social and biological deter-
minants in contrasting environments, coupled with understanding
and improving the health and social systems that can address these
effectively. Such issues pose a profound challenge to scientists and
practitioners within and across disciplines. They shine a light on
the sociopolitical context in which such issues are being addressed.
Whereas scientific and social progress hold the promise of major
advances in personal and population health, persisting disparities
can exacerbate inequalities and render ‘catch-​up’ all but out of reach.
Fostering medical research in resource-​constrained countries is
not only essential but also requires institutional and individual sci-
entific development—​with attention to career paths, research lead-
ership, sound ethics, the sustaining of versatile research platforms
(molecular, clinical, service and population-​based), and institu-
tional support. Broadly put, the goal is to:
•​	 Bring outstanding science to bear on the medical and health prob-
lems of resource-​constrained countries and communities.
•​	 Invest in building the requisite medical and health research sys-
tems based on sound governance and long-​term institutional cap-
acity strengthening.
•​	 Build an evidence base to inform policies and programmes, re-
source allocation, and health practice from primary care to ad-
vanced clinical settings; these should focus on national priorities
and tackle inequities.
•​	 Enhance scientific processes, findings, explanation, and ‘dis-
covery’ by drawing on local know-​how and indigenous know-
ledge and expertise.
•​	 Gain sustained engagement with policymakers and senior man-
agers as well as community leadership, to foster the uses of evi-
dence and encourage enhanced investment in health research
and healthcare.
The work of the Commission on Health Research for Develop­ment
was a landmark which, in 1990, published ‘Health Research: Essential
Link to Equity in Development’. This far-​reaching report envisaged
that every country, no matter how small or resource-​constrained,
should have the capacity to conduct essential national health re-
search in order to identify, prioritize, and respond to its own health
problems. In addition, that all countries, no matter how limited their
resource base, should contribute to the global health research enter-
prise. These aspirations—​widely shared and as timely today—​clearly
align with aspirations for the new ‘global health’. They highlight the
national and scientific competencies needed to ensure deep and sus-
tained progress towards the Millennium Development Goals and
their recent successor, the Sustainable Development Goals.
Changing profiles of health, disease, and
well-​being: Setting priorities for health research
Health problems facing resource-​constrained countries—​where poverty
is widespread amid income and social inequalities—​can occur with un-
expected intensity, evidenced by the recent Ebola epidemic in the West
African countries of Guinea, Liberia, and Sierra Leone. While the infec-
tion itself carried an overwhelming mortality rate, the knock-​on effects
were similarly devastating: primary care services were hit hard, affecting
immunizations and maternal and child care, while household liveli-
hoods were severely undermined contributing to vulnerability of the do-
mestic economies. The emergency reaction of Médecins Sans Frontières
was critical, but a generally sluggish global response prompted soul-​
searching on the future role of high-​income countries like the United
Kingdom, agencies such as the World Health Organization (WHO), and
international strategies for vaccine and drugs development. A resurgent
epidemic in the Democratic Republic of Congo (DRC), aggravated by
community mistrust of measures introduced to limit the outbreak, high-
lights the complexity of epidemic containment.
Reflected in recent assessments of the global burden of disease,
complex health transitions are underway in LMICs. Rapid changes in
national health profiles are being fuelled by the socioeconomic forces
driving (for example) labour migration within and across national
boundaries, and new forms of transport and communication that
link people and places at unprecedented speed and scale. Together,
these act to juxtapose infectious conditions (malaria, TB, HIV/​AIDS,
neglected tropical diseases) with unfolding non-​communicable dis-
eases—​notably cardiovascular disease, stroke, and diabetes, respira-
tory conditions, cancers, and mental illness. The natural history of
these diseases and their attendant risks pose challenges to personal
development along the life course—from conception, birth and child-
hood to puberty, and adolescence to adulthood—​in a world where
populations are ageing while national incomes remain uncertain.
This ‘dual burden’ of disease—​graphically described as a colli-
sion of epidemics—​amounts to a ‘quadruple burden’ when the un-
finished agenda of maternal, neonatal and child health is included
along with persistently high levels of injury (violent and uninten-
tional) and rising accident rates (from motor vehicles especially).
Such a rapidly changing disease burden must profoundly impact
on the framing of national and global research priorities—​and the
resources and capabilities required to enable effective research.
Context is all-​important
Our long-​standing African experience brings home the overriding
importance of social and political context in fostering, conducting,
2.18  Fostering medical and health research in resource-constrained countries
183
interpreting, and then ensuring the applications of research. Indeed,
given many countries’ history and experience with the deliberate
marginalizing of poorer communities (now in the voting majority
post-​independence), research in LMICs cannot be separated from
broader public concerns with national and social development. This
manifests in several ways.
Contestation between science and politics
This translates to a pervasive culture of resorting to political au-
thority to resolve issues that elsewhere would be addressed by public
sector executives. Research findings, whatever the evidence, and
particularly if results conflict with prevailing social norms or pol-
itical ideologies, may be actively contested with appeals to popular
leadership for guidance and direction. Leading politicians in LMICs
face high, at times unrealistic, pressure from their constituencies to
resolve deeply entrenched social and development concerns—​hence
politicians’ and policymakers’ preoccupation with answers in con-
trast to researchers’ emphasis on understanding.
Examples from our recent history illustrate vividly the interplay
between politics and bio-​sciences: Adolf Hitler was a dominant pol-
itical figure who, on rising to power in Germany, used science to
justify the political ideology on which the Nazi state was based, and
which justified barbarous treatment of ‘non-​Aryans’. Trofim Lysenko
was an influential Soviet agricultural scientist, with a powerful ally
in Joseph Stalin, who for over 40 years undermined, then destroyed,
biological science (notably genetics) teaching and research in the
Soviet Union. Thabo Mbeki and Manto Tshabalala-​Msimang, for-
merly president and minister of health in South Africa, confused a
nation and led it into denial about HIV/​AIDS causation for a decade,
with devastating consequences for human lives (see next).
Deep respect for human and community rights (reflected
in the UN charter as well as national constitutions)
Disparities in wealth and education—​as much within as between
countries—​raise concerns about the rights and protections of vulner-
able persons (such as children, the disabled, or unemployed) and com-
munities in all spheres, and notably in research. Concern for the dignity
of research participants, confidentiality regarding personal informa-
tion, and equitable access to findings and resultant benefits (e.g. new
diagnostic tests, therapeutics, equipment) weigh heavily on public rep-
resentatives who increasingly are held to account by an activist citizenry.
Sensitivity to purpose and relevance
Keen interest in the focus of research—​what research is conducted,
how, and for what purpose—​lies behind the strenuous debates on
research priorities in LMICs and presents great opportunity to
work at the frontiers of socially responsive science. Such research
environments support interdisciplinary efforts that engage with
community perspectives; and tend to a more holistic approach
to both the conduct and translation of research including health
systems development and all aspects of professional practice.
It follows that while concern for the impacts of research is plainly
evident in industrialized countries, in LMICs it is becoming an im-
perative. While working relationships with public leadership bear
on the applications of research to policy development, as important
are scientists’ links with public representatives for drawing attention
to the social importance of the research enterprise. This will help
counter the tendency for findings to play but little part in public
sector investment decisions—​whereas findings and understanding
should figure prominently in development priorities and resource
allocation.
An unlikely alliance: Science and politics
At first glance, the apparently neutral terrain of research, scien-
tists, and the public good seems a world away from the contested,
at times cut-​throat world of national politics. Yet history is replete
with examples—​from Stalinist Russia (and the Lysenko era of eu-
genics, discussed earlier) to Hitler’s Germany (that posited Aryan
races as the apex of human development and condoned human
experimentation on those judged ‘lesser’ persons)—​where science
served the goals of a political master plan. In response, ethical
codes and principles governing the conduct of research—​most re-
cently, the ninth revision of the Helsinki Declaration—​are upheld
across the globe.
Nevertheless, in all settings there are examples where political
goals have undermined ethical practice. In South Africa at the close
of the 20th century, the emergence of HIV/​AIDS confronted politi-
cians with an intolerable dilemma: face up to a profound health and
social challenge evolving in concert with national liberation, and sup-
port a talented scientific community determined to address this; or
buy the argument of the AIDS ‘denialists’ who rejected the existence
of a human immunodeficiency virus and attributed AIDS solely to
widespread deprivation and related nutritional and immunological
compromise.
President Thabo Mbeki’s ‘flirtation’ with HIV/​AIDS denialists,
notably Peter Duesberg, ushered in a decade of national confu-
sion and denialism reflected in a crisis in the governance and
independence of medical research; a health policy denial of anti-
retroviral therapy to HIV-​positive pregnant mothers and patients,
with consequences estimated at 330 000 deaths; and society’s pro-
foundly compromised ability to impact on interpersonal and peer
or community behaviour and halt the spread of infection. Thus,
though difficult to accept, the consequence of the delayed rollout
of antiretroviral medication was the loss of hundreds of thousands
of lives.
Today, some 20 years later, South Africa and its Department of
Health can take credit for the widespread and sustained take-​up
of the largest antiretroviral therapy programme worldwide. This
is saving countless lives, rapidly improving life expectancy, and
preserving the livelihoods of many of the country’s poorest citi-
zens. Indeed, this recent but hardly isolated example illustrates how
powerful political figures influencing health research can determine
the future direction and success (or not) of nations.
Fig. 2.18.1 provides a generic framework for the dynamic inter-
actions involving science and research in national contexts. The
political environment, prevailing culture and social norms, and the
pursuit of human rights all act on the conduct of research and use of
the findings/​evidence that result.
Acting together, these influences can ensure that the applications
of research serve as a tremendous power for social good and human
development—​but equally, they hold the potential for political ma-
nipulation of scientists who may be used (even coerced) to legitimate
state actions in support of ends that fly in the face of obligations to
advance population health and well-​being.
184
section 2  Background to medicine
Conclusion
While much in our past experience emphasizes ‘divides’—​between
rich and poor, north and south or east and west, stark disciplinary
divisions, and a divide between the social and biological determin-
ants of health—​we have entered a world of ‘convergences’ that is
fostered and reinforced by a global health movement that empha-
sizes our common future. Work that is ‘convergent’ lies at the heart
of global improvements in child health and the great reductions in
mortality from HIV/​AIDS and malaria.
Critical to fostering medical research in resource-​constrained coun-
tries is the need for balanced, equitable collaborations that emphasize
longer-​term building of human and institutional capacity. This vision
needs to be tempered by the reality that the gap in resources—​including
statistical capability, computational and technical expertise—​remains
profound and is paralleled by a limited clinical-​physiological-​molecular
evidence base. Addressing this as a common, collaborative enterprise
would bring returns to research and practice in all settings and fields.
Beyond this, positive recognition is justified for the sophisticated, lo-
gistically complex field-​based research that increasingly extends from
molecular to population levels, and similarly for generating the ‘public
good’ data being made widely accessible.
In summary, we are at the cusp of a new paradigm of research
collaboration that fulfils the value-​based aspirations of ‘global
health’, bringing the best science to bear on the complex challenges
facing vulnerable communities in resource-​constrained countries,
while recognizing the pressures to ‘deliver’ that are faced daily by
scientists, policymakers, and politicians. This speaks to the true
opportunity offered by the promise and practice of research in
resource-​constrained countries:  to contribute widely to both na-
tional and global health development and, in so doing, to major and
equitable progress in the human condition.
Acknowledgements
We thank Carren Ginsburg for sourcing comparative data to reflect
national differences.
FURTHER READING
Abubakar I, et al. (2018). The UCL-Lancet Commission on Migration
and Health: the health of a world on the move. Lancet, 392, 2606–54.
Chigwedere P, et al. (2008). Estimating the lost benefits of antiretro-
viral drug use in South Africa. J Acquired Immune Deficiency
Syndrome, 49(4), 410–​15.
Commission for Africa (2005). Leaving no-​one out:  investing in
people: our common interest. Penguin, London.
Commission on Health Research for Development (1990). Health
research: essential link to equity in development. Oxford University
Press, New York.
Future Earth (2015). Urban Health and Wellbeing: research for global
sustainability. http://​www.futurearth.org/​
Jamison DT, et al. (eds) (2006). Disease and mortality in sub-​Saharan
Africa. The World Bank, Washington DC, pp. 1–​387.
Jamison D, et al. (2017). Universal Health Coverage and Intersectoral
Action for Health: Key messages from Disease Control Priorities,
3rd edition. Lancet.
Klausner RD, et al. (2003). The need for a global HIV vaccine enter-
prise. Science, 300, 2036–​9.
Koplan JP, et al. (2009). Towards a common definition of global health.
Lancet, 373, 1993–​5.
Makgoba MW (1999). The South African Medical Research Council:
Africanizing health research. Nat Med, 5(4), 367–​70.
Makgoba MW (2000). HIV/​AIDS: the perils of pseudoscience. Science,
288, 1171.
Makgoba MW (2002). Politics, the media and science in HIV/​AIDS:
the peril of pseudoscience. Vaccine, 20, 1899–​904.
Nattrass N (2008). AIDS and the scientific governance of medicine in
post-​apartheid South Africa. African Affairs, 107, 157–​76.
OECD (2014). Factbook 2014: Economic, Environmental and Social Statistics.
OECD Publishing 2014. http://​dx.doi.org/​10.1787/​factbook-​2014-​en
Robertson LJ, Makgoba MW (2019). Mortality analysis of people with
severe mental illness transferred from long-stay hospital to alter-
native care in the life Esidimeni tragedy. SAMJ, 108, 813–17.
Streatfield PK, et al. (2014). Cause-​specific mortality in Africa and
Asia: evidence from INDEPTH health and demographic surveil-
lance system sites. Glob Health Action, 7, 25362.
The Lancet (2015). 1 year on—​lessons from the Ebola outbreak for
WHO. Lancet, 385, 1152.
The Lancet (2015). GBD 2013 mortality and causes of death collaborators.
Global, regional, and national age-​specific all-​cause and cause-​specific
mortality for 240 causes of death, 1990–​2013: a systematic analysis for
the Global Burden of Disease Study 2013. Lancet, 385, 117–​1.
Tollman SM, et al. (2008). Implications of mortality transition for pri-
mary health care in rural South Africa: a population-​based surveil-
lance study. Lancet, 372, 893–​901.
World Bank (2015). World Bank Data:  Research and Development
Expenditure. World Bank, Washington DC. http://​data.worldbank.
org/​indicator/​GB.XPD.RSDV.GD.ZS
World Bank (2015). World Bank Data: Physicians (per 1,000 people).
World Bank, Washington DC. http://​data.worldbank.org/​indicator/​
SH.MED.PHYS.ZS
World Medical Association (2013). World Medical Association
Declaration of Helsinki:  ethical principles for medical research
involving human subjects. JAMA, 310 (20) 2191–​4.
Human rights
Politics and
the
political
environment
Culture and
social norms
Research
Knowledge/Evidence
on health and
well-being
Fig. 2.18.1  Framework highlighting social drivers and interactions that
influence the conduct of research and use of evidence to advance human
health and well-​being.

2.19 Regulation versus innovation in medicine 185
2.19 Regulation versus innovation in medicine 185
ESSENTIALS
Two tragedies created drug regulatory authorities in the form they
exist today. The marketing in 1937 of an elixir of sulphanilamide using
diethyl alcohol as the solvent led to over 100 children dying from
acute kidney injury. The use from 1957 of thalidomide as a treatment
for morning sickness led to the development of phocomelia (and
other abnormalities) in the fetuses of about 10 000 mothers who had
been given it during their pregnancies.
Drug regulatory authorities initially tended to be risk averse, but
now recognize concern that the process of drug regulation might
inhibit the development and licensing of novel products. They now
pursue a more risk-​based approach to regulation and seek to support
safe innovation, with arrangements including scientific advice pro-
grammes, orphan drug approval processes, and expedited approval
processes, although these are not without their critics.
Regulation of medicines
The regulation of medicines started, in a modest way, during the
18th and 19th centuries with the publication of pharmacopoeias
intended to ensure the quality of medicinal products sold to the
public. These did not specify the purposes for which products
should be used, and they were advisory rather than compulsory,
although they probably had some impact. The 20th century, how-
ever, saw substantial changes arising from a combination of scandal
and tragedy.
In 1906 the British Medical Association published a slim volume
entitled Secret Remedies: What they Cost and What they Contain,
which was an exposé of the patent medicines industry. Beechams
Pills—​containing only aloes, powdered ginger, and powdered
soap—​were promoted for numerous purposes that ranged from
‘constipation’ to ‘maladies of indiscretion’ and even ‘menstrual de-
rangements’, which was interpreted as a covert claim as an abortifa-
cient. The public in both the United States and Britain were horrified.
In America, where the US Congress was discussing a ‘Pure Foods
Act’, the legislation was extended to medicines, thus creating the
Food and Drugs Administration (FDA). In the United Kingdom
things moved rather more slowly. After the publication, in 1909, of a
sequel entitled More Secret Remedies, Parliament established a Select
Committee to advise on what measures should be put in place to
protect the public. Its report recommended that legislation be en-
acted to regulate the supply of medicines to the public. Sadly, the
Committee published its findings the day before the start of the First
World War and it was largely forgotten.
Tragedies that shaped drug regulation
Two tragedies created drug regulatory authorities in the form
they exist today. In 1937 an American pharmaceutical company,
Massengill, began marketing an elixir of sulphanilamide using di-
ethyl alcohol as the solvent. Over 100 children died from acute
kidney injury a result of the solvent’s nephrotoxicity. As a conse-
quence, the FDA’s powers were enhanced, but virtually nothing was
done in Europe.
In 1957 the German drug company Grünenthal placed thalidomide
on the market as a treatment for morning sickness without under-
taking any reproductive toxicology tests. It became apparent, after
a delay of almost three years, that thalidomide caused phocomelia
­(malformations of the arms and/or legs), as well as other abnormal-
ities, in the fetuses of mothers who had been given it during their preg-
nancies. It is estimated that, worldwide, 10 000 babies were afflicted,
although the thalidomide disaster was largely avoided in America as a
result of the legislative response to the sulphanilamide tragedy.
In response to these disasters—​particularly the thalidomide
tragedy—​national drug regulatory authorities were established
in almost all countries. They all have legal powers—​within their
jurisdictions—​to ensure that medicines placed on their markets are
of good quality, are effective in the uses promoted by their manu-
facturers, and that they are safe in relation to their efficacy. Within
the European Union almost all new drugs are now authorized by
the so-​called ‘centralized procedure’ which means that they become
licenced, simultaneously, across all 28 member states. The European
Medicines Agency (EMA) ​undertakes a detailed scientific scrutiny
of the evidence through its Scientific Committees and makes a
­recommendation whether or not new products should be granted
so-​called ‘marketing authorization’.
Because of their historical origins, drug regulatory authorities
initially tended to be risk averse. Although they were not opposed
2.19
Regulation versus innovation in medicine
Michael Rawlins
186
section 2  Background to medicine
to pharmaceutical innovation, there had been some concern that
the process of drug regulation might inhibit the development and
licensing of novel products. However, regulatory authorities have
increasingly recognized that they have an important role in not
only protecting public health, but supporting innovation. They
have therefore taken considerable steps over recent years to ensure
a more risk-​based approach to regulation, reducing regulatory bur-
dens where possible, and putting in place measures to support safe
innovation.
Support for innovation
Provision of advice
Regulatory agencies have, for many years, offered the sponsors of
new products the opportunity to seek scientific advice on the de-
sign of their development programmes. At face-​to-​face meetings,
sponsor companies meet with the scientific staff of drug regu-
latory agencies to discuss their proposed development plans and
protocols, including the outcome measures they plan to adopt in
demonstrating efficacy. More recently the FDA and EMA have in-
stituted arrangements for the two agencies to provide joint scien-
tific advice.
The scientific advice programme is entirely voluntary, with agen-
cies covering their costs by charging fees for the services they offer.
The scientific advice that is provided is ‘non​binding’ on either
party, but nevertheless these scientific advice services are widely
used by the sponsors of new products, and few companies would
now complete a drug development programme without having
sought regulatory advice from the FDA and the EMA or National
Authorities in Europe.
In the United Kingdom the Medicines and Healthcare Products
Regulatory Agency (MHRA) has, in addition to its scientific advice
provisions, recently established an Innovation Office. This offers
help to all those developing particularly innovative products or tech-
niques/​processes, including small and medium-​sized enterprises as
well as academics who have developed a novel medicine or device,
or a novel approach to the development or manufacture of a product.
Examples of such innovative products include Advanced Therapy
Medicinal Products (cell-​based therapies), nanotechnologies,
stratified medicines, novel drug/​device combinations, and advanced
manufacturing techniques.
Orphan drugs
In the early 1980s it was recognized, initially in the United States,
that manufacturers were being deterred from developing drugs for
rare diseases. The development and ongoing costs of such products,
to meet the demands of regulation to ensure appropriate standards of
quality safety and efficacy, could never be recouped from their sales.
The US Congress, in 1983, passed legislation that allowed the US
FDA to offer various incentives to sponsors seeking to develop prod-
ucts for rare (so-​called ‘orphan’) diseases. Comparable measures
were subsequently introduced in other jurisdictions including the
EU, Japan, Singapore, and Australia. In the United States an orphan
disease is defined as a condition with less than 200 000 affected per-
sons: in the EU it is defined as a condition with a prevalence of not
more than 5 per 10 000 population. To further reduce the burden on
manufacturers applying for orphan drug status, the FDA and EMA
now use a common application process for both agencies, although
each maintains separate approval processes.
The orphan drug programme has been an undoubted success. In
the EU, for example, more than 100 orphan drugs have been granted
marketing authorization since the programme began in 2000.
Expedited approval programmes
In the mid-​2000s, the FDA began to introduce arrangements that
would permit promising new products to become more rapidly
available to patients. To meet the FDA’s requirements for expedited
approval, the product must fulfil three criteria. First, the product
should be intended to treat a ‘serious’ condition that has substantial
impact on patients’ day-​to-​day functioning. Secondly, the product
should expect to be an improvement any ‘available therapies’ that
are approved for the same indication as the new drug. Thirdly,
the product should meet ‘unmet clinical need’, defined as a condi-
tion whose treatment is not adequately met by currently available
treatments.
In the United Kingdom, the MHRA have recently introduced
analogous arrangements. A promising innovative medicine desig-
nation gives sponsors an indication that a product may be eligible
for the Early Access to Medicines Scheme (EAMS) based on early
clinical data. Under the scheme, the MHRA provides an opinion on
the benefit/​risk balance of the medicine based on the data available
when the EAMS submission is made. If successful, the MHRA gives
a positive opinion that supports prescribers in deciding whether
to use the product in patients with life threatening or seriously
debilitating conditions before marketing authorization is granted.
Discussions are currently taking place with a view to extending the
principle of early designation of a promising medicine in Europe.
The EMA is currently undertaking a pilot study of the use of ‘adap-
tive pathways’ to accelerate the availability of novel products for
unmet clinical need. In this scheme a sponsor first targets a narrow
indication in an area of high unmet need and subsequently develops
the product for broader indications. During the development pro-
cess input is obtained from other stakeholders such as health tech-
nology assessment (HTA) bodies. The scheme is within the existing
legislative framework and will likely use provisions such as con-
ditional approval. Once the first approval is granted, the sponsor
will then be required to collect additional information about the
product’s performance in clinical use so that—​provided the early
promise is fulfilled—​full marketing authorization is granted or the
indication broadened.
Are these mechanisms to support
innovation useful?
National drug regulatory authorities are clearly making serious ef-
forts to promote the development of safe and effective innovative
medicines. In doing so they must continue to balance a product’s
benefits against its risks, but the approaches that authorities are
taking is not without controversy.
Some commentators have criticized the arrangements for providing
scientific and regulatory advice to companies, whether they be small
or large. They believe that doing so will (or has) lead to ‘regulatory
2.19  Regulation versus innovation in medicine
187
capture’, whereby the agendas of regulatory authorities are manipu-
lated to the commercial advantage of pharmaceutical companies.
There is no evidence that this has occurred in either Europe or North
America, but the fact that such accusations have been made can leave
regulatory authorities feeling—​at best—​uneasy. The converse is that,
without appropriate advice, those developing novel products run the
risk of failing to understand what is required of them, undertaking
inappropriate or unnecessary studies, and wasting substantial sums
of money.
Other commentators of a utilitarian persuasion have suggested
that (at least in the EU) the orphan drugs legislation should be re-
pealed. They argue that their development costs are still very high,
even with incentives, and as a consequence the costs of providing pa-
tients with orphan products is likely to deprive those with common
conditions of cost-​effective care. This attitude is countered by egali-
tarians who claim that society has a duty—​within reason—​to help
those with any condition achieve the quality of life to which they
are entitled. There is no substantial appetite at present in the United
Kingdom, the European Union, or North America to repeal the
orphan drugs arrangement.
The arrangements for expedited approval in the United States
have also been criticized on two grounds. First, it has been claimed
that the number of safety warnings issued after the launch of prod-
ucts undergoing expedited approval have been greater than those
approved under conventional approaches. This may well be correct
because the data on which expedited approvals are made are likely
to be less. Nevertheless, if patients with unmet clinical need are to
have access to products that may have a profoundly beneficial effect
on their lives, a reduced safety database would seem to be a price
worth paying. The second criticism is that the benefits have been
limited, at least for some products that have gone through an exped-
ited approval. This has been particularly alleged in respect of some
anticancer agents, when approvals for products used in later stages
of the disease have been based on intermediate endpoints (such as
reductions in tumour size), with only a modest increase in longevity.
Despite these criticisms regulatory authorities are continuing to
develop arrangements that would allow patients earlier access to
promising new products. It is in the best interest of patients and the
public that they succeed.
FURTHER READING
Carpenter D (2010). Reputation and power: organizational image and
pharmaceutical regulation at the FDA. Princeton University Press,
Princeton.
Emanuel M, et  al. (2012). Thalidomide and its sequelae. Lancet,
380(9844), 781–​3.
Stephens MDB (2010). The dawn of drug safety. George Mann
Publications, Easton, Winchester.

2.2 Evolution Medicine’s most basic science 39
2.2 Evolution: Medicine’s most basic science 39
ESSENTIALS
The role of evolutionary biology as a basic science for medicine is
expanding rapidly. Some evolutionary methods are already widely
applied in medicine, such as population genetics and methods
for analysing phylogenetic trees. Newer applications come
from seeking evolutionary as well as proximate explanations for
disease.
Traditional medical research is restricted to proximate studies of
the body’s mechanism, but separate evolutionary explanations are
needed for why natural selection has left many aspects of the body
vulnerable to disease. There are six main possibilities: mismatch,
infection, constraints, trade-​offs, reproduction at the cost of health,
and adaptive defences. Like other basic sciences, evolutionary biology
has limited direct clinical implications, but it provides essential re-
search methods, encourages asking new questions that foster a
deeper understanding of disease, and provides a framework that or-
ganizes the facts of medicine.
Physicians who understand evolution recognize that bodies
are not designed machines but jury-​rigged products of millions of
years of natural selection that work remarkably well, given that no
trait can be perfect, and that selection maximizes reproduction, not
health.
Introduction
This medical textbook is, as far as we know, the first to offer a chapter
on evolutionary biology. The occasion of the 160th anniversary of
the publication of The Origin of Species makes it fitting, albeit some-
what delayed. Medical students are taught how the human body
is (anatomy), and how it works (physiology), but seldom are they
taught why it works (natural selection) or whence it comes (evo-
lution). It is as though car mechanics were taught how a car works,
and how to fix breakdowns, but never where it came from (factories
and designers’ drawing boards) nor the purpose for which it was de-
signed (transport along roads).
Things are beginning to improve. The past 15 years have seen a
series of books, articles, and meetings that report new applications
of evolutionary biology to medicine. Evolution is as fundamental
to medicine as physics or chemistry. This chapter cannot review its
whole scope. We can only illustrate a few core principles in hopes of
encouraging further reading.
Core evolutionary principles for medicine
Natural selection and adaptation
When individuals in a population vary in ways that influence their
genetic contribution to future populations, the average character-
istics of the population will change. This is not a theory; it is neces-
sarily true. Natural selection involves no design, no planning, and
no goal. The word ‘evolution’ refers more generally to any changes
over time in a population, whether from selection, mutation, genetic
drift, or migration.
Notwithstanding his most famous title, Darwin’s greatest contri-
bution was not his explanation of speciation, but his explanation
of adaptation. Recent research on the Galapagos finches known as
‘Darwin’s finches’ illustrates the point. During drought, only larger
seeds are available, so individuals with larger beaks get more food
and have more offspring. In just a few generations, the average beak
in the population became significantly larger after a drought. When
the rains came, and small seeds again became plentiful, selection
switched to favouring smaller beaks. No trait is adaptive except in
relation to a specific environment.
Levels of selection
Non​specialists often assume that natural selection should shape traits
to benefit groups. After all, if a species goes extinct, all the individuals
and their genes are lost. This ‘group selection’ fallacy was unmasked
over 40 years ago, but it continues to cause confusion in medicine.
For instance, one might expect pathogens to evolve low viru-
lence: killing off the host is surely not good for the group! However,
even long association of a host and pathogen does not necessarily
decrease virulence. People who are out of bed transmit a rhinovirus
2.2
Evolution: Medicine’s most basic science
Randolph M. Nesse and Richard Dawkins
Acknowledgement: Thanks to the Berlin Institute for Advanced Study for pro-
viding a fellowship to RMN that made preparation of this chapter possible.
40
SECTION 2  Background to medicine
faster; this selects for low virulence. The story is very different for
insect-​borne diseases. Plasmodium is transmitted faster from pa-
tients who are too sick to slap mosquitoes, so virulence is high for
malaria in humans (infected mosquitoes feel just fine).
Ageing can be similarly misunderstood. One might think that
senescence could speed the evolution of the species by making room
for new individuals. The species, however, is not the level at which
selection acts. Consider a lethal or deleterious gene that is expressed
only late in life. Many carriers will have passed on the gene before it
kills them. The same gene would be quickly selected out if it killed
individuals before they reproduced. We are all descended from indi-
viduals who died after having children. Not one of our direct ances-
tors ever died in childhood! Moreover, a pleiotropic gene that gives
a benefit early in life may be favoured, even if it causes deleterious
effects later, when selection is weaker. This evolutionary explan-
ation for senescence is now confronting remarkable new evidence
that single-​gene effects in the insulin signalling pathways can have
huge effects. The reasons why selection has not incorporated such
changes will prove most interesting.
Established applications
Some methods from evolutionary biology have long been applied
to medicine. Population genetics describes how natural selection,
mutation, migration, and drift account for shifting gene frequencies.
This body of knowledge has been a foundation for medicine since
the middle of the twentieth century, so we will only note a few new
applications.
It is now clear that the ability to digest lactose as an adult is the
exception, rather than the rule. In our ancestors, milk was a food
for babies only. New analyses show that the ability to digest lac-
tose as an adult has emerged on at least three separate occasions
in human prehistory, always in dairying cultures. Remarkably, the
selective advantage in these cultures has been huge, of the order
of 5 to 15%. The exact benefits remain to be fully understood but
calcium and vitamin D may be important, as well as getting more
calories.
Another example is the prevalence of mutations influencing the
acetaldehyde dehydrogenase genes in some populations (especially
in southeast Asia). Carriers get sick when they drink alcohol. Is
the prevalence of this mutation a result of random genetic drift, or
does it give some advantage, perhaps by decreasing the risk of alco-
holism? New data show that it does protect against alcoholism and
that strong selection has acted at this locus; it is at the centre of one of
the largest haplotypes in some populations. This supports the role of
alcohol, but the geographical distribution suggests that diet or other
cultural variations may be responsible.
Genetic methods for tracing phylogenies of pathogens have long
been available. Influenza strains are tracked so assiduously that it is
possible now to predict some characteristics of likely future epidemic
strains—​invaluable information for vaccine design. Epidemics of
pathogen-​contaminated food are now routinely traced back to the
source using genetic data. It has even been possible to trace specific
cases of HIV back to a specific source, because rapid mutations leave
a clear trail.
Evolutionary methods also can also be applied to somatic cell
lines within a body, for instance, to determine if the cells in a tumour
are all identical or if subclones are competing in the tumour. The im-
plications for customizing chemotherapy are substantial.
Evolutionary aetiology
Most medical research provides proximate explanations based on
the anatomical and chemical details of the body’s mechanisms.
However, even knowing every detail about a trait offers only one half
of a complete biological explanation. The other half is provided by
an evolutionary explanation of how that trait came to exist in the
first place. There are two kinds of evolutionary explanations: the first
is a phylogenetic explanation based on the sequence of prior traits
across evolutionary history; the other is an explanation of what evo-
lutionary forces account for the changes across time. Most often,
this requires an understanding of how the trait gives a selective
advantage.
Explain vulnerabilities, not diseases
Evolution can explain why aspects of the body have been left vul-
nerable to disease. Why do we have wisdom teeth, and a small birth
canal? Why do we so often develop lower back pain and hip problems?
Why hasn’t selection shaped our immune systems to better eliminate
pathogens and cancer cells? Answering such questions in an evolu-
tionary way is often challenging. A framework can help to organize
the effort. There are six main reasons why bodies have vulnerabilities
to disease despite the actions of natural selection (Box 2.2.1).
Mismatch
Chronic ‘diseases of civilization’ such as obesity, hypertension, and
diabetes are now pandemic. The motivations that make us eat too
much and exercise too little were shaped for an environment where
sweet, fatty, or salty foods were good for us, and excess exercise
could be fatal. Recognizing the origins of our unhealthy preferences
does not change them, but it illuminates the source of the problem
and possible solutions.
Similarly, allergies and autoimmune disorders are more common
in developed societies. Our immune systems evolved when people
were routinely exposed to intestinal parasites and pathogens. In
their absence, inhibitory immune cells are not stimulated, leaving
the system overactive and responsive to self. An attempt to recreate
the original intestinal environment by administering whipworm ova
has proved remarkably effective as a treatment for Crohn’s disease.
Coevolution
We remain vulnerable to infections because pathogens evolve faster
than us. Just how fast is demonstrated by the rapid rise of resistance
Box 2.2.1  Six kinds of evolutionary explanations for vulnerability
•	 Mismatch between aspects of our bodies and novel environments
•	 Pathogens that evolve faster than we do, and resulting costly defences
that cause harm themselves
•	 Constraints on what natural selection can do
•	 Trade-​offs that keep any trait from being truly perfect
•	 Traits that increase reproduction at the cost of health
•	 Protective defences such as pain and fever
2.2  Evolution: Medicine’s most basic science
41
to every antibiotic. Evolutionary analysis of the phenomenon shows
that initial intuitions may not be right. For instance, rotating the
first-​choice antibiotic in a hospital every few months does little to
decrease multidrug resistance, and taking all of an antibiotic pre-
scription may not prevent resistance. Most of our antibiotics are
products of natural selection sifting through a vast range of mol-
ecules during a billion years of competition between microbes.
Pathogens also have strong selection effects on hosts, particularly
in shaping defences such as fever, vomiting, diarrhoea, cough, and
the many manifestations of inflammation. These adaptive responses
often have harmful effects because they are products of an evolu-
tionary arms race. Every defence creates selection for ways to es-
cape it, and this shapes yet more expensive and dangerous defences.
At equilibrium, we would expect the defences to become nearly as
dangerous as the pathogens (natural selection would be expected to
amplify them until they approach the danger level), a principle that
should inform studies of anti-​inflammatory agents in infection.
Constraints
Many of the body’s limitations reflect the limits on what natural
selection can do. It cannot maintain an information code without
errors, nor can it start afresh to correct a poor ‘design’. For instance,
the eye’s nerves and vessels are between the light and the retina, and
their exit causes a blind spot. Such constraints can never be fixed,
because intermediate stages do not work. Human engineers can,
literally, go back to the drawing board, evolution cannot (imagine
if the jet engine had had to ‘evolve’ from the propeller engine, step
by step).
Trade-​offs
Not only does selection result in many suboptimal ‘designs’, but it
cannot make any trait perfect. All traits involve trade-​offs. Thicker
wrist bones would break less easily, but they would inhibit free wrist
rotation. Muscles fatigue, but careless use of a new drug that blocks
fatigue may reveal just what damage fatigue prevents.
Bilirubin is, according to some medical teaching, a waste product
from haem metabolism. However, an intermediate molecule, bili-
verdin, is relatively water soluble. Why not excrete biliverdin?
Because bilirubin is an effective antioxidant.
If there are no such specific trade-​offs to be seen, economics al-
ways furnishes an ultimate trade-​off. Individuals could be built with
thickened bones that never break, but they would spend extra en-
ergy moving those big bones while individuals with thinner bones
would have more offspring because they divert the economic goods
saved (e.g. calcium and energy) elsewhere in the economy of the
body (e.g. milk) where they can do more good. Engineers know
this as the principle of ‘overdesign’, in which risks of failure are min-
imized within available budgets. But whereas engineering budgets
are arbitrary—​civilian aviation standards are more risk averse than
military, for example—​evolutionary budgets are set by the compe-
tition. Individuals whose bones are ‘too good’ will end up having
fewer children than rivals whose ‘spending policy’ accepts the in-
creased risk of breakage.
Reproduction at the expense of health
A related point explains the differences in mortality between the
sexes. A trait that increases reproduction will tend to spread, even
if it harms health. Investments in competitive ability give greater
reproductive pay-​offs for males than for females, so men have been
shaped to take more risks and to invest less in bodily repair. Data
from developed societies shows that mortality rates for men at
the age of sexual maturity are about three times higher than that
for women.
Defences
The final explanation is not really a reason for vulnerability, but
it is on the list because defences against disease are so often inad-
equately distinguished from direct manifestations of disease. Pain,
fever, nausea, and vomiting are adaptations useful in certain situ-
ations. Unfortunately, they are often expressed as ‘false alarms’ when
they are not essential. From a physician’s point of view, it seems that
selection has done a poor job. After all, much of general medicine
involves of blocking normal defence reactions such as pain, fever,
vomiting, and anxiety, and few patients expire as a result.
However, selection has not made a mistake. The costs of not ex-
pressing a response when it is needed are so huge relative to the
costs of false alarms that the optimal threshold allows for many false
alarms. This ‘smoke detector principle’ explains why blocking a de-
fence is usually safe: the doctor can judge if the response is necessary.
Nonetheless, we should expect that defences have been shaped to be
expressed when they were needed on the average, in the long run.
Utility
In the clinic
Upon hearing about new evolutionary approaches to medicine,
most journalists and many doctors ask how it can improve treat-
ment in the clinic today. This is the wrong question. There are some
direct clinical applications, such as hesitating before blocking
a defensive response such as a raised temperature or vomiting.
However, theory should not change practice directly. Instead,
evolution offers established methods such as population genetics,
new questions about why the body is vulnerable, strategies for an-
swering them, and a scientific foundation for an integrative under-
standing of the body.
Research implications
Revisions and extensions of evolutionary methods will make them
even more valuable. As extensions of the Human Genome Project
move us towards individualized genetic medicine, an evolutionary
view of genetic variations can get us beyond simply labelling some
‘defective’ and others ‘normal’. There is, after all, no normal genome.
There are just genes that construct phenotypes that result in more or
fewer offspring in a given environment.
As outlined above, an evolutionary approach also suggests a new
class of questions about the aetiology of disease. Research to answer
these questions should eventually allow a book like this to provide
an additional evolutionary section for each disease. The chapter on
gout will describe comparative data which tests the hypothesis that
uric acid’s benefits as an antioxidant in a long-​lived species justify its
raised levels, despite the pain to some individuals. The chapter on
jaundice will mention the costs, benefits, and evolution of bilirubin.
The chapter on infectious disease will describe the arms races that
shape pathogens and defences, and the costs and benefits of blocking
defensive responses. The chapter on anxiety and depression will not
42
SECTION 2  Background to medicine
treat them simply as pathological states, but as potentially useful
responses, prone to dysregulation. So far, however, the benefits of
seeking the evolutionary aetiology for every disease is only begin-
ning to be recognized.
Teaching implications
There is more to teach than can be taught, so medical educators try
to provide students with core facts, general understanding, and crit-
ical skills that allow them to learn more. Evolutionary knowledge
is invaluable not only for itself, but because it offers a framework
that can organize and relate the thousands of facts. It helps students
realize why bodies fail, and therefore what disease really is. Evolution
also offers opportunities for designing courses that provide deeper
understanding. For example, a biochemistry course could emphasize
the origins of certain pathways, and how adaptation is constrained
by the limits of natural selection. Students in physiology would learn
the evolutionary reason why the respiratory system relies on carbon
dioxide, not oxygen, to regulate respiration.
A deeper understanding of the body
Physicians are increasingly being educated as if they are technicians,
identifying problems and applying officially approved solutions.
This makes very poor use of medicine’s most valuable resource.
We select medical students carefully because we want—​or should
want—​doctors who think. Providing them with a deep evolutionary
understanding of the body will foster clear thinking. Instead of
viewing the body as a designed machine, they will see it as a product
of natural selection with traits more exquisite than in any machine,
some of which nonetheless leave us vulnerable to diseases. Doctors
who understand the body in evolutionary terms will make better de-
cisions for their patients because they will have a better sense of what
it is that they are actually doing.
Opportunities
Interested physicians will want to be in touch with The International
Society for Evolution, Medicine & Public Health (https://​isemph.
org/​). It sponsors an annual meeting, an Oxford University Press
Journal (Evolution, Medicine, & Public Health), The Evolution and
Medicine Review, and an email list and network to connect clinicians
and researchers interested in evolutionary approaches to medicine.
FURTHER READING
Evolution, Medicine, and Public Health. https://academic.oup.com/emph
Gluckman D, et al. (2011). How evolutionary principles improve the
understanding of human health and disease. Evol Appl, 4, 249–​63.
International Society for Evolution, Medicine, and Public Health.
http://isemph.org
Nesse RM (2011). Ten questions for evolutionary studies of disease
vulnerability. Evol Appl, 4, 264–​77.
Nesse RM, Stearns SC (2008). The great opportunity: evolutionary
applications to medicine and public health. Evol Appl, 1, 28–​48.
Nesse RM, Williams GC (1994). Why we get sick: the new science of
Darwinian medicine. Vintage Books, New York.
Perlman, R (2013). Evolution and medicine. Oxford University Press,
Oxford.
Stearns SC, Medzhitov R (2016). Evolutionary medicine. Sinauer
Associates, Sunderland, MA.

2.20 Human disasters 188
2.20 Human disasters 188
ESSENTIALS
Human disasters, as massive misfortunes long recorded over history,
have great importance for medicine, rightly prompting the call for
prevention, relief, and practical intervention by medical personnel.
But why do human disasters happen? A sharp distinction is some-
times drawn between natural disasters (e.g. earthquakes), and social
disasters (e.g. wars), but detailed knowledge often shows that this
contrast is not always clear: many disasters have mixed causes.
Taking the example of famines. These are popularly understood
in terms of food output decline, yet many famines have occurred
without any decline in food production. Such misunderstanding has
been responsible for the loss of millions of lives, mainly by under-
mining the role of social intervention. Starvation is a characteristic of
some people not having enough food to eat, not of there being not
enough food in the economy.
Even when nature plays a part in human disasters, society can
make a huge difference. The death toll from so-​called natural
disasters is much less in higher than lower-​income countries, even
when they do not have any fewer disastrous natural events. The type
of political rule can also be a factor of importance: authoritarian
governments and military dictatorships are generally insensitive to
vulnerable people. Interventions that can be effective include envir-
onmental policies, those that mitigate the effects of disasters on the
economy, and healthcare actions.
Introduction
‘A sudden or great misfortune’ is the way the New Shorter Oxford
English Dictionary defines a disaster. That is not a bad starting point
in trying to understand the idea of ‘human disasters’. Disaster is a
common enough term in public discussion. The addition of the ad-
jective ‘human’ to the word points specifically to the misfortunes of
human beings, rather than those of animals or of the perils of our
natural surroundings (except to the extent that these non​human
­adversities yield human mishaps). It also guides us towards focusing
particularly on the predicaments of groups of people with shared
and interlinked predicament, rather than on the personal tragedy or
sadness of particular individuals, seen on their own.
The term could be used differently of course. There is nothing in-
herently odd in talking, for example, about the human disaster that
befell King Lear, and indeed Lear’s woeful elderly life can, of course,
be of interest for a textbook of medicine. It would belong, however,
to some other section of this book linked, for example, with psycho-
logical adversity, the trauma of dejection and betrayal, or perhaps
geriatric deprivation, with their own clinical challenges.
In contrast, what are standardly called ‘human disasters’ have
come to be seen as sudden developments of huge misfortunes, in the
form of death or debilitation or displacement or impoverishment, of
substantial groups of people simultaneously affected by such events
as earthquakes, floods, epidemics, wars, or famines, which call for
imperative social action for prevention and relief. That is the sense in
which the idea of human disasters will be addressed in this chapter.
It is, however, important to mention that the disastrous effects of
terrorism and targeted killing are increasingly becoming another
source of what can be called human disaster. Such specialized vio-
lence in the contemporary world demands attention as well, and they
take us in the direction of examining the influence of exploitation—​
and sometimes deliberate ‘planting’—​of a sharp and confrontational
human identity in a highly belligerent form, used for exciting violent
feelings towards other groups with different identities—​other reli-
gions, other nationalities, other races, or other ethnicities. Studies
are needed not only about the way identity politics can be made to
serve the cause of human violence, but also about how organized in-
flammation can become a source of danger for human security and
an ingredient of a form of human disaster.
Problems of a natural-​social dichotomy
In the vast literature on the subject of human disasters, a sharp
distinction is sometimes made between natural disasters, like
earthquakes, floods or droughts, and social disasters, like wars or
genocide. There is some rudimentary logic in that distinction: polit-
ical sagacity can obviously have more success in averting genocide or
wars than in, say, preventing earthquakes. Yet the contrast is not so
2.20
Human disasters
Amartya Sen
2.20  Human disasters
189
total, since in the transformation of a natural calamity into a human
disaster, social factors play a big role.
For example, though floods are natural phenomena in an obvious
sense, their effects—​and indeed even their incidence—​are also in-
fluenced by social and economic events, such as the making of canals
or drainage systems, not to mention the climatic effects of global
warming that may be significantly influenced by preventable human
activities. Even though in understanding the immediate parameters
of such events as floods or droughts or storms, we can begin well
enough by ‘rounding up the usual suspects’ in the natural world
(such as rainfall, temperatures, gales), it is important that our ana-
lysis should not end there.
Some other examples of human disaster are quintessentially
mixed bags, such as epidemics, which depend on the working of
nature (e.g. on the properties of viruses, bacteria, and other conta-
gions), and yet are massively influenced by human behaviour (like
contacts and exposures) and social arrangements (like immuniza-
tion and medical care). In such cases it would be very difficult even
to begin the analysis of disasters of this mixed type—​epidemics and
others—​only as phenomena of nature, since the social-​natural en-
tanglement is rampant in every aspect of this type of human disaster.
In some cases, the inescapably mixed causation of disaster has
tended to escape attention because of the popularity of apparently
‘obvious’ causal explanations that fail to look far enough and are
easily satisfied with finding a natural correlate. A good example of
this kind of neglect can be seen in the popular understanding of the
causation of famines in terms of food output decline, with much-​
repeated explanations that do not go beyond the proximate fea-
tures of droughts, floods, or storms. This causal confusion, which is
contradicted by the great many cases of famines that have occurred
without any decline in food production, has been responsible for the
loss of millions of lives, mainly through undermining the necessity
and urgency of social intervention.
The subject demands some discussion here, and to that I turn
next. Indeed, by concentrating on famines, which have been much
more extensively studied than other kinds of disasters, it is possible
to illustrate some general points about the relationship between
social and natural aspects of human disasters.
Famines: Causes and prevention
Famines are gigantic events of carnage in which millions of people
die from starvation and from diseases linked with human debili-
tation, social disruption, and movements of destitutes in search of
something to eat, and the spread of communicable ailments asso-
ciated with these phenomena. Even though a great many people die
from a famine, the misfortune that a population experiences in a
famine can go well beyond the mortality of those who succumb to
the famine, since famines can leave an inheritance of huge disrup-
tion for many years to come.
The crude logic that if people are dying of starvation, there must be
a shortage of food, has had much influence on traditional thinking
across the world. That theory has had a particularly damaging effect
in persuading governments to do nothing even when many people
lose their livelihood (for one reason or another), which makes them
unable to buy food. This hypothesis was relevant to the British Raj’s
strange inactivity as the Bengal famine of 1943 slowly gathered
momentum. It would eventually kill close to 3 million people. The of-
ficial statistics of food output in Bengal indicated that there had been
no significant change in the availability of food there, and sticking to
their theory, the government did little to relieve the relentless emer-
gence of a large famine. Instead, it tried to deny the existence of the
unfolding famine, sustained by the censorship of the ‘native’ press,
and the complicit silence of the British-​owned English papers.
That gigantic cover-​up ended when the leading English paper
of India, The Statesman (also British owned), broke rank with the
decision of the highly agonized editor (Ian Stephens) that he could
no longer be a party to the concealment of a huge human disaster.
Stephens supplemented his factual reports of starvation by stinging
editorials on official policy. Once the news went into the public do-
main it got large attention in the British press and in Parliament in
London, making the government admit the existence of a famine,
which also led to the starting of much-​delayed relief work. However,
ever loyal to the ‘theory’ that famines are caused only by food avail-
ability decline, the same imperial government would later revise the
statistics downward to bring the revised ‘facts’ into line with their
deluded theory, even though the previous food statistics had been
broadly right.
What went wrong, of course, was the inability to understand that
starvation is a characteristic of some people not having enough food
to eat: it is not a characteristic of there being not enough food in the
economy. The critical connection of the command over food—​what
is sometimes called the ‘entitlement’ to food—​is with the purchasing
power to buy food (in a market economy), linked to jobs and in-
comes and relative prices of what people have to sell (their labour
and the commodities they can make, such as services and crafts) to
buy staple food. A famine is, in this sense, an economic phenom-
enon, not just one of agricultural production.
However, going further, famines are not only manifestations of
an economic phenomenon, they also have political aspects and
medical correlates as well. Famines are easy to prevent since only
a relatively tiny proportion of the population is affected, and it is
easy to regenerate the purchasing power of those who are destitute
through emergency employment. This can provide incomes in a re-
gion in which many people have lost their purchasing power be-
cause of job loss or price rise. Indeed, the fact that independent
India, despite considerable regular undernutrition at a chronic
level, has not had a famine since the end of the Raj in 1947, is only
partly linked with the progress of food production and agriculture
in post-​independence years; it is also critically connected with the
state’s immediate willingness to start relief programmes, mainly in
the form of emergency projects, giving jobs and incomes to the af-
flicted population.
The urgency of preventive intervention is a political imperative
for the government in office in a functioning democracy. If a famine
is allowed to develop then no ruling party (or coalition) has much
chance of winning a democratic election (or even to survive blis-
tering criticism in the media and in parliaments or assemblies).
Indeed, famines in the recent years, as in the past, have continued
to occur only in countries without a functioning democracy (e.g.
military dictatorships, one-​party states, or countries that do have
elections but do not have other necessary features of a democracy
such as a free media and room for public discussion and social criti-
cism). Famines are thus manifestations of political as well as eco-
nomic phenomena.
190
section 2  Background to medicine
Famines also belong to the domain of medicine and public
healthcare. This is because most people who die from famines die
from diseases of the region, some of which are locally endemic.
While starvation is the prime mover of famine mortality, actual
deaths very often come from people getting ill and succumbing to
their illness. Debilitation due to severe undernourishment plays a
part in this in making people vulnerable to fall ill and sometimes
to die of it. But no less importantly, hunger precipitates illness and
death because of (1) the propensity of hungry people to eat what-
ever scraps they can pick up from anywhere; (2) vast population
movements induced by search for jobs and food that spread con-
tagions with great speed; and (3) breakdown of essential services
including healthcare and medical attention. Famines typically kill
not just directly as a result of starvation but through intensifying
and aggravating the forces of illness-​based mortality common to
the region.
Table 2.20.1 gives a breakdown of the causes of death in the ex-
cess mortality in the Bengal famine of 1943, as presented in my 1981
book Poverty and Famines, for the period 1943–​1946, taking the
prefamine 1941–​1942 average as the standard for comparison.
More than four-​fifths of the death toll resulting from the Bengal
famine, in this estimate, was directly connected with diseases
common to the region, with pure starvation death accounting for no
more than only a fifth of the total. A similar picture emerges from
most other famines.
This aspect of famines is the centre of attention in Alex de Waal’s
important study of famines in Darfur in Sudan. This is certainly a
rich and policy relevant perspective on famines, particularly be-
cause even with the failure of food entitlements, the magnitude
of deaths can be very substantially reduced through health inter-
ventions. If the emergence and survival of a famine is largely an
economic and political issue, the mortality it generates is also
very significantly a matter of healthcare and medical attention
(including timely prevention, particularly through immuniza-
tion). The allegedly ‘natural’ phenomenon of famines tends to be
intensely ‘social’ at many different levels—​economic, political, and
medical.
Before I move on from the subject of famines to other types of
human disasters, let me briefly note the variety of circumstances
in which famines have occurred and have killed with abandon
across the world. It would be good to associate firm mortality es-
timates with each famine, but this would be a difficult thing to
do in many cases. Even though death tolls have been thoroughly
studied for some famines, using reasonably good statistics, there is
always an element of uncertainty in any such estimate, no matter
how carefully it may have been prepared. Part of the difficulty lies
in the underlying conceptual issues. The death toll of a famine has
to be calculated by contrasting the actual number of deaths in that
period with the number of deaths that would have taken place in the
absence of that famine. That is, famine mortality is the difference
between actual mortality in the famine period and the estimated
number of people who would have died anyway, even if the famine
had not occurred at all. Thus the uncertainties involved relate only
partly to problems in getting good statistics of actual deaths, which
is not an easy task as it is, especially at a time when many of the
normal functioning of social institutions, including registration of
deaths, are disrupted (sometimes because of death or migration of
the staff involved in the relevant social institutions). They also relate
to the uncertainties inherent in the ways and means of estimating—​
as a ‘counterfactual’—​what would have happened had the famine
not occurred at all.
One can nevertheless get some general impression of the size
and reach of a famine by looking at the estimates of death tolls that
we have, along with reports of other events and predicaments con-
nected with the respective famines. Table 2.20.2 relates to many
‘recent’ famines that have occurred over the last couple of hundred
years, beginning with the Irish famines of 1845–​1851 and drawing
on a large number of sources.
The association of famines with political authoritarianism and
civil disruption is well illustrated by the incidence of such events
in countries with military governments (Ethiopia, Sahel countries),
or one-​party rules (Soviet Union, China, Cambodia, North Korea),
or alien governance (preindependence India and politically dom-
inated Ireland), or civil disruption (Nigeria, and to some extent
Bangladesh, even though the death toll was kept in check, particu-
larly in the latter, through organized famine relief).
Disasters and policy intervention
Human disasters can be of many types, varying from floods,
droughts, wind storms, extreme temperatures, and earthquakes,
where natural factors have some clear role, to wars and indus-
trial accidents (like the Union Carbide disaster in Bhopal in
1984) where the story is almost entirely social in the broad sense.
But the important point in the context of policy intervention is to
recognize that, even when nature plays a part, society can make a
huge difference.
Consider natural calamities such as floods or droughts, which
have been associated with processes that have led to great many
disaster deaths. What can intervention achieve? There are at least
three different ways of prevention or amelioration.
•	First, environmental policies can make floods, droughts, and
other natural adversities less common. This is a subject that is
being energetically discussed today—​at long last—​moved largely
by the evidence of massive global warming and the related long-​
run environmental hazards. The recent ‘COP21 summit’ in Paris
in December 2015 is a good example of attempting disaster pre-
vention through concerted international efforts. There was,
Table 2.20.1  Proportionate breakdown of excess mortality in the
Bengal famine: 1943–​1946 over 1941–​1942
Ailment types
Percentage
contribution
Malaria
36.7
Cholera
7.1
Smallpox
5.0
Dysentery, diarrhoea, and enteric group of fevers
5.0
Other fevers (often undiagnosed)
27.4
Respiratory diseases
0.4
Percentage share of ailment-​related deaths
81.6
Source: Sen A (1981). Poverty and Famines: An Essay on Entitlement and Deprivation.
Oxford University Press, Table D.3, p. 204.
2.20  Human disasters
191
contrary to earlier fears, some limited success, and yet much was
left unresolved in Paris—​what George Monbiot described in The
Guardian as ‘by comparison to what it could have been, it’s a mir-
acle’, but ‘by comparison to what it should have been, it’s a disaster’
(12 December 2015). There will no doubt be many efforts in future
years to avert the climatic catastrophe-​in-​making.
•	Second, any human disaster that kills directly will also tend to
disrupt the economy through its effects on jobs and incomes and
prices, thereby affecting the entitlement to food and other es-
sentials that people need. The discussion on famines above has
already illustrated this general connection. That analysis can be
extended. For example, an earthquake may not lead to a famine,
but can leave an economy unsettled, making lives precarious for
those who are not killed in the physical calamity itself. These ef-
fects can be prevented or at least reduced in their impact through
careful social intervention. The type of housing and the arrange-
ments of city planning can also reduce the effects of earthquakes,
floods, storms, and other hazards.
Third, the morbidity and mortality associated with a human
disaster can be large and even extend far beyond the immediately
affected population who are directly hit by the physical events.
However, taking note of the likely dangers from local conta-
gions, the scale of both illness and death can be radically altered
through a variety of preventive measures, such as immunization,
ensuring safe water, influencing the routes of contact and the
spread of infection, and also through health education and med-
ical preparedness.
The divisiveness of disasters
The death toll from so-​called natural disasters reduces sharply as
we move to higher and higher income countries, even when they
do not have any fewer natural events of that type. The contrast can
be very sharp indeed, as Kofi Annan, the former Secretary General
of the United Nations, has noted (International Herald Tribune, 19
September 1999):
‘Ninety per cent of the disaster victims worldwide live in developing
countries where poverty and population pressures force growing
numbers of poor people to live in harm’s way on flood plains, in
earthquake prone zones and on unstable hillsides. Unsafe buildings
compound the risks. The vulnerability of those living in risk prone
areas is perhaps the single most important cause of disaster casualty
and damage.’
A similar remark can be made about the role of medical services and
epidemiology, which are often rudimentary in the poorer countries.
If low income is a predisposing condition for the penalty of
disasters, the type of political rule can also be a factor of importance.
The impact of the insensitivity of ruling governments in authori-
tarian countries on the interests of vulnerable people was illustrated
earlier in the specific context of famine prevention, but the problem
of authoritarian insensitivity is a widespread phenomenon and goes
well beyond the terrible record of authoritarian rules with famines.
The neglect of medical services and healthcare that can be seen in
many military dictatorships, for example in sub-​Saharan Africa, is
Table 2.20.2  Some recent famines
1845–​1951, Ireland
A series of famines linked with a blight in potato farming, with loss of jobs, incomes, and staple food; famine mortality around one
million deaths (the largest proportionate mortality among all ‘recent’ famines).
1928–​1929, China
Famines, particularly affecting farming population, especially in Shensi, Honan, and Kansu provinces, with mortality estimates that
are in excess of 5 million deaths.
1932–​1934, Soviet Union
Connected with agricultural turmoil, at least partly linked with collectivization, especially severe in Ukraine, with mortality
estimates in excess of 5 million deaths.
1943, India
In the province of Bengal in British India, with a fairly normal food availability; mortality numbers are now placed between 2 and
3 million.
1958–​1961, China
Linked with the disastrous failure of the so-​called Great Leap Forward, there was a sharp decline in food production, disruption
of normal economic processes, and huge chaos; the famine was well hidden from those not directly exposed to it. The death toll
has been later estimated to be at least around 30 million (making it the largest absolute size of famine mortality in the recorded
history of famines in the world).
1967–​1968, Nigeria
Connected with civil war and the blockade of Biafra; no reliable estimate of deaths has been made.
1972–​1974, Sahel region
Affecting several African countries in the Sahel belt (Burkina Faso, Chad, Mali, Mauritania, Niger, Senegal), with some reduction in
agricultural production, but also affecting pastoralists both through animal death and through a fall in the relative prices of animal
products against staple food from agriculture; no reliable mortality estimates can be made.
1973, Ethiopia
The drought in the Wollo province reduced the purchasing power of the poor, and even though there was no significant
reduction in food output at all for Ethiopia as a whole (the reduction was confined mainly to Wollo), food did not move into
Wollo, and some moved out of it, because of the relatively larger purchasing power of the rest of Ethiopia; no reliable mortality
estimates can be made.
1974, Bangladesh
This famine, which followed shortly after the disruption of civil strife that led to the break-​up of Pakistan, occurred in a year of
peak food availability. Floods that would reduce food output the following year immediately affected employment and incomes
of rural labourers, and also an exaggerated anticipation of a coming food crisis made food prices rises very steeply (followed
by a price decline later on, after the famine). Mortality estimates vary but it was kept very much in check by a huge food relief
programme arranged by the government through which 4.35 million people (about 6 per cent of the total population) were fed.
1975–​1979, Cambodia
Linked with the Khmer Rouge ravages of the rural economy, deportation of parts of the urban population, and systematic
slaughter of unfavoured people; the death toll is estimated to be around one million.
Since early
1990s, North Korea
Intermittent famines with economic disruption by an authoritarian regime with a strong ideological agenda; death estimates are
often placed around one million—​so far.
192
section 2  Background to medicine
itself a problem for disaster management. There are political param-
eters related to the medical, social, and economic interventions
that can make disasters less likely, and less devastating when they
do occur. The effects of similar natural phenomena can be widely
different, depending on income levels, political reactions, medical
arrangements, and other variable conditions.
Human disasters are very divisive events, not only between coun-
tries but also within them. Even in the same country, and sometimes
even within the same region of a country, a disaster can ruin some
groups of people while leaving others almost completely untouched.
Famines, for example, tend to affect those who are already vulner-
able because of poverty and the absence of safety nets on which they
can rely, private or social. Policy intervention has to take note of the
divisions along the lines of class and occupation groups.
There are also differences related to age and gender. Children may
be particularly vulnerable in some disasters, including in epidemics
of some diseases. The need for immunization of children and arran-
ging medical attention for children are important requirements in
disaster management, and so is the institution of economic safety
nets through state provision and other social arrangements for child
support and income supplement at moments of dire need.
In some disasters (for example in the present phase of AIDS epi-
demic), women are increasingly more affected. It is usually the case
that the opposite is true in wars, which typically kill more men than
women or children, although women may be particularly vulner-
able when there is a civil war targeted against a specific community
where women are systematically raped and injured, as has happened
for example in Darfur in Sudan. Even though the mortality rate may
be higher among men, leaving a lower ratio of men in the surviving
population, the huge incidence of rapes amounts to a gigantic vul-
nerability for women in particular. There are other kinds of special
vulnerabilities in disasters, like breakdown of families in famines,
from which women tend especially to suffer.
There is also a more subtle point that tends to be neglected in
journalistic reports. The life expectancy of women is typically longer
than that of men, unless this is reversed through gender bias in social
care. There is some evidence that the life expectancy of women is
comparatively more reduced by many types of so-​called natural
disasters than that of men. This greater vulnerability of women is
sometimes hidden by the fact that even after the larger reduction,
women may still have a higher life expectancy than men, but—​as was
discussed in the context of famine analysis—​the right comparison is
between what actually happens with a disaster and what could have
been expected to have happened in its absence. That distinction is
as relevant for medical practitioners and public health experts as it
is for social and economic investigators, as is the recognition that
human disasters are typically highly divisive events.
FURTHER READING
Bose S (2009). Pondering poverty, fighting famines: towards a new
history of economic ideas. In: Basu K, Kanbur R (eds) Arguments
for a better world, Vol. II. Oxford University Press, Oxford.
Depoortere E, et al. (2004). Violence and mortality in West Darfur,
Sudan (2003–​2004): epidemiological evidence from four surveys.
Lancet, 364, 1315–​20.
Dreze J, Sen A (1989). Hunger and public action. Oxford University
Press, Oxford.
Guha-​Sapir D, Hargitt D, Hoyois P (2004). Thirty years of natural
disasters 1974–​2003:  the numbers. Centre for Research on the
Epidemiology of Disasters, Presses Universitaires de Louvain,
Louvain-​la-​Neuve, Belgium.
Kahn ME (2005). The death toll from natural disasters: the role of
income, geography, and institutions. Review of Economics and
Statistics, 87, 271–​84.
Malik K (2014). Multiculturalism and its discontents: rethinking diver-
sity after 9/​11. Seagull Books, Kolkata, India.
Neumayer E, Plumper T (2007). The gendered nature of natural
disasters: the impact of catastrophe events on the gender gap in life
expectancy, 1981–​2002. Ann Assoc Am Geogr, 87, 551–​66.
Ravallion M (1987). Markets and famines. Oxford University Press,
Oxford.
Sen A (1981). Poverty and famine: an essay on entitlement and depriva-
tion. Oxford University Press, Oxford.
Sen A (2006). Identity and violence:  the illusion of destiny. Norton,
New York and Penguin, London.
Stephens I (1966). Monsoon morning. Ernest Benn, London.
Stern N (2007). The economics of climate change. Cambridge University
Press, Cambridge.
Vaughan M (1987, 2007). The story of an African famine. Cambridge
University Press, Cambridge.
de Waal A (2005). Famine that kills: Darfur, Sudan. Oxford University
Press, Oxford.
World Disaster Report 2005 (2006). International Federation of Red
Cross and Red Crescent Societies, Geneva, Switzerland.

2.21 Humanitarian medicine 193
2.21 Humanitarian medicine 193
ESSENTIALS
Humanitarian medicine addresses the human consequence
of crises such as conflict, disaster, or displacement, and serves
to assist those whose lives and health are impacted by such
events. It is practised in challenging settings where resources
are limited and environments unstable, and requires a clinical
skillset which is both near limitlessly broad and context specific.
The humanitarian sector has expanded significantly during the last
two decades as a result of climate-​related crises and increasing
complex humanitarian emergencies due to protracted and multi-
faceted conflicts. The picture today is evolving rapidly to ad-
dress expanding humanitarian demands and a changing global
sociopolitical reality.
Introduction
Humanitarian medicine is a discipline in which the overarching
aim is to provide relief to those whose lives and health are harmed
by crises. This work strives to uphold the core humanitarian prin-
ciples of humanity, neutrality, impartiality, and independence,
even when the reality on the ground is dangerous and chaotic.
Unlike most clinical disciplines, humanitarian medicine is not
defined by technique but rather by the setting in which it oc-
curs:  complicated environments, such as conflict, disaster, or
displacement, that are inherently resource-​limited and insecure.
Within this context, however, its scope is broader than clinical
medicine and requires a coordinated effort between multiple dis-
ciplines. Moreover, the practice of humanitarian medicine in-
volves adherence to principles which go beyond the precepts of
the Hippocratic Oath.
Humanitarian Medicine has been proposed as a practice that
‘goes beyond the usual therapeutic act and promotes, provides,
teaches, supports, and delivers people’s health as a human right, in
conformity with the ethics of Hippocratic teaching, the principles of
the World Health Organization, the Charter of the United Nations, the
Universal Declaration of Human Rights, the Red Cross Conventions
and other covenants and practices that ensure the most humane and
best possible level of care, without any discrimination or considera-
tion of material gain’.1
Humanitarian providers practise medicine in one of its truest
forms. Humanitarian medicine occurs in settings with low re-
sources, where providers need to rely on clinical skills rather than
extensive ancillary tests and imaging. It would be hard to find
a practice of clinical medicine that more closely demonstrates
Hippocrates’ assessment that ‘the art (of medicine) consists of three
elements: the disease, the patient and the doctor’. At a period when
Western-​trained physicians are becoming increasingly specialized,
humanitarian providers often need to diagnose and treat patients
that cross all specialties, including those with rare or atypical pres-
entations, and they often conduct their practice without the safety
net of referral.
Although not a new discipline, the sphere of humanitarian medi-
cine has expanded over the last 20 years, primarily in response to an
increase in the scale and frequency of humanitarian emergencies.
Climate change and natural disasters have become more frequent,
violent conflicts are less defined and more protracted, and forced
displacement has reached critical levels. With this reality comes the
human consequence and the humanitarian needs of those affected.
The United Nations Office for the Coordination of Humanitarian
Affairs (OCHA) estimates that at least 141  million people in 37
countries were in need humanitarian assistance in 2017—​the highest
level in history (Fig. 2.21.1). The totality of need is staggering, and
humanitarian medicine is a practice needed now more than ever.
Origins and evolution of the humanitarian sector
History of humanitarian response
Humanitarian medicine is increasing in prominence (as demon-
strated by its new inclusion in this established textbook!). This is
related to many factors including increasing global humanitarian
2.21
Humanitarian medicine
Amy S. Kravitz
1  Masellis M, Gunn S (2017). International Association for Humanitarian
Medicine (IAHM) Brock Chisholm. Iahm.org. http://​www.iahm.org/​eng/​home_​
medicinaumanitaria.htm
194
section 2  Background to medicine
Democratic Republic
of the Congo (DRC)
Central African
Republic (CAR)
Chad
Libya
Cameroon
Nigeria
Burkina Faso
Senegal
Mauritania
Haiti
Mali
Niger
Nigeria regional refugee
response plan
Burundi regional refugee
response plan
DRC
Uganda
Chad
Cameroon
Niger
Rwanda
Burundi
Nigeria
Tanzania
Sudan
Myanmar
Afghanistan
Occupied Palestinian territory (oPt)
Syrian Arab Republic
Iraq
South Sudan
Burundi
Somalia
Djibouti
Yemen
Ethiopia
Ukraine
Syria regional refugee
and resilience plan
South Sudan regional
refugee response plan
Sudan
Ethiopia
Egypt
Turkey
Iraq
Lebanon
Jordan
Kenya
Uganda
South Sudan
Syrian Arab Republic
Countries with humanitarian
response plans (HRPS) or
other appeals
Countries included in regional
refugee response plans
Fig. 2.21.1  Response plans around the world; Global Humanitarian Review 2017.
Reproduced from Global Humanitarian Review 2017, United Nations Office for the Coordination of Humanitarian Affairs (OCHA).
2.21  Humanitarian medicine
195
needs, the expanding size of the overall humanitarian sector, and
the role that the mass media have played in raising awareness of
global crises and increasing the profile of those who are engaged in
the emergent responses. However, the humanitarian concept is far
from a new phenomenon: as the Greek historian, Thucydides, wrote
in his classical account of the war between Sparta and Athens in the
5th century BCE:
‘If it had not been for the pernicious power of envy, men would not so
have exalted vengeance above innocence and profit above justice . . . in
these acts of revenge on others, men take it upon themselves to begin
the process of repealing those general laws of humanity which are there
to give a hope of salvation to all who are in distress.’
(Thucydides, History of the Peloponnesian War)
This understanding of human nature and behaviour in crises nearly
two and a half millennia ago is one of the earliest discourses on hu-
manitarian action. Humanitarian principles have been enshrined
in many religious teachings, and limits to warfare, including the
acceptable conduct of wars, have been articulated in texts from
the time of Thucydides: examples include The Art of War by Sun
Tzu. But despite its underlying presence, for millennia the conduct
of humanitarian medicine lacked codification and the practical
frameworks that it now enjoys.
The origins of contemporary humanitarianism are traditionally
considered to be rooted in the actions of Henri Dunant, a Geneva-​
born businessman, who in June 1859 travelled to northern Italy to
see Emperor Napoleon III, arriving in time to witness the imme-
diate aftermath of one of the bloodiest battles of the century, the
Battle of Solferino in the Second Italian War of Independence. With
over 5000 soldiers killed and estimates of over 23 000 left wounded
on the battlefield, Dunant’s experiences led to the publication of a
small book entitled Un Souvenir de Solférino (A Memory of Solferino)
in 1862. This book focused attention on the value of having a pre-
pared cadre of responders to provide support in emergencies and
proposed a plan for the formation of government sponsored inter-
national relief societies to care for those wounded in wartime. He
put forth the idea that aid workers should be permitted to assist the
wounded on the battlefield, unencumbered and unharmed, pro-
vided they remained neutral to the conflict.
In 1863, Dunant helped to establish The International Committee
for Relief to the Wounded, providing the framework for the even-
tual establishment of the Red Cross. The following year, on 22
August 1864, 12 nations agreed to the principles of protection for
the wounded and the neutrality and independence of medical per-
sonnel on the battlefield. By signing the Geneva Convention for the
Amelioration of the Condition of the Wounded in Armies in the Field,
now commonly known as the Geneva Conventions, the concepts
of rules in war and protection for humanitarian assistance were en-
shrined in law. For his efforts, Dunant was awarded the first ever
Nobel Peace prize in 1901.
From this origin, formal international humanitarian law has been
designed to specifically cover the duties of those engaged in armed
conflicts as well as the rights of humanitarian organizations involved
in relief operations. The laws serve as a careful equipoise between
respect for state sovereignty with the recognized rights of inde-
pendent organizations to provide aid to those in need. In total,
the four Geneva Conventions of 1949, the two 1977 Additional
Protocols, and a final additional Protocol in 2005, form the funda-
mental basis for international humanitarian law.
Derived directly from these laws are the fundamental principles
of humanitarian action: humanity, neutrality, impartiality, and inde-
pendence. The OCHA summarizes these key principles as follows:
•	Humanity:  Human suffering must be addressed wherever it is
found. The purpose of humanitarian action is to protect life and
health and ensure respect for human beings.
•	Neutrality: Humanitarian actors must not take sides in hostilities
or engage in controversies of a political, racial, religious, or ideo-
logical nature.
•	Impartiality:  Humanitarian action must be carried out on the
basis of need alone, giving priority to the most urgent cases of dis-
tress and making no distinctions on the basis of nationality, race,
gender, religious belief, class, or political opinions.
•	Independence:  Humanitarian action must be autonomous from
the political, economic, military, or other objectives that any actor
may hold with regard to areas where humanitarian action is being
implemented.
International humanitarian law, combined with medical ethics and
other legal frameworks for healthcare providers, was intended to de-
fend the autonomy of medical personnel, to protect them against
interference by security or military personnel, and to ensure that
healthcare is provided on the basis of need, and neither refused nor
used against groups.
Aid architecture
The formation of the Red Cross, and the principles upon which
it was based, set the standard for the development of subsequent
non​governmental organizations (NGOs). The League of Nations
High Commission for Refugees, under the newly formed League
of Nations, was founded in 1921 and served as a forerunner of the
United Nations High Commissioner for Refugees (UNHCR). This
was the first international agency that focused upon legal and polit-
ical protections and assistance to refugees. Other humanitarian or-
ganizations followed, and by the 1980s the key emergency offices
within the United Nations had been formed, followed by the prolif-
eration of NGOs.
Today, humanitarian medicine is practised within a framework
of a larger aid system, with an UN-​led coordination mechanism
established in 1991 through General Assembly Resolution. Key
pillars of the UN-​led humanitarian system included the forma-
tion of the Inter-​Agency Standing Committee (IASC) as the main
policy and decision-​making body guiding international humani-
tarian response, and the establishment of the OCHA, intended
to increase coordination among the multitude of actors now en-
gaged in humanitarian relief. No individual organization has suf-
ficient capacity to address the comprehensive needs within even a
single humanitarian crises, let alone the multitude simultaneously
ongoing, and efforts to improve coordination between agencies,
as well as with the host authorities, and the affected communities
remains an ongoing priority. As part of that effort, the UN intro-
duced the cluster system in 2005 to increase accountability, or-
ganization, and professionalism within the global humanitarian
196
section 2  Background to medicine
community. With health serving as one of the 11 established clus-
ters, the purpose of this system was to ensure ‘that international
responses to humanitarian emergencies are predictable and account-
able and have clear leadership by making clearer the division of la-
bour between organizations, and their roles and responsibilities in
different areas’. (Fig. 2.21.2).
The practice of humanitarian medicine
Precipitants and preparations
All crises, including violent conflict and natural disaster, can af-
fect the health of a population, ranging from injuries and trauma
to the exacerbation of underlying medical conditions. These events
can weaken already poor infrastructure and affect access to basic
services such as healthcare, clean water, and sanitation. Population
displacement can result, often precipitating the spread of infectious
diseases that are exacerbated by overcrowding. The diseases with
the highest epidemic potential in such settings—​cholera, dysen-
tery, measles, and meningitis—​are uncommonly seen in most clin-
ical practices today, but where there is displacement they can have
devastating individual and population level effects.
Humanitarian assistance is essential in any crisis situations
where the host government is unable to meet the demands within
its own country or is unwilling to provide for the needs of its popu-
lation. Before the establishment of a humanitarian space, how-
ever, even in emergent situations, assessments, and planning must
occur. Most experienced humanitarian organizations have teams
of medical, logistics, security, and epidemiological experts on
standby to deploy to any affected area to immediately carry out
rapid assessments. The inclusion of local partners who can trans-
late and provide an understanding of local context is essential, and
mapping platforms such as geographic information systems can
serve as tools which allow a better understanding of the locations,
relationships, and patterns.
Logistical considerations can be complex, particularly in un-
stable environments, and often require the development of a supply
chain with more agility than that required for typical commercial
endeavours. Most agencies initially operate by a ‘push’ system,
using prepacked and stockpiled kits ready for rapid deployment
to an emergency, and switch to a ‘pull’ system, when supplies
are requested as per need, within about 2–​3 weeks of operations.
Humanitarian doctors who join these relief efforts, either in the ini-
tial stages or once the programme has become established, should
Humanitarian
& Emergency
Relief
Coordinator
Early
recovery
UNDP
Education
UNICEF &
Save the
Children
Emergency
telecom-
munications
WFP
Health
WHO
Food security
WFP & FAO
Logistics
WFP
Nutrition
UNICEF
Camp
coordination and
camp management
IOM/UNHCR
Shelter
IFRC/
UNHCR
Protection
UNHCR
Water,
sanitation,
and hygiene
UNICEF
Pre
ve
nt
io
n
Mit
ig
at
io
n
Pre
pa
re
d
n
e
s
s
D
is
as
ter
Res
po
ns
e
Reco
ve
ry
Re
co
ns
tr
uc
tio
n
Fig. 2.21.2  UN cluster system.
Reproduced from Reference Module for Cluster Coordination at Country Level (revised July 2015), Inter-​Agency Standing Committee.
2.21  Humanitarian medicine
197
gather information about the context and strive to form a basic
framework of the background clinical and epidemiological picture.
The setting they will enter may be chaotic and tense, and prepar-
ations will serve to align expectations and streamline focus onto the
clinical challenges which lie ahead.
Complex realities
Simply put, providing clinical care in humanitarian settings is
a challenge; resources are limited, advanced diagnostics are not
readily available, and the environment is unfamiliar and chaotic.
Colleagues may have different backgrounds, cultural norms, and
professional practices. Language barriers often require the use of
translators. Despite these realities, the breadth of possible caseloads
is nearly limitless. A humanitarian provider may need to manage
acute emergencies, respond to victims of sexual violence, treat non-​
communicable diseases, or support war surgery. A caseload could
include ophthalmological, renal, and orthopaedic presentations,
while concurrently working to ensure the provision of clean water
and support for hospital management. Without extensive adminis-
trative or axillary staff, humanitarian providers may have a broad
range of duties that stretch their capacity and test their skills to
the limit.
Humanitarian medicine must be practised with a nuanced under-
standing of the specific context in which it occurs; mastery in one
setting does not imply universal expertise in another. The skills re-
quired for effective working (for example) in the Central African
Republic, where there is context of protracted conflict, include an
understanding of malnutrition, epidemic control, and of sexual vio-
lence. In contrast, providing assistance to Iraqi refugees involves
supporting emergency health needs and a shifting focus from war
injuries to long-​term rehabilitation. In any setting these clinical
scenarios would pose a challenge, but working in unstable environ-
ments adds further complexity, since before even starting it may be
necessary to negotiate humanitarian space with armed combatants,
coordinate across agencies, and gain the trust of a broad community.
Rather than a ‘simple’ mastery of clinical medicine, effective practice
in humanitarian settings demands of the provider that they under-
stand disciplines such as security, law, epidemiology, and logistics,
and can operate practically in highly politicized domains
Despite the innumerable complexities that present themselves,
there is as yet no overarching licensing body or specific clinical
review board to instruct and regulate this emerging discipline.
There have been coordinated efforts to standardize best practices
in humanitarian aid, such as with the Sphere Standards, initiated
in 1997, and the Core Humanitarian Standards on Quality and
Accountability (CHS), launched in 2014 (Box 2.21.1), but there
are clear limitations to the use of universal standards in diversely
varied settings. The lack of a definitive training programme to pre-
pare the physician for the enormous variability of contexts and
scenarios means that humanitarian providers must depend upon
their medical education, clinical experiences, and preparations, and
be willing to listen and learn from others. National or ‘local’ col-
leagues are often the source of indispensable information, and the
best humanitarian doctors are those who have the capacity to ac-
knowledge their own shortcomings and learn from others. Having
‘gold standards’ might remain the goal, but the reality of circum-
stances ensures that these remain at best highly challenging or, in
effect, impossible.
The humanitarian system
Size and economics
In response to a changing global landscape there has been an unpre-
cedented expansion in the humanitarian sector over the last decade.
Between 2007 and 2010, the fieldworker population rose by 4–​6%
annually, with financial contributions to the international emer-
gency response efforts rising nearly threefold. By 2010, there were
an estimated 274 000 humanitarian providers who collectively re-
sponded to 103 natural disasters and 43 complex emergencies. By
2014, this number had grown to 450 000 humanitarian providers,
working across more than 4480 operational aid organizations.
Box 2.21.1  Humanitarian response review and
sector strengthening
Throughout the history of humanitarian action, nearly every major
international intervention had raised questions about how to measure
the impact and effectiveness of the response and how to make the
sector more coordinated, efficient, and effective. When the contextual
variability of humanitarian interventions can include sociopolitical con-
straints and security limitations, developing a uniform approach to
improvement can be difficult to achieve. Despite the challenges, over
the past two decades, great strides have been made to further the ef-
fectiveness of humanitarian action, and the evaluation of humanitarian
responses, such as that to the 2010 earthquake in Haiti, remain an im-
portant component driving this effort.
After the devastating 2010 earthquake in Haiti, which killed more than
220 000 people and affected nearly a third of the Haitian population,
almost $9 billion in aid was donated in what amounted to one of the
largest emergency relief responses ever mounted. Countless lives were
saved in the immediate search-​and-​rescue efforts and medical response,
and in subsequent relief efforts which reduced food insecurity and con-
trolled an ensuing cholera epidemic. Despite such a large financial com-
mitment and global response, by 2013, acute vulnerabilities persisted
with over 10% of the population remaining displaced in camps and 30%
continuing to have chronic and acute health needs.
A review of the international response to the earthquake showed that,
significantly, the Government of Haiti, was estimated to have received
only 1% of the immediate humanitarian aid, and between 15 and 21%
of the longer term recovery funding for reconstruction and develop-
ment. NGOs and private contractors were the primary recipients of the
aid in Haiti, and despite building extensive infrastructure with the use of
public funds, had limited accountability to the national health authorities
and lacked transparency as to quality and services offered. Further, with
limited infusion of funds into the existing public health system, Haitian
clinics and hospitals continued to lack the necessary resources to func-
tion. With this in context, the 2014 Haiti Humanitarian Action Plan,
developed through OCHA, sought to address the ongoing critical hu-
manitarian needs but also focused at strengthening national capacities.
The review of the response also demonstrated inconsistent capaci-
ties and performance between agencies, and, (as concluded by Sir John
Holmes, former Under-Secretary General for Humanitarian Affairs and
Emergency Relief Coordinator) ‘the influx of many hundreds of hu-
manitarian organizations, many of whom, while well-​meaning, were not
necessarily professional and well-​informed in their approach, posed a
huge challenge to coherence’. In addition to reforms aimed at coord-
ination, the review highlighted the need ‘to develop principles, cri-
teria and standards for medical teams that respond to emergencies
and disasters’. This led to the development of the Classification and
Minimum Standards for Emergency Medical Teams in sudden-​onset
disasters, and fed into the formation, in 2014, of the Core Humanitarian
Standards on Quality and Accountability.
198
section 2  Background to medicine
Although international actors receive the most media attention,
local and national agencies and authorities actually deliver the
greater assistance in conflict and disaster situations, particularly in
the critical initial phase. Over 80% of operational aid organizations
are local NGOs that work only within their home country. These na-
tional providers are the ‘first in’ and ‘last out’ and form the backbone
of the aid delivery system, but they are often underrecognized and
excluded from much of the available funding. Despite a consensus
on the essential role that national and local NGOs play within the
humanitarian sector, recent data suggest that these sectors receive
only a meagre share of global funding—​0.3% in 2017.
Funding of the sector has risen in absolute amount, but the
growth has not kept pace with need, indeed the increase in need
has stretched the capacity of the sector to its apparent limit. In 1992,
the first interagency humanitarian appeals by OCHA requested
(US) $2.7 billion, a figure which, by 2007, had only risen to $5.5
billion. The International Red Cross and Red Crescent Movement
(RCRC) set out its requirements separately, maintaining independ-
ence from the UN-coordinated appeals. By 2016, the UN joint ap-
peal was for $20.5 billion yet yielded only $12.4 billion in committed
funding, a 40% shortfall, and—irrespective of any shortfalls—needs
have continued to increase (Fig. 2.21.3). As of June 2017, the UN
joint appeal estimated a total requirement of $23.5 billion, to target
101.2 million people across 37 countries.
Changing humanitarian picture
The dizzying increase in global humanitarian needs show no signs
of abating. Today, the primary cause of human suffering and the re-
sultant humanitarian response is intricately linked to conflict, par-
ticularly those which are increasingly complex and protracted. With
comparison to the historic precedence, today’s crises are increas-
ingly referred to ‘complex humanitarian emergencies’, defined by the
World Health Organization as ‘situations of disrupted livelihoods
and threats to life produced by warfare, civil disturbance and large-​
scale movements of people, in which any emergency response has to
be conducted in a difficult political and security environment’.
Modern conflicts increasingly occur in urban rather than rural
­settings, and now affect middle income countries as often as poorer
ones. The impact of such conflicts includes a massive surge in displace-
ment, a regression on development, and a general increase of human
suffering. The humanitarian consequences are massive and persistent.
The International Committee of the Red Cross (ICRC) spends most of
its budget on protracted conflicts, and in 2013, two thirds of the hu-
manitarian aid from the Development Assistance Committee (DAC)
donors was spent in countries which had been recipients of substantial
humanitarian support for at least the previous 8 years.
Protracted conflicts and other man-​made crises have already con-
tributed to unparalleled levels of displacement and 2016 marked an
unprecedented challenge for the humanitarian community. UNHCR
reported 65.6 million people forcibly displaced due to persecution,
conflict, violence, or human rights violations, exceeding even the
levels seen after the Second World War (Fig. 2.21.4) This equates
to nearly 1% of the global population and the rate of new displace-
ments is calculated to be 20 people every minute. Most refugees ori-
ginate from the three zones of continued violence, Syria (5.5 million),
In billion US$
20
15
10
5
0
2004
2016
2014
Funding
Gap
Requested
2012
2010
2008
2006
Fig. 2.21.3  Funding trends 2004-​2016.
97
0
10
20
30
40
50
60
70
Displaced population (millions)
Proportion displaced
(number displaced per 1000 world population)
98
99
00
01
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
0
1
2
3
4
5
6
7
8
9
10
Refugees and asylum-seekers
Internally displaced persons
Proportion displaced
Fig. 2.21.4  Trends in global displacement and proportion displaced (1997–​2016).
From Global Trends: Forced Displacement in 2016, © 2017 United Nations High Commissioner for Refugees.
2.21  Humanitarian medicine
199
Afghanistan (2.5 million) and South Sudan (1.4 million). Despite the
predominance of news coverage on the European refugee influxes,
the reality is that the least wealthy countries are actually hosting
the vast majority of those displaced. In 2016, 84% of the refugees
were hosted in developing countries, with Turkey assuming the lar-
gest number worldwide for the third consecutive year (2.9 million).
Lebanon, meanwhile, hosted the largest number of refugees relative
to population size, and current estimates are that one out of every
six people in that country is a refugee. Pakistan, Iran, Uganda, and
Ethiopia were the other primary countries of asylum for refugees.
The challenges of displacement for the humanitarian community
relate not simply to the sheer magnitude of the issue but also to the
changing contextual picture. Displacement today no longer adheres
to traditional precedents that were set in rural areas or in camp set-
tings. Today’s response requires evolving practices, approaches, and
interventions, in particular those that recognize and are sensitive to
often dire historical contexts. Current crises in settings such as Yemen
and Syria, where prewar health systems were functional, means that
humanitarian providers must adapt traditional responses and con-
sider the voids left by previously-​existing tertiary level or specialist
care. The expectations placed on humanitarian providers in such set-
tings can be untenably high. Requests to support dialysis units or ter-
tiary specialist care are totally different propositions from traditional
refugee responses, factors which have substantially increased the
complexity of humanitarian programming in recent years. Further,
working in dense urban areas poses different challenges with regard
to shelter and housing, water and sanitation, engagement with es-
tablished municipal and governmental systems, and even locating
marginalized or displaced populations. Despite years of operational
experience, humanitarian practices are currently, and rapidly, being
adapted to today’s picture. But this takes time, and the humanitarian
community may not have the desired resources for this investment as
the needs of millions are generally urgent and pressing.
Instability and insecurity in humanitarian space
Addressing the humanitarian consequences of today means
engaging in a sector that is more complex and hazardous than ever
before. With increased involvement of non​state actors, conflicts can
often no longer be considered to be between two recognizable sides.
Fragmented structures can mean it is often difficult to know who is
in charge, and the laws of war and the associated humanitarian pro-
tections are often forgotten.
Perhaps the most fundamental issues facing humanitarian pro-
viders today is a decrease in respect for and enforcement of humani-
tarian principles. Violations against humanitarian law, including a
lack of respect for the protections of civilians, humanitarian, and
medical workers, has risen to unprecedented levels and become a
threat to the foundations of the sector. Analysis of data by the World
Health Organization (WHO) showed that within the 2 year period
of 2014 and 2015, there were 594 reported attacks on health care
­facilities or personnel across 19 different countries with emer-
gencies, resulting in 959 deaths and 1561 injuries. Médecins Sans
Frontière (MSF), meanwhile, reported attacks on 63 of their facil-
ities in Syria in 2015, resulting in the total destruction of 12 facilities
and the injury or death of more than 80 MSF-​supported medical
staff in that year. There has been an appalling lack of accountability
against the perpetrators of such crimes, allowing violations of inter-
national law to occur with seeming impunity.
The campaign #NotATarget was launched by MSF in 2015 to stand
in solidarity with those targeted and to demand the ‘right of humani-
tarian workers to provide humanitarian care to people in need, par-
ticularly in conflict zones’. The laws to provide these protections are
already in place, but without enforcement they hold little power.
When nations and warring parties disregard the international ob-
ligations and the protections in place for the impartial provision of
healthcare, humanitarian action itself is threatened. Now more than
ever, efforts must be made to strengthen and protect this system,
and those who work within in it. In May 2016, UN Security Council
Resolution 2286 passed, which condemned the attacks on medical
facilities and demanded compliance with international law. At pre-
sent, however, global leaders have not provided much substantive
action to support the rhetoric.
Political change and forward uncertainty
Humanitarian Medicine has grown in prominence because it is im-
portant, indeed essential. Those who practise humanitarian medi-
cine have the fundamental ability to reaffirm humanity and help
relieve human suffering. Yet the humanitarian sector now appears to
be at a crossroads, with capacity stretched to a limit and insufficient
funding, and today’s geopolitical context portends uncertainty.
Throughout modern history, humanitarian principles have played
a critical role in the geopolitical sphere, from Britain’s commitment
to end the slave trade in early 1800s to the international response
for those affected by the Northern Chinese Famine of 1876-​1879.
In 1904, United States President Theodore Roosevelt used the op-
portunity of his Annual Message to Congress to reaffirm a place in
politics for humanity, stating that:
‘ . . . there are occasional crimes committed on so vast a scale and
of such peculiar horror as to make us doubt whether it is not our
manifest duty to endeavor at least to show our disapproval of
the deed and our sympathy with those who have suffered by it . . .
(and) . . . in extreme cases action may be justifiable and proper’.
After the Second World War, these commitments were repeated
and appeared firmly rooted in the foundations of liberal govern-
ance, but these sentiments appear disconnected with the realities of
the present day, and today’s global world order no longer conveys
any semblance of solidarity. The predominant views from leaders
is often decidedly more inward-​looking and self-​centred. Decisions
made in response to the refugee influx in Western countries include
the tightening of entry of refugees to the point of near stranglehold.
The EU-​Turkey deal, signed by 28 European Union Heads of State
in March 2016, agreed to the return to Turkey of ‘all new irregular
migrants’ arriving on Greek shores, despite concerns about poten-
tial violations of both EU and International Humanitarian Law. The
traditions of humanitarian principles, which once ran deep in world
politics, now appear reduced to strategic interest. The moral respon-
sibilities of today’s world leaders can no longer be assumed.
At this profoundly transformative time, those practising hu-
manitarian medicine continue to uphold moral imperatives. They
are demonstrating a commitment to medical excellence, with a
broadening and strengthening of their capacities to meet the de-
mands of those in need. But it is crucial that these and other active
groups realize that the practice of a humanitarian provider should
200
section 2  Background to medicine
go well beyond the therapeutic acts of relieving suffering and saving
lives, as it serves to exert the general principles of humanity. The
mere presence of humanitarian providers indicates a solidarity with
those in need and helps to restore dignity for those impacted by
crises. By bearing witness and demanding accountability, they re-
ject the placid acceptance of suffering. Through their actions the
capacity of compassion and the potential of humanity displayed,
which at this particular time in history, perhaps more than ever be-
fore, is imperative.
FURTHER READING
Bradol J-​H, Vidal C (eds) (2011). Medical interventions in humani-
tarian situations: The work of Médecins Sans Frontières. Translated
by Christopher Brasher, Nina Friedman, Philippa Bowe Smith,
Karen Stokes, and Karen Tucker. Médecins Sans Frontières (MSF),
Geneva, Switzerland.
Kravitz A (2019). Oxford handbook of humanitarian medicine. Oxford
University Press, Oxford.

2.22 Complementary and alternative medicine 201
2.22 Complementary and alternative medicine 201
ESSENTIALS
Complementary and alternative medicine can be defined as diagnosis,
treatment, and/​or prevention which complements mainstream medi-
cine by contributing to a common whole, by satisfying a demand not
met by orthodoxy, or by diversifying the conceptual frameworks of
medicine. It is popular; hence doctors should know about it.
Why is complementary and alternative medicine popular?
The following motivations may be important: (1) to leave no thera-
peutic option untried; (2) to take control over one’s own health; (3) to
accord one’s healthcare with one’s global outlook; (4) to benefit from
natural and, by implication, safe treatments; (5) to be given time,
understanding, and empathy by a practitioner; (6) disenchantment
with conventional medicine/​science.
Types of complementary and alternative medicine
The term covers a vast array of treatments and diagnostic techniques
which have little in common except that they are not part of main-
stream medicine. The most important modalities are (1) acupuncture—​
probably effective for some painful conditions and for nausea/​
vomiting; rarely causes severe adverse events. (2)  Phytotherapy—​
treatment with herbal extracts; can be evaluated by assessing each of
the many remedies separately; some phytomedicines are supported
by sound evidence. (3) Homeopathy—​based on irrational concepts
of ‘like cures like’ and ‘potentizing’ (shaking and stepwise dilution of
drugs); trial data fail to show efficacy for any condition. (4)  Spinal
manipulation—​may be mildly effective for back pain as practised by
chiropractors, osteopaths, physiotherapists, and other healthcare pro-
fessionals; claims that it also works for many other conditions are not
supported by sound evidence; can cause significant side effects (e.g.
manipulation of the cervical spine causes transient adverse events in
about half of all patients and has been associated with serious compli-
cations such as dissection of the vertebral artery).
Definition
Most healthcare professionals feel they know intuitively what is
meant by complementary and alternative medicine (CAM), yet an
adequate definition is hard to find. Often CAM is described by char-
acteristics that exclude it from mainstream medicine, for example:
•	not taught in medical school
•	not scientifically proven
•	not based on a scientific rationale
•	not used in routine healthcare
CAM can be positively defined as ‘diagnosis, treatment, and/​or pre-
vention which complements mainstream medicine by contributing
to a common whole, by satisfying a demand not met by orthodoxy or
by diversifying the conceptual frameworks of medicine’.
CAM encompasses a large variety of techniques which have little
in common except that they are not part of mainstream medicine,
claim to offer help for most conditions, and pride themselves on a
holistic approach to patient care (Table 2.22.1). Some relate to thera-
peutic modalities (e.g. herbalism), some to diagnostic techniques
(e.g. iridology), and some include both diagnostic and therapeutic
modalities (e.g. acupuncture).
There are considerable local differences in what are regarded as
CAM or mainstream medicine. In Germany, for instance, massage
therapy and herbalism are orthodox, whereas in English-​speaking
countries they are usually regarded as CAM. Acupuncture is CAM
in the West, while in China it is a widespread, traditional, and gener-
ally accepted form of treatment.
Since most CAM therapies are used as adjuncts to conventional
treatments, ‘complementary’ is a more appropriate term than ‘al-
ternative’. When used as a true alternative to mainstream medicine,
CAM can become a hazard to patients even if the treatment itself is
without risks. In many countries, including the United Kingdom,
CAM is practised mostly by healthcare professionals who are not
medically trained, often in the absence of stringent regulation,
leading many to be concerned that vulnerable patients may be
exploited.
Prevalence
The 1-​year prevalence of CAM usage by the general population
ranges from 10% in the United Kingdom to 62% in Germany.
In patient populations, these figures can be considerably higher.
2.22
Complementary and alternative medicine
Edzard Ernst
202
section 2  Background to medicine
For instance, most cancer patients try one form of CAM or another.
The annual expenditure for CAM exceeds (US) $20 billion in the
United States of America and £1.6 billion in the United Kingdom.
In industrialized countries, typical users of CAM are middle-​
aged, female, well-​educated members of a high socioeconomic class.
Indications for CAM range from chronic benign conditions where
mainstream medicine does not offer a cure (e.g. back pain) to life-​
threatening diseases like cancer and AIDS. Most patients try CAM
in parallel with conventional treatment yet 30 to 50% do not tell their
conventional healthcare providers that they do so. A medical history
should include specific questions about CAM.
Reasons for CAM’s popularity
There is not one but a complex mix of reasons; the following motiv-
ations may be important:
•	to leave no therapeutic option untried
•	to take control over one’s own health
•	to align one’s healthcare with one’s global outlook
•	to benefit from natural and, by implication, safe treatments
•	to be given time, understanding, and empathy by a practitioner
•	disenchantment with conventional medicine/​science
Table 2.22.1  Various other forms of therapeutic and diagnostic methods
Name
Principle
Main indications/​
reasons for use
Efficacy
Risks
Alexander technique
Training process of ideal body posture and
movement; developed by F M Alexander
Musculoskeletal
problems
Very few clinical trials
No serious adverse
effects on record
Applied kinesiology
Diagnostic technique using muscle strength
as an indicator; developed by G. Goodheart
n.a.
Repeatedly shown to be not valid
Can delay reliable
diagnoses
Aromatherapy
Application of essential oils usually through
gentle massage techniques; developed by
R.M. Gattefossé
Relaxation
improvement of
well-​being
Systematic reviews are mostly
inconclusive
Allergic reactions to oils
Autogenic training
Form of self-​hypnosis for relaxation and
stress reduction; developed by J. Schultz
Stress management
Some evidence for effectiveness
No serious adverse
effects on record
Chelation therapy
Intravenous infusion of EDTA used in CAM
for ‘deblocking’ arteries from arteriosclerotic
lesions
Circulatory disorders
Shown to be ineffective
Serious adverse effects,
even deaths, reported
Chiropractic
Popular manual therapy based on the
assumption that most health problems
are due to misalignment of the spine and
treatable through spinal manipulation;
developed by D Palmer; seen as mainstream
by many proponents
Back pain, neck pain,
and many others
Cochrane reviews of chiropractic
for back pain show it is not
superior to standard therapies,
no good evidence for other
indications
Serious adverse effects
have been reported,
their exact incidence is
not known
Colonic irrigation
(or colon therapy)
Cleansing of the colon through enemas with
water or coffee for ‘detoxication’
Various
No sound evidence for
effectiveness
Serious adverse effects
reported
Hypnotherapy
Induction of trance-​like state to influence the
unconscious mind
Various
Some evidence for effectiveness
Serious adverse effects
probably infrequent
Iridology
Diagnostic technique using signs and
impurities on the iris
n.a.
Repeatedly shown to be not valid
Can delay reliable
diagnoses
Macrobiotic diet
Diet based on the yin/​yang principle using
whole grains and vegetables
Disease prevention
Positive effects on cardiovascular
risk factors
Serious adverse effects
reported
Massage
Various techniques of manual stimulation
of cutaneous, subcutaneous, or muscular
structures (deemed mainstream on the
European continent)
Musculoskeletal
problems, anxiety, and
many others
Some evidence for effectiveness in
musculoskeletal and psychological
problems
No serious adverse
effects on record
Osteopathy
Health problems are thought to be due to
misalignment of the spine and corrected
through spinal mobilization; developed
by T Still; seen as mainstream by many
proponents
Back pain, neck pain,
and many others
Systematic reviews of osteopathy
for back pain are inconclusive
Adverse effects less
than with chiropractic
Reflexology
Internal organs correspond to areas on
the sole of the feet and can be influenced
through massaging these
Relaxation
Systematic review was inconclusive
No serious adverse
effects on record
Spiritual healing
Channelling of ‘healing energy’ through a
healer into a patient
Re-​establishing a
wholesome balance
Best evidence fails to show
effectiveness
No serious adverse
effects on record
Yoga
Meditative, postural, and breathing
techniques from ancient India
Various
Some evidence for effectiveness
in asthma, or cardiovascalur risk
factors, for instance
No serious adverse
effects on record
n.a., not applicable.
2.22  Complementary and alternative medicine
203
Examples of CAM methods
Acupuncture
Description
Traditionally, the Chinese believed that the life energy (Qi or Chi)
flowing in particular channels (meridians) governs human health.
The energy is a balance of opposite characteristics: yin and yang.
Illness is understood as an expression of an imbalance between yin
and yang. One way of re-​establishing the proper equilibrium would
be to insert needles in acupuncture points located along the merid-
ians. Instead of or in addition to needles, acupuncturists also use
pressure (acupressure), laser light (laser acupuncture), electrical
currents (electroacupuncture), heat (moxibustion), or other stimuli
to stimulate acupuncture points. Neither the meridians nor the acu-
puncture points have a morphological basis, and the theory of yin
and yang is not supported by facts.
Mode of action
Neurophysiological research has created a (hypothetical) rationale
for acupuncture: activation of brainstem nuclei, and the release of
neural transmitters and endorphins in the brain and descending in-
hibitory control systems.
There are considerable differences between traditional Chinese
and Western acupuncture. In traditional Chinese medicine, con-
ventional diagnoses are not normally sought, treatment is highly
individualized according to each patient’s particular yin/​yang im-
balance, and acupuncture is employed as a ‘cure all’. In contrast,
Western acupuncturists tailor their treatment to the conventional
diagnosis established beforehand and use acupuncture for selected
indications for which it might be efficacious.
Efficacy
Many trials of acupuncture exist but are fraught with methodological
problems, such as placebo effect and blinding of patients or therap-
ists. Over 200 systematic reviews and meta-​analyses of acupuncture
trials for various conditions have been published. As they are often
based on biased primary data, their conclusions are not always reli-
able. According to such reviews, the most convincing evidence exists
for the following conditions:
•	chronic back pain
•	dental pain
•	gastrointestinal endoscopy
•	idiopathic headache
•	osteoarthritis of knee
•	postoperative nausea and vomiting
For other indications, the data remain inconclusive because of
highly contradictory findings, poor quality research, insufficient
quantity of primary data, or a total absence of reliable studies.
Recently, new non​penetrating acupuncture devices have become
available which best control for placebo effects in clinical trials.
The results of clinical trials employing such devices tend to sug-
gest that most of the therapeutic response to acupuncture relies on
­placebo effects. Hardly any acupuncture studies are conducted with
masking the therapist according to group allocations of patients.
Safety
Serious complications of acupuncture include:
•	trauma (e.g. cardiac tamponade, pneumothorax)
•	infections (e.g. viral hepatitis)
With well-​trained therapists, such complications are rare. However,
mild adverse effects (e.g. pain or bleeding at the site of needling)
occur in about 10% of all patients. In addition, there are indirect
risks. For instance, some acupuncturists advise their patients about
prescription drugs without having the medical competence to do so.
Phytotherapy
Description
Medical herbalism (phytotherapy) is treatment with whole plants,
parts of plants, or plant extracts. The term does not cover treatment
with single active constituents such as acetylsalicylic acid, origin-
ally derived from willow bark. Since all plants contain a multitude
of chemicals, phytotherapy involves treatment with a mixture of poten-
tially active compounds. In many cases there is uncertainty about
the most important active ingredients and their pharmacological ac-
tions. The claim of herbalists that the whole plant (extract) will yield
more beneficial effects than any single isolated ingredient (synergy)
is largely unproven.
Most medical cultures have their version of traditional herb-
alism. Traditional Chinese medicine has a long history of employ­
ing mixtures of herbs to prevent and treat disease. This tradition
was modified by the Japanese and resulted in Kampo medicine.
The Indian tradition has generated Ayurvedic medicine which
relies heavily on plant-​based remedies. Likewise, European herb-
alism has a tradition which is as old as European medicine itself.
The scientific investigation of medicinal herbs is, however, a rela-
tively recent innovation.
Mode of action
There are few differences in principle between pharmacotherapy
and phytotherapy except that herbal remedies are multicomponent
systems which render them pharmacologically more complex.
There is no reason why the rules of pharmacokinetics and pharma-
codynamics should not apply. Discernible modes of action exist for
every plant-​based medicine. In some cases, these have been eluci-
dated; in many other cases, they remain hypothetical.
Efficacy
Based on authoritative systematic reviews and meta-​analysis, good
or at least encouraging evidence exists for the efficacy of the fol-
lowing herbal remedies:
•	Andographis paniculata for upper respiratory tract infections
•	black cohosh (Actaea racemosa) for alleviating menopausal
symptoms
•	cranberry (Vaccinium macrocarpon) for prevention of urinary
tract infections
•	devil’s claw (Harpagophytum procumbens) for treating musculo-
skeletal pain
•	garlic (Allium sativum) for hypercholesterolaemia
•	Ginkgo biloba for intermittent claudication
204
section 2  Background to medicine
•	green tea (Camellia sinensis) for prevention of cancer and cardio-
vascular disease
•	hawthorn (Crataegus spp.) for treatment of chronic heart failure
•	horse chestnut (Aesculus hippocastanum) seed extract for primary
venous insufficiency
•	kava (Piper methysticum) as an anxiolytic drug
•	nettle (Urtica dioica) for benign prostate hyperplasia
•	red clover (Trifolium pratense) for hot flushes during menopause
•	saw palmetto (Serenoa repens) for benign prostatic hyperplasia
•	St John’s wort (Hypericum perforatum) for mild to moderate
depression
•	valerian (Valeriana officinalis) for insomnia
•	willow (Salix spp.) bark for pain
For many other popular medicinal herbs, too few clinical trials have
been carried out, the studies are methodologically flawed, or their
results are contradictory. No good evidence exists for the efficacy of
traditional approaches to herbal medicine where plant mixtures with
a multitude of ingredients are used depending, not on a conventional
diagnosis, but on the individual patient’s set of symptoms, consti­
tution, or other circumstances. Yet these traditional approaches are
the ones likely to be applied if a patient consults a medical herbalist.
Safety
Many medicinal herbs have been associated with serious adverse ef-
fects (see Chapter 10.4.3), for example:
•	aconite (Aconitum) cardiotoxic
•	Aristolochia nephrotoxic
•	black cohosh (Actaea racemosa) hepatotoxic
•	broom (Cytisus scoparius) cardiotoxic
•	chaparrall (Larrea tridentate) nephrotoxic
•	comfrey (Symphytum officinale) hepatotoxic
•	kava (Piper methysticum) hepatotoxic
•	liquorice root (Glycyrrhiza glaba) induces hypokalaemia
•	pennyroyal (Mentha pulegium) hepatotoxic
•	skullcap (Scutellaria lateriflora) hepatotoxic
Herbal remedies can interact powerfully with synthetic drugs
(Table 2.22.2), and Asian herbal medicines have been shown re-
peatedly to be adulterated with synthetic drugs or contaminated
with heavy metals. In many countries (e.g. United Kingdom and the
United States of America) herbal medicines are marketed as food
(or dietary) supplements in the absence of stringent quality control.
Homeopathy
Description
Samuel Hahnemann, a German physician, believed in two major
principles which formed the basis of an entirely new school of medi-
cine: homeopathy. The ‘like cures like’ principle postulates that, if a
given drug induces symptoms (e.g. a headache) in healthy individ-
uals, it can be employed to treat headaches in patients who suffer
from it. The second principle holds that ‘potentizing’ (i.e. shaking
and stepwise diluting) drugs makes them more potent for the
treatment of illness. Homeopathic dilutions prepared in this way
are believed to be clinically effective even if not a single molecule
of the original medicine is contained in the potentized remedy.
For 200 years, scientists have pointed out that these principles fly in
the face of science and that therefore homeopathy cannot possibly
work beyond a placebo effect. Homeopaths, however, insist that
their remedies act via ‘energy’ transfer from the original substance
to the diluent (the theory of a ‘memory of water’).
Homeopaths do not treat diseases but claim to treat the whole in-
dividual. They take a detailed history with the aim to match the to-
tality of the symptoms and characteristics of that patient with a ‘drug
picture’ (the ‘like cures like’ principle). This homeopathic remedy,
given in the correct potency, should then be the optimal treatment
for that patient. Clinical improvement may, however, take weeks or
months, and, in about 20% of all cases, symptoms may deteriorate
before they become better, a phenomenon termed ‘homeopathic
aggravation’.
At the time of Hahnemann there were very few effective treat-
ments and many that were overtly harmful. Homeopathic remedies
had virtually no adverse effects. Hahnemann can therefore be cred-
ited with clinically exploiting the placebo effect to the best benefit
of his patients. It is hardly surprising then that homeopathy con-
quered many countries (e.g. France, the United States, India, South
America) by storm. The advent of effective synthetic drugs and other
efficacious treatments lead to the sharp decline of homeopathy; the
recent boom of CAM, however, has brought about a revival.
Mode of action
Several hypotheses have been developed to explain the transfer of
‘energy’ from the mother tincture to the diluent. However, none has
so far withstood the scrutiny of independent assessment. Neither
has the ‘energy’ ever been defined in physical terms, nor are there
rational explanations as to how this ‘energy’ (if it exists) might af-
fect human health. Therefore, homeopathy remains among the least
plausible forms of CAM.
Efficacy
A meta-​analysis of all 89 randomized and/​or placebo-​controlled
clinical trials published by 1995 calculated an overall odds ratio
of 2.45 in favour of homeopathy. When only the 26 most rigorous
studies were meta-​analysed, the odds ratio fell to 1.66 but remained
statistically significant. This publication was criticized (e.g. for
pooling data for all medical conditions and all homeopathic rem-
edies, and for including trials that were not randomized or placebo-​
controlled and studies of material (low dilution) remedies for which
efficacy is not disputed). The results of about a dozen subsequent
systematic reviews generally fail to demonstrate effects beyond pla-
cebo. Therefore, the best evidence available to date fails to suggest
efficacy.
Safety
Highly diluted homeopathic remedies cannot cause pharmacological
adverse effects. Homeopaths claim that ‘homeopathic aggravations’
(an exacerbation of presenting symptoms after administration of the
optimal remedy) occur in about 20% of cases; if that were true, they
might represent a safety issue. ‘Indirect’ safety problems include the
substitution of effective interventions by homeopathy. For instance,
non-​medically qualified homeopaths tend to advise their clients
against immunization and advocate homeopathic remedies instead.
If this happens on a large scale, it jeopardizes herd immunity against
serious infectious diseases.
2.22  Complementary and alternative medicine
205
Table 2.22.2  Possible interactions between some popular herbal remedies and synthetic drugs
Herbal remedya
Usage or pharmacological effectb
Possible interaction
Aloe
(Aloe spp.)
Various
With chronic use, potentiation of cardiac glycosides or antiarrhythmic drugs due to loss of
potassium
Black cohosh
(Actaea racemosa)
Oestrogenic
Increased effects of antihypertensives
Borage
(Borago officinalis)
Anti-​inflammatory
Interaction with antiepileptics, may increase risk of seizure
Broom
(Cytisus scoparius)
Antiarrhythmic, diuretic
Increased effects of antidepressants, β-​blockers, and cardiac glycosides
Cascara
(Rhamus purshiana)
Laxative,
cathartic
Loss of potassium with chronic use, potentiation of cardiac glycosides, or antiarrhythmic
drugs
Chamomile
(Matricaria recutita)
Spasmolytic, anti-​inflammatory
May potentiate effects of anticoagulants through its coumarin content
Chasteberry
(Vitex agnus castus)
Hormonal effects
Increased effects of other hormonal drugs
Cranberry
(Vaccinium macrocarpon)
Urinary tract infections
May enhance elimination of drugs normally excreted in urine
Ephedra
(Ephedra sinica)
Central nervous system (CNS)
stimulant, sympathomimetic
Cardiac glycosides/​halothane: arrhythmias, guanethidine: enhanced sympathomimetic
effect,
monoamine oxidase enzyme (MAO) inhibitors: enhanced sympathomimetic effect,
secale alkaloids/​oxytocin: hypertension
Garlic
(Allium sativum)
Hypocholesterolaemic
Increased effects of anticoagulants and antiplatelet drugs
Ginger
(Zingiber officinale)
Antiemetic
Increased effects of anticoagulants
Ginseng
(Panax
ginseng)
Various
Interaction with MAO inhibitors, interaction with stimulants and phenelzine, increased
effect of hypoglycaemics
Hawthorne
(Crataegus spp.)
Digitalis-​like
Can increase hypotensive effects of nitrates, antihypotensives, cardiac glycosides, and CNS
depressants
Hops
(Humulus lupulus)
Hypnotic
Antagonism with antidepressants, can increase effects of CNS depressants and hypnotics,
interference with hormonal drugs
Horse chestnut
(Aesculus hippocastanum)
Anti-​inflammatory
Increased effects of anticoagulants
Kava
(Piper methysticum)
Anxiolytic
Potentiation with other anxiolytics, can increase parkinsonian symptoms with levodopa
Lavender
(Lavandula angustifolia)
Sedative
Increased effects of CNS depressants
Liquorice
(Glycyrrhiza glaba)
Corticosteroid activity for gastric
irritation
Potassium loss (e.g. with thiazide diuretics, water, and sodium retention with
corticosteroids), increased effects of digoxin, decreased effects of antihypertensives
Lily of the valley
(Convallaria majalis)
Congestive heart failure
Increased (side) effects of quinidine, calcium, saluretics, laxatives, glucosteroids, β-​blockers,
calcium channel blockers, and digitalis
Mistletoe
(Viscum album)
Anticancer drug
Increased effects of CNS depressants, antihypertensives, and cardiac drugs
Nettle
(Urtica dioica)
Diuretic
May potentiate effects of other diuretics
Pumpkin seed
(Curcubita pepo)
Anthelmintic, diuretic
Can increase effect of diuretics
Sage
(Salvia officinalis)
Antispasmodic
Interaction with antiepileptics, may increase risk of seizure, decreased effect with
antiglycaemics
St John’s wort
(Hypericum perforatum)
Antidepressant
Increased effects of digoxin MAO inhibitors or serotonin uptake inhibitors, decreased
effect of drugs metabolized by the cytochrome P450 enzyme system
Valerian
(Valeriana officinalis)
Hypnotic
Increased effects of CNS depressants and hypnotics
Yew
(Taxus spp.)
Antirheumatic, anticancer
Chemotherapeutic agents may potentiate its effects
a Plant source in brackets.
b Not comprehensive.
206
section 2  Background to medicine
Spinal manipulation
Description
In most cultures, spinal manipulation has been practised by bone­
setters for centuries. Today such therapeutic techniques are practised
by chiropractors, osteopaths, physiotherapists, doctors, and other
health professionals. Spinal manipulation is the hallmark therapy
for chiropractors who use it to adjust ‘subluxations’, malalignments
of the spine claimed to be at the root of all health problems. During
spinal manipulation vertebral joints are often moved beyond their
physiological range of motion but not far enough to destroy joint
structures. A  typical technique is the short-​lever, high-​velocity
thrust which is used by chiropractors on most patients.
Mode of action
Many chiropractors believe that vertebral ‘subluxations’ adversely
affect human health and that consequent spinal manipulation will
improve it. The mechanism of action is, however, unclear. Some
theories hold that it breaks fibrous adhesions within joints, that it
affects mechanoreceptors of the joint, or that it modulates central
nervous system excitability.
Efficacy
Most of the trial data pertain to back pain. A Cochrane review of
spinal manipulation found no evidence that it is superior to standard
treatments for acute or chronic back pain, but some evidence that
it is better than harmful interventions or sham treatments. For all
other indications (e.g. neck pain, headache, dysmenorrhoea, colic,
asthma), the current best evidence fails to indicate effectiveness.
Safety
Several prospective studies have shown that spinal manipulation
leads to transient, mild adverse effects such as local pain in about
50% of all patients. In addition, serious adverse effects such as arterial
dissection, stroke, and death have been reported in about 700 cases.
Chiropractors claim that the incidence of such complications is
exceedingly low. Due to significant underreporting, this may not be
so. At present, the true incidence of complications is not known and
no adequate reporting systems for adverse effects are in place.
Other forms of CAM
CAM is a highly diverse field comprising more than 150 different
forms of therapeutic and diagnostic methods (see Table 2.22.1).
FURTHER READING
Assendelft WJJ, et al. (2003). Spinal manipulative therapy for low back
pain. A  meta-​analysis of effectiveness relative to other therapies.
Ann Intern Med, 138, 871–​81.
Capasso F, et al. (2003). Phytotherapy: a quick reference to herbal medi-
cine. Springer Verlag, Berlin.
Cochrane reviews related to complementary medicine. https://​cam.
cochrane.org/​cochrane-​reviews-​related-​complementary-​medicine
Derry CJ, et al. (2006). Systematic review of systematic reviews of acu-
puncture published 1996–​2005. Clin Med, 6, 381–​6.
Ernst E (2002). A systematic review of systematic reviews of homeop-
athy. Br J Clin Pharmacol, 54, 577–​82.
Ernst E (2006). Acupuncture—​a critical analysis. J Intern Med, 259,
125–​37.
Ernst E (2008). Chiropractic:  a critical evaluation. J Pain Symptom
Manage, 35, 544–​62.
Ernst E, Smith K (2018). More Harm than Good?: The Moral Maze of
Complementary and Alternative Medicine. Springer.
Ernst E, et al. (2008). Oxford handbook of complementary medicine.
Oxford University Press, Oxford.
Mathie R, Fok Y, Viksveen P, To A, Davidson J (2019). Systematic Review
and Meta-Analysis of Randomised, Other-than-Placebo Controlled,
Trials of Non-Individualised Homeopathic Treatment. Homeopathy,
108(2), 88–101.
Newman M (2018). Is cancer crowdfunding fuelling quackery? BMJ,
362, k3829.
Reisman S, Balboul M, Jones T (2019). P-curve accurately rejects
evidence for homeopathic ultramolecular dilutions. PeerJ, 23(7),
e6318.
Rubinstein S, et al. (2019). Benefits and harms of spinal manipula-
tive therapy for the treatment of chronic low back pain: systematic
review and meta-analysis of randomised controlled trials. BMJ,
364, l689.
Singh S, Ernst E (2008). Trick or treatment? Alternative medicine on
trial. Bantam, London.
SECTION 3
Cell biology
Section editors: John D. Firth, Christopher P. Conlon, and Timothy M. Cox
3.1	 The cell  209
George Banting and Jean Paul Luzio
3.2	 The genomic basis of medicine  218
Paweł Stankiewicz and James R. Lupski
3.3	 Cytokines  236
Iain B. McInnes
3.4	 Ion channels and disease  246
Frances Ashcroft and Paolo Tammaro
3.5	 Intracellular signalling  256
R. Andres Floto
3.6	 Apoptosis in health and disease  266
Mark J. Arends and Christopher D. Gregory
3.7	 Stem cells and regenerative medicine  281
Alexis J. Joannides, Bhuvaneish T. Selvaraj,
and Siddharthan Chandran
3.8	 The evolution of therapeutic antibodies  296
Herman Waldmann and Greg Winter
3.9	 Circulating DNA for molecular diagnostics  299
Y.M. Dennis Lo and Rossa W.K. Chiu

2.3 The Global Burden of Disease Measuring the hea
2.3 The Global Burden of Disease: Measuring the health of populations 43
ESSENTIALS
To make the best decisions to improve health, policymakers need
reliable, up-​to-​date information on the major challenges facing
their country. The Global Burden of Disease study facilitates this
by providing comprehensive and scientifically rigorous estimates
of the causes of death and illness across the globe. It examines
a total of 84 risk factors and the amount of health loss attribut-
able to each or combinations of them. Analysis over time reveals a
shift from disease burden dominated by communicable, maternal,
neonatal, and nutritional causes to a burden increasingly made up
of non​communicable diseases and injuries. By making compari-
sons between countries or subnational units like states or coun-
ties, the Global Burden of Disease can highlight areas of particular
success or challenge, providing opportunities to examine what is
working, or what is not.
Introduction
To improve health by making the best decisions, policymakers need
reliable, up-​to-​date information on the major challenges facing their
country. The Global Burden of Disease (GBD) is a large-​scale en-
terprise dedicated to expanding the quantitative evidence base for
health by producing estimates of deaths, prevalence, and incidence
by disease for all countries and by identifying the major risks that
impinge on population health.
The GBD study was created to provide comparable, compre-
hensive, and scientifically rigorous estimates of the causes of death
and illness across the globe and aims to answer the following ques-
tions:  What are the world’s major health problems? How well is
society addressing these problems? Additionally, how do we best
dedicate resources to get the maximum impact in improving popu-
lation health in the future?
In the early 1990s, when the World Bank commissioned the ori-
ginal GBD study, researchers were surprised to discover that the an-
swers to these questions were elusive. For some parts of the world,
data were sparse or non​existent. In summing disease-​specific mor-
tality estimates by organizations whose mission was to combat a
given disease, GBD researchers found a total that was considerably
greater than the number of people, globally, who had actually died
in a particular year.
The GBD approach not only makes sure that deaths attributed to
different causes sum to the total number of deaths—​it goes beyond
mortality estimates and looks at causes of disability, or non​fatal con-
ditions. Increasingly, around the world, people are living longer and
are exposed to disease risks that are more common with increasing
age. Examining disabling conditions, such as back pain, dementia,
or diabetes is therefore more important than ever. With the diversity
of health conditions facing different populations, having compar-
able measurements is essential. The GBD approach allows policy-
makers to directly compare the public health impact of diseases that
often are fatal, such as cancers, and conditions that primarily disable,
such as depression or anaemia.
Underlying some of these conditions are risk factors including al-
cohol and tobacco use, dietary factors, occupational exposures, and
air pollution. The GBD 2017 study examined 84 risk factors and the
amount of health loss attributable to each or combinations of them.
This information can help decision-​makers to identify opportunities
to promote population health by preventive measures.
After the original GBD study, disease burden estimates by world
regions were produced on a regular basis in the 2000s by the World
Health Organization. In 2007 the Institute for Health Metrics and
Evaluation (IHME) was established at the University of Washington
in Seattle with funding from the Bill & Melinda Gates Foundation
to make these estimates by country and to regularly update esti-
mates. IHME published a comprehensive update, the Global Burden
of Diseases, Injuries, and Risk Factors Study 2010, in December
2012 in a dedicated issue of The Lancet. Since then, IHME has com-
mitted to producing annual updates of these results for the whole
time period from 1990s onward and for an increasing number of
countries and subnational units such as the provinces of China and
the states of India, Mexico, Brazil, and the United States. New risks,
diseases, and their disabling outcomes are added based on policy
interest. For instance, the GBD 2017 includes new estimates for type
1 and type 2 diabetes mellitus, and bullying victimization as a risk
factor for depression and anxiety.
Governments and funding partners want to maximize the im-
pact of their healthcare spending. By pinpointing the leading causes
2.3
The Global Burden of Disease:
Measuring the health of populations
Theo Vos, Alan Lopez, and Christopher Murray
44
SECTION 2  Background to medicine
of health loss, including risk factors for diseases, GBD can guide
how health services are planned and how resources are directed to
diseases and risks. By making comparisons between countries or
subnational units like states or provinces, GBD can highlight areas of
particular success or challenge, providing opportunities to examine
what is working, or what is not.
Methods
GBD quantifies health loss from hundreds of diseases, injuries, and
risk factors. In order to achieve this, a large effort is made to col-
lect all available data on deaths and the epidemiology of more than
300 diseases and injuries in 195 countries and territories, by age and
sex, from 1990 to the present. This large-​scale effort is coordinated
by IHME.
Health data for many countries are sparse, and they are often
messy. Researchers at IHME with guidance from a global con-
sortium of researchers—​more than 3300 in over 140 countries—​
identify every possible source of data for a country or condition and
work to gain access to them. Once in hand, the data are often ad-
justed to account for known differences in measurements between
data sources. The next step is synthesizing the data—​taking what
is known and making sense of it. Sophisticated statistical model-
ling approaches are used to estimate health loss for every disease
and country. Every estimate in GBD is estimated 1000 times over to
account for the impact of uncertainty from sampling errors, meas-
urement errors, and the choice of models.
Causes of health loss are classified at a first level into three broad
categories in the GBD:  communicable, maternal, neonatal, and
nutritional diseases; non​communicable diseases; and injuries.
The first of these encompasses diseases like HIV, tuberculosis,
malaria, diarrhoea, lower respiratory infections, iron-​deficiency
anaemia, and preterm birth complications. Non​communicable
diseases include heart disease, stroke, diabetes, cancer, depression,
and asthma. Injuries range from self-​harm (suicide) to road in-
juries to animal bites.
In order to account for death and disease, GBD uses the disability-​
adjusted life year (DALY). DALYs equal the sum of years of life lost
to premature mortality and years lived with disability. To calculate
years of life lost, a choice was made to set a norm for long life. This
norm was set by looking at the lowest observed mortality rates at
any given age in any population greater than five million. Years of
life lost are computed as the remaining life expectancy if an indi-
vidual had not died but continued to live at low risk of dying. This
works out as almost 88 years of remaining ‘ideal’ life expectancy for
an infant that has died, 39 years for someone who died at age 50,
and still another 2.2 years for someone dying at age 100. It reflects
an assumption that everyone, all over the world, deserves to live a
long life in full health.
Years lived with disability take into account the prevalence of
disability in a population as well as its severity. To make more than
2500 disabling outcomes of diseases and injuries included in the
GBD comparable, researchers mapped these into 234 homogenous
‘health states’ for which they wrote short lay descriptions of the main
symptoms and functional limitations such as pain, inability to move,
see, or hear, or changes in mood. These lay descriptions were put
to over 60 000 respondents in country and internet surveys, each
time in a randomly chosen pair, followed by the question ‘who is
the healthier’. From over half a million of these judgements, dis-
ability weights were derived that indicate the relative severity of a
health state as a number between 0 and 1. The disability weights in
GBD vary from 0.003 for mild vision loss to 0.778 for a person with
schizophrenia with fulminant psychosis. As we combine years of life
lost (YLLs) and years lived with disability (YLDs) into the DALY
measure by simple addition, it means that we equate the health loss
experienced by 333 cases of mild vision loss and a little more than
one person with schizophrenia in a year to a year of life lost due to
a death.
Another factor that is considered in creating GBD estimates is
that the world’s population is growing. An increase in the number
of deaths from a particular disease, then, might simply be explained
by the fact that there are more people alive who may then contract
the disease. One way to account for these changes in population is
to calculate rates of each condition, usually expressed as deaths or
DALYs per 100 000 people. In some cases, an apparent increase is
revealed as a decrease when considered this way. For example, in the
Democratic Republic of Congo between 1990 and 2017, the number
of deaths from tuberculosis increased by 45% but the rate of tuber-
culosis deaths per 100 000 actually decreased by 32% over the same
period. While the absolute burden in terms of number of people suf-
fering from TB increased, the proportion of people in the population
who are suffering actually decreased. The increase in the number of
deaths from TB indicates to health service planners that they need to
expand tuberculosis treatment resources even though the decline in
rates indicates that the existing TB control measures may be having
success.
Age-​standardizing also allows for more accurate comparison
across countries and time frames. This technique applies observed
age-​specific rates of a condition to a standard age distribution and
avoids distortion when one population is older or younger than
another.
Results
GBD 2017 generated 38 billion data points. In order to make these
results more accessible to researchers, policymakers, students, health
workers, journalists, and others, IHME created a group of interactive
data visualization tools that allow people to explore health trends
for different countries and regions. One of these, GBD Compare,
shows numbers and rates of deaths, YLLs, YLDs, and DALYs by dis-
eases and risk factors. The main view is a ‘tree map’, which is basic-
ally a square pie chart showing the proportional distribution of the
overall burden by underlying causes. Additional views can be added
showing a map of the world, time plots, age plots, and ranking lists.
Each can interactively be explored by country, age, sex, year, cause,
or risk factor. They can be found at https://vizhub.healthdata.org/
gbd-compare/.
Looking at global patterns in causes of death and disability over
time reveals a shift from disease burden dominated by communic-
able, maternal, neonatal, and nutritional causes (shown in red) to a
burden increasingly made up of non​communicable diseases (blue)
and injuries (green, Fig. 2.3.1). This pattern has been partly driven
by improvements in healthcare and greater access to vaccines and
life-​saving medical interventions that contributed to healthcare
2.3  The Global Burden of Disease: Measuring the health of populations
45
successes, such as the nearly 90% decrease in health loss from mea-
sles and tetanus globally between 1990 and 2017. Meanwhile, heart
disease, stroke (intracerebral haemorrhage and ischaemic stroke),
type 2 diabetes, low back pain, and chronic obstructive pulmonary
disease (COPD) climbed in ranking over this period to occupy 6 of
the top-​10 slots for causes of DALYs (per 100 000).
The rise in non​communicable diseases can be explained by sev-
eral factors. Economic development has propelled some countries
from low-​ to middle-​income status, often bringing improvements
in healthcare, water quality and sanitation, and living conditions. At
the same time, as countries gain higher economic status, air pollu-
tion may increase, and changes in diet and levels of physical activity
typically occur. Factors like these have led to an epidemiologic tran-
sition away from communicable, maternal, neonatal, and nutritional
disorders and towards a much greater burden of non​communicable
diseases. The exception to this trend has been the HIV/​AIDS epi-
demic: in 1990, HIV/​AIDS caused the 39th-​highest rate of deaths
(per 100 000) globally. In 2005, at the peak of the epidemic, this rank
had jumped to fifth place, and by 2017 it had dropped to 13th due to
increased access to antiretroviral treatment.
Greater economic development has also led to a demographic
transition, meaning that birth and death rates, typically high in low-​
income countries, decreased as industrialization occurred. This re-
sults in population growth, as death rates decrease faster than birth
rates, and later, population ageing. Policymakers must plan for the
results of these changes by expanding health services and antici-
pating the need to care for the changing spectrum of prevalent dis-
eases and comorbidities that are more common at older ages.
At the regional and country levels, more variation can be seen.
High-​income countries, including the United States, Australia,
Western Europe, and Japan, exhibit a disease burden dominated by
non​communicable diseases. Life expectancy in these countries is
typically high: between 78 and 84 years for both sexes combined in
2017 (Fig. 2.3.2), and the leading causes of health loss are all non-​
communicable diseases:  ischaemic heart disease, low back pain,
lung cancer, COPD, and Alzheimer’s disease and other dementias
are the top five. By contrast, life expectancy in sub-​Saharan Africa
ranged from 52 to 71 years; this region also has the greatest burden
of communicable diseases including malaria, diarrheal diseases,
HIV/​AIDS, and tuberculosis among the top ten causes of DALYs
Leading causes 1990
Leading causes 2017
1 Lower respiratory infections
1 Ischaemic heart disease
2 Diarrhoeal diseases
2 Lower respiratory infections
3 Preterm birth complications
3 COPD
4 Ischaemic heart disease
4 Diarrhoeal diseases
5 Neonatal encephalopathy
5 Preterm birth complications
6 COPD
6 Low back pain
7 Drug-susceptible tuberculosis
7 Intracerebral haemorrhage
8 Measles
8 Diabetes type 2
9 Malaria
9 Neonatal encephalopathy
10 Intracerebral haemorrhage
10 Ischaemic stroke
11 Other neonatal disorders
11 Migraine
12 Low back pain
12 Malaria
13 Protein-energy malnutrition
13 HIV/AIDS other
14 Drowning
14 Lung cancer
15 Ischaemic stroke
15 Drug-susceptible tuberculosis
16 Congenital heart anomalies
16 Falls
17 Self-harm by other means
17 Age-related hearing loss
18 Dietary iron deficiency
18 Major depression
19 Migraine
19 Other neonatal disorders
20 Pedestrian road injuries
20 Self-harm by other means
23 Dietary iron deficiency
22 Falls
27 Pedestrian road injuries
23 Lung cancer
29 Congenital heart anomalies
26 Diabetes type 2
36 Drowning
28 Major depression
38 Protein-energy malnutrition
30 Age-related hearing loss
73 Measles
38 HIV/AIDS other
Cause type:
Communicable, maternal,
neonatal, and nutritional
Noncommunicable
Injuries
Fig. 2.3.1  Top 20 causes of DALYs per 100 000 globally, both sexes, all ages, 1990 and 2017.
46
SECTION 2  Background to medicine
<55
55-59
60-64
65-69
70-74
75-79
≥80
Caribbean
LCA
DMA
ATG
TTO
GRD
VCT
TLS
MDV
BRB
SYC
MUS
COM
Persian Gulf
W Afr
E Med
MLT
SGP
Balkan Peninsula
TON
WSM
FSM
KIR
FJI
VUT
SLB
MHL
Fig. 2.3.2  Life expectancy at birth, both sexes, 2017. (ATG, Antigua and Barbuda; VCT, St Vincent and the Grenadines; BRB, Barbados; COM, the Comoros;
DMA, Dominica; E Med, Eastern Mediterranean; FJI, Fiji; FSM, Federated States of Micronesia; GRD, Grenada; KIR, Kirbati; MDV, Maldives; MHL, Marshall Islands;
MLT, Malta; MUS, Mauritius; LCA, St Lucia; SGP, Singapore; SLB, Solomon Islands; SYC, Seychelles; TON, Tonga; TTO, Trinidad and Tobago; TLS, Timor Leste; VUT,
Vanuatu; W Afr, West Africa; WSM; Western Samoa.)
2.3  The Global Burden of Disease: Measuring the health of populations
47
lost in 2017. Preterm birth complications, protein-​energy malnutri-
tion, neonatal encephalopathy from birth trauma and asphyxia, and
neonatal sepsis were also among the top 10 causes of health loss in
sub-​Saharan Africa in 2017.
Between 1990 and 2017, the absolute number of deaths from most
causes increased (Fig. 2.3.3). For example, cardiovascular disease
caused 11.9 million deaths in 1990 and 17.8 million in 2017, a 50%
increase (Fig. 2.3.3). Looking at the numbers of DALYs, mental dis-
orders were responsible for 82 million DALYs in 1990 and 123 mil-
lion in 2017, a 50% increase (Fig. 2.3.4). In terms of age-​standardized
rates, however, deaths from cardiovascular disease dropped by 30%
over this period, and DALY rates from mental and substance abuse
disorders changed marginally by 1.7%. The explanation for this dif-
ference is that cardiovascular deaths mainly occur in older people.
Despite a drop in the age-​by-​age rates of disease, population ageing
is causing the number of deaths to increase. In the case of mental
disorders, the rates of disease have not changed much over time and
the increase in DALY numbers is largely due to the increase in the
world population size.
China, in the period 1990 to 2017, provides a clear example of
the epidemiological transition. In 1990, the top 10 causes of DALYs
per 100 000 included two neonatal conditions along with lower re-
spiratory infections and diarrheal diseases (Fig. 2.3.5). By 2017,
however, all of the top-​10 causes were non​communicable diseases,
including ischaemic heart disease, stroke, diabetes, lung cancer,
stomach cancer, low back and neck pain, and age-​related hearing
loss. Injuries follow a less predictable pattern as countries move
through the epidemiological transition. In this example of China,
the decrease in DALY rates from drowning, by 79%, and suicide by
means other than firearms, by 67%, were remarkable.
Risk factors have undergone a similar transition globally. In 1990,
the top five risks, in terms of rates of DALYs, were child wasting,
short gestation, low birth weight, smoking, and high blood pres-
sure. Twenty-​seven years later, short gestation had dropped to
number five, while high blood pressure had taken over the number-​
one ranking. Smoking, high fasting plasma glucose, and high body
mass index came in at numbers two, three, and four, respectively
(Fig. 2.3.6). This trend in risk factors reflects the epidemiological
transition, where more health loss now results from poor diets
and overweight and obesity than inadequate nutrition. Deaths
and DALYs related to smoking have increased in countries like
Bangladesh and Indonesia but have decreased in most high-​income
countries.
A notable exception to these trends is sub-​Saharan Africa, where
many countries continue to experience a large burden of com-
municable, neonatal, and nutritional conditions. In 1990, these
made up all of the ten highest-​ranked diseases in terms of DALYs
per 100 000. By 2017, a few prominent changes had occurred:
HIV/​AIDS shot to number four, and measles dropped from
4th place to 18th, but the leading causes remained the same: lower
respiratory infections, malaria, diarrheal diseases, HIV/​AIDS, neo-
natal encephalopathy, preterm birth complications, tuberculosis,
protein-​energy malnutrition, and neonatal sepsis. Trends in risk
factors diverge from most other regions as well, with childhood
wasting at the top of the list, followed by short gestation, low birth
weight, unsafe sex, unsafe water, and unsafe sanitation. These find-
ings underscore the continuing need for health services focused on
communicable diseases and maternal and child health, as well as
infrastructure for clean water and sanitation in sub-​Saharan Africa
(Fig. 2.3.7).
0
10M
20M
30M
40M
50M
60M
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
Year
Deaths in millions
Cause of death
HIV/AIDS and sexually transmitted infections
Respiratory infections and tuberculosis
Enteric infections
Neglected tropical diseases and malaria
Other infectious diseases
Maternal and neonatal disorders
Nutritional deficiencies
Neoplasms
Cardiovascular diseases
Chronic respiratory diseases
Digestive diseases
Neurological disorders
Mental disorders
Substance use disorders
Diabetes and kidney diseases
Skin and subcutaneous diseases
Musculoskeletal disorders
Other noncommunicable diseases
Transport injuries
Unintentional injuries
Self−harm and interpersonal violence
Fig. 2.3.3  Global number of deaths by cause, both sexes, all ages, 1990–​2017.
48
SECTION 2  Background to medicine
Epidemiological trends in disease burden are best identified
by examining rates of deaths and DALYs; however, the absolute
number of people dying and suffering from various conditions also
informs healthcare needs. The distribution of medical specialties
might need to shift to accommodate changes in population and
in the leading causes of death and illness. A decreasing burden of
childhood illnesses would mean less need for paediatricians and
children’s hospitals over time. Larger numbers of elderly people
Leading causes 1990
Leading causes 2017
1 Lower respiratory infections
1 Ischaemic heart disease
2 COPD
2 Intracerebral haemorrhage
3 Intracerebral haemorrhage
3 COPD
4 Preterm birth complications
4 Ischaemic stroke
5 Ischaemic heart disease
5 Lung cancer
6 Drowning
6 Diabetes type 2
7 Neonatal encephalopathy
7 Neck pain
8 Self-harm by other means
8 Age-related hearing loss
9 Diarrhoeal diseases
9 Stomach cancer
10 Ischaemic stroke
10 Low back pain
18 Preterm birth complications
13 Stomach cancer
21 Neonatal encephalopathy
15 Lung cancer
23 Lower respiratory infections
16 Low back pain
26 Self-harm by other means
19 Neck pain
28 Drowning
21 Age-related hearing loss
78 Diarrhoeal diseases
29 Diabetes type 2
Cause type:
Communicable, maternal,
neonatal, and nutritional
Noncommunicable
Injuries
Fig. 2.3.5  Leading causes of DALYs per 100 000, both sexes, all ages, China, 1990 and 2017.
0
500M
1B
1.5B
2B
2.5B
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year
DALYs in billions
Cause of DALY
HIV/AIDS and sexually transmitted infections
Respiratory infections and tuberculosis
Enteric infections
Neglected tropical diseases and malaria
Other infectious diseases
Maternal and neonatal disorders
Nutritional deficiencies
Neoplasms
Cardiovascular diseases
Chronic respiratory diseases
Digestive diseases
Neurological disorders
Mental disorders
Substance use disorders
Diabetes and kidney diseases
Skin and subcutaneous diseases
Sense organ diseases
Musculoskeletal disorders
Other noncommunicable diseases
Transport injuries
Unintentional injuries
Self−harm and interpersonal violence
Fig. 2.3.4  Global number of DALYs by cause, both sexes, all ages, 1990–​2017.
2.3  The Global Burden of Disease: Measuring the health of populations
49
Occupational injury
Low physical activity
Lead
Vitamin A deficiency
Low omega−3
Child underweight
Iron deficiency
Low vegetables
Secondhand smoke
Handwashing
Unsafe sanitation
Drug use
Low nuts and seeds
Unsafe sex
Household air pollution
Impaired kidney function
Unsafe water
Low fruit
High sodium
Low whole grains
Ambient particulate matter
Child wasting
High LDL
Alcohol use
Low birth weight
Short gestation
High body mass index
High fasting plasma glucose
Smoking
High blood pressure
0
500
1k
1.5k
2k
2.5k
DALYs per 100 000
Risk factor
Cause of DALY
HIV/AIDS and sexually transmitted infections
Respiratory infections and tuberculosis
Enteric infections
Other infectious diseases
Maternal and neonatal disorders
Nutritional deficiencies
Neoplasms
Cardiovascular diseases
Chronic respiratory diseases
Digestive diseases
Neurological disorders
Mental disorders
Substance use disorders
Diabetes and kidney diseases
Sense organ diseases
Musculoskeletal disorders
Other noncommunicable diseases
Transport injuries
Unintentional injuries
Self−harm and interpersonal violence
Fig. 2.3.6  Leading risk factors in terms of DALYs per 100 000, both sexes, all ages, 2017.
with non​communicable diseases will require expansion of specialty
care for conditions affecting this population, including neurologists,
geriatricians, and elder care facilities.
Subnational studies, which are done at the state or province level,
provide an even more granular level of detail, revealing greater
disparities within countries than between neighbouring countries
and allowing decision-​makers to implement more targeted inter-
ventions. These have been conducted so far in China, Mexico, the
United Kingdom, Brazil, India, Russia, Iran, Indonesia, Japan, the
United States, Kenya, South Africa, Ethiopia, Sweden, Norway, and
New Zealand. China’s subnational study uncovered some striking
differences between provinces. In Shanghai, the age-​standardized
death rate for communicable, maternal, neonatal, and nutritional
diseases was 15.1 per 100 000 in 2017, similar to rates in high-​income
countries like Switzerland and Italy, but lower than rates in the United
States and Canada. In China’s western provinces like Xinjiang,
however, the death rate is several times higher than Shanghai’s and
comparable to death rates in Morocco, Vietnam, and El Salvador.
 
Leading risk factors 1990
Leading risk factors 2017
1 Child wasting
1 Child wasting
2 Unsafe water
2 Short gestation
3 Short gestation
3 Low birth weight
4 Child underweight
4 Unsafe sex
5 Vitamin A deficiency
5 Unsafe water
6 Unsafe sanitation
6 Unsafe sanitation
7 Low birth weight
7 No access to handwashing
8 No access to handwashing
8 Household air pollution
9 Child stunting
9 Child underweight
10 Household air pollution
10 Vitamin A deficiency
14 Child stunting
12 Unsafe sex
Risk factor:
Behavioural
Environmental
Fig. 2.3.7  Leading risk factors in terms of DALYs per 100 000, both sexes, all ages, sub-​Saharan
Africa, 1990 and 2017.
50
SECTION 2  Background to medicine
Conclusion
The Global Burden of Disease enterprise provides powerful tools for
clinicians and policymakers alike, to identify both large-​scale trends
and local disparities. Being aware of these factors can help decision-​
makers plan for future healthcare needs and make changes that will
reduce exposure to harmful risk factors. Healthcare providers, staff
at health ministries and international health agencies, researchers,
and others can also contribute to the ongoing GBD effort as collab-
orators. For more information on joining the GBD network, please
visit http://​www.healthdata.org/​gbd/​call-​for-​collaborators.
Collaborating with a worldwide network of researchers strengthens
both the data-​gathering effort and the quantitative analysis by
bringing together experts from a variety of disciplines. IHME and
its collaborators are expanding the list of diseases, injuries, and risk
factors included in GBD and routinely updating the GBD estimates.
Continual updates will ensure that the international community has
access to high-​quality estimates in the timeliest fashion. Through
sound measurement, it will provide the foundational evidence that
will lead to improved population health.
FURTHER READING
Foreman KJ, et  al. (2018). Forecasting life expectancy, years of life
lost, and all-​cause and cause-​specific mortality for 250 causes of
death: reference and alternative scenarios for 2016–​40 for 195 coun-
tries and territories. Lancet, 392(10159), 2052–​90.
GBD 2016 Healthcare Access and Quality Collaborators (2018).
Measuring performance on the Healthcare Access and Quality
Index for 195 countries and territories and selected subnational
locations: a systematic analysis from the Global Burden of Disease
Study 2016. Lancet, 391, 2236–​71.
GBD 2017 DALYs and HALE Collaborators (2018). Global, regional,
and national disability-adjusted life-years (DALYs) for 359 diseases
and injuries and healthy life expectancy (HALE) for 195 countries and
territories, 1990–2017: a systematic analysis for the Global Burden
of Disease Study 2017. Lancet, 392, 1859–1922.
IHME (2018). GBD Compare. https://vizhub.healthdata.org/gbd-
compare/
Murray CJL, Lopez AD (2017). Measuring global health: motivation
and evolution of the Global Burden of Disease Study. Lancet, 390,
1460–​4.

2.4 Large- scale randomized evidence Trials and me
2.4 Large- scale randomized evidence: Trials and meta- analyses of trials 51
ESSENTIALS
Reliable detection or refutation of realistically moderate
effects on major outcomes often requires large-​scale
randomized evidence
As long as doctors start with a healthy scepticism about the many ap-
parently striking claims and counterclaims that appear in the medical
literature, trial results do make sense. The main enemy of common
sense is overoptimism:  there are a few striking exceptions where
treatments for serious disease really do turn out to work extremely
well, but in general most claims of vast improvements from new ther-
apies turn out to be evanescent.
Clinical trials generally need to be able to detect or to refute real-
istically moderate (but still worthwhile) differences between treat-
ments in long-​term disease outcome. Large-​scale randomized
evidence should be able to detect such effects, but medium-​sized
trials or medium-​sized meta-​analyses can, and often do, yield false-​
negative or exaggeratedly positive results. If the results from such
studies seem too good to be true then they probably are: conversely,
unpromising evidence can be misleading if it is from a study of inad-
equate size, or from one particular subgroup of a large study with a
clearly favourable overall result.
Realistically moderate expectations of what a treatment might
achieve (or, if one treatment is to be compared with another, of how
large any difference between the main effects of these two treatments
is likely to be) should foster studies that can discriminate reliably be-
tween (1) a difference in outcome that is realistically moderate but
still worthwhile, and (2) a difference in outcome that is too small to
be of any material importance.
To assess moderate effects reliably, avoid both moderate
biases and moderate random errors
To demonstrate or refute realistically moderate differences in out-
come, studies must guarantee (1)  strict control of bias—​which, in
general, requires proper randomization and appropriate statistical
analysis, with no unduly ‘data-​dependent’ emphasis on specific parts
of the overall evidence; and (2) strict control of the play of chance—​
which, in general, requires large numbers of patients with the out-
come of interest, rather than a lot of detail on each patient. The
conclusion is obvious: moderate biases and moderate random errors
must both be avoided if moderate benefits are to be assessed reliably.
This leads to the need for large numbers of properly randomized pa-
tients with properly analysed data, which in turn should lead to some
large but simple randomized trials (or ‘mega-​trials’) and to large sys-
tematic overviews (or ‘meta-​analyses’) of all related randomized trials.
Other forms of evidence may be untrustworthy
Non​randomized evidence, unduly small randomized trials, unduly
small meta-​analyses of trials, and undue emphasis on particular sub-
groups (or on particular trials) are all inferior sources of evidence
about current patient management or as foundations for future
research strategies because they often cannot discriminate reliably
between moderate (but worthwhile) differences and negligible dif-
ferences in outcome, and the mistaken clinical conclusions that they
engender could well result in the undertreatment, overtreatment, or
other mismanagement of millions of future patients worldwide.
Benefits of large-​scale randomized evidence
Hundreds of thousands of premature deaths each year could be
avoided by seeking appropriately large-​scale randomized evidence
about various widely practicable treatments for the common causes
of death, and by disseminating this evidence appropriately. The
value of large-​scale randomized evidence is illustrated by the trials
of fibrinolytic therapy for acute myocardial infarction; of antiplatelet
therapy for a variety of vascular conditions; of endocrine therapy
for early breast cancer; and of drug therapy for lowering blood
pressure. In these examples, proof of benefit that could not have
been achieved by either small-​scale randomized evidence or non-​
randomized evidence has led to widespread changes in practice that
are now preventing hundreds of thousands of premature deaths
each year, and appropriately large-​scale randomized evidence could
substantially improve the management of many important, but non-​
fatal, medical conditions.
Moderate (but worthwhile) effects on major
outcomes are generally more plausible than
large effects
Some treatments have large, and hence obvious, effects on survival:
for example, it was clear without the need for any randomized trials
that prompt treatment of diabetic coma or cardiac arrest can save
2.4
Large-​scale randomized evidence:
Trials and meta-​analyses of trials
Colin Baigent, Richard Peto, Richard Gray, Natalie Staplin, Sarah Parish, and Rory Collins
52
SECTION 2  Background to medicine
lives, and more recently the introduction of protease inhibitors for
the treatment of HIV infection led to a reduction in AIDS-​related
morbidity and mortality that was large enough to be obvious even
without randomized evidence; indeed, the remarkable effectiveness
of antiretroviral drugs can be seen from the sudden reversal, after
the mid-​1990s, of the upward trend in mortality among US men
aged 30–​34 (see Fig. 2.4.1), the chief cause of which was HIV/​AIDS.
However, perhaps in part because of these striking successes, for
the past few decades the hopes of large treatment effects on mortality
and major morbidity in many serious diseases have been unrealis-
tically high. Of course, treatments do quite commonly have large
effects on various less fundamental measures: certain drugs clearly
reduce blood pressure, blood cholesterol, or blood glucose; many tu-
mours or leukaemias can be controlled temporarily by radiotherapy
or chemotherapy; and, in acute myocardial infarction, lidocaine
(lignocaine) can prevent many arrhythmias and fibrinolytic therapy
can dissolve many thrombi. However, although such effects on inter-
mediate outcomes may be large, the net effects on mortality may be
much more modest.
In general, if substantial uncertainty remains about the efficacy of
a practicable treatment, its effects on major endpoints are probably
either negligibly small, or only moderate, rather than large. Indirect
support for this possibly pessimistic conclusion comes from many
sources, including: the previous few decades of disappointingly slow
progress in the curative treatment of common chronic diseases of
middle age; the heterogeneity of each single disease, as evidenced
by the unpredictability of survival duration even when apparently
similar patients are compared with each other; the variety of dif-
ferent mechanisms in certain diseases that can lead to death, only
one of which may be appreciably influenced by any one particular
therapy; the modest effects often suggested by meta-​analyses (see
later) of various therapies, and, in certain special cases, observa-
tional epidemiological studies of the strength of the relationship
between a particular disease and the factor that the treatment will
modify (e.g. blood pressure, blood cholesterol, or blood glucose; see
later on in this chapter).
Having accepted that only moderate reductions in mortality are
likely with many currently unevaluated interventions, how worth-
while might such effects be if they could be detected reliably? To some
clinicians, reducing the risk of early death in patients with myocar-
dial infarction from 10 per 100 patients down to 9 or 8 per 100 pa-
tients treated may not seem particularly worthwhile, and if such a
reduction was only transient, or involved an extremely expensive or
toxic treatment, this might well be an appropriate view. Worldwide,
however, several million patients a year suffer an acute myocardial
infarction, and if just one million were to be given a simple, non​toxic,
and widely practicable treatment that reduced the risk of early death
from 10% down to 9 or 8% (that is, a proportional reduction of 10
or 20%), this would avoid 10 or 20 thousand deaths. (For example,
at least a million patients a year now receive fibrinolytic therapy for
acute myocardial infarction, which is avoiding about 20 000 early
deaths a year.) Such absolute gains are substantial, and might consid-
erably exceed the number of lives that could be saved by a much more
effective treatment of a much less common disease.
Reliable detection or refutation of moderate
differences requires avoidance of both moderate
biases and moderate random errors
If realistically moderate differences in outcome are to be reliably
detected or reliably refuted, then errors in comparative assessments
of the effects of treatment need to be much smaller than the differ-
ence between a moderate, but worthwhile, effect and an effect that
is too small to be of any material importance. This in turn implies
that moderate biases and moderate random errors cannot be tol-
erated. The only way to guarantee very small random errors is to
study really large numbers, and this can be achieved in two main
ways: by making individual studies large, and by combining infor-
mation from as many relevant studies as possible in a systematic
meta-​analysis (Table 2.4.1). However, it is not much use having
very small random errors if there are moderate biases, so even the
large sizes of some non​randomized analyses of computerized hos-
pital records, where the complex factors involved in the decision to
treat a person with a particular drug may not be recorded in suffi-
cient detail, cannot guarantee medically reliable comparisons be-
tween the effects of different treatments (see later). For example,
the choice of treatment may be strongly affected by subtle patient
characteristics that are correlated with the prognosis. (A crude
1950
1960
1970
1980
1990
2000
2010
0%
0.1%
0.2%
0.3%
0.4%
0.5%
0.6%
Male
Female
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
5-year
risk
Death rate/1000
UNITED STATES 1950−2015: Males & Females
All medical mortality at ages 30−34
Source: WHO mortality &
UN population estimates
Fig. 2.4.1  Mortality trends in the United States of America among men
and women aged 30–​34 during the period 1950–​2015. Antibacterial
drugs caused a big decrease in mortality around the middle of the
century in both sexes. The increase in AIDS-​related mortality since the
early 1980s caused a sharp increase in all-​cause mortality, particularly
in men, which continued until it was spectacularly reversed by effective
antiretroviral drug combinations in the mid-​1990s.
2.4  Large-scale randomized evidence: Trials and meta-analyses of trials
53
illustration of such problems is provided by the old joke ‘What’s the
most dangerous place in the world?’ ‘Bed—​look at the number of
people who die in bed!’.)
Avoiding moderate biases
Proper randomization avoids systematic differences between the
types of patient in different treatment groups. The fundamental
reason for randomization is to avoid moderate bias, by ensuring
that each type of patient can be expected (but for the play of
chance) to have been allocated in similar proportions to the dif-
ferent treatment strategies that are to be compared. This means
that only random differences should affect the final comparisons of
outcome. Non​randomized methods, in contrast, cannot generally
guarantee that the types of patient given the new study treatment
do not differ systematically in any important ways from the types
of patient given any other treatment(s) with which the new study
treatment is to be compared. For example, moderate biases might
arise if the study treatment was novel and doctors were afraid to use
it for the most seriously ill patients, or, conversely, if they were more
ready to use it for those who were desperately ill. There may also
be other ways in which the severity of the condition differentially
affects the likelihood of being assigned to different treatments by
the doctor’s choice (or by the patient’s choice, or by any other non-​
random procedure).
It might appear at first sight that by collecting enough infor-
mation about various prognostic features it would be possible to
make some mathematical adjustments to correct for any such dif-
ferences between the types of patients who, in a non​randomized
study, receive the different treatments that are to be compared.
The ill-​conceived hope is that such methods, which are often car-
ried out on routinely collected healthcare data, might achieve
comparability between those entering the different treatment
groups, but they cannot be guaranteed to do so, and often fail ser-
iously. The difficulty is that some important prognostic factors
may be unrecorded, while others may not have been discovered
yet or be difficult to assess exactly and hence difficult (or even
impossible) to adjust for reliably. Although there are examples of
non​randomized studies in which the estimated effects of treat-
ment appear quantitatively close to those observed in analogous
randomized trials, there are many examples where they are not,
being either quantitatively incorrect—​so that drugs appear ei-
ther misleadingly promising or of misleadingly low efficacy—​or
even qualitatively incorrect, when a harmful drug might appear
effective (or vice-​versa).
The machinery of a properly randomized trial
No foreknowledge of what the next treatment will be
In a properly randomized trial, the decision to enter a patient is
made in ignorance of which of the trial treatments that patient
will, once entered, be allocated. The treatment allocation is then
made known after trial entry has been decided upon. (The purpose
of this sequence is to ensure that foreknowledge of what the next
treatment is going to be cannot affect the decision as to whether
to enter the patient; if it did, those to be allocated one treatment
might differ systematically from those to be allocated another.)
Ideally, any major prognostic features should also be irreversibly
recorded before the treatment is revealed, particularly if these are
to be used in any treatment analyses. For, if the recorded value
of some prognostic factor might be affected by knowledge of the
trial treatment allocation, then treatment comparisons within
subgroups defined by that factor might be moderately biased.
In particular, treatment comparisons just among ‘responders’
or just among ‘non​responders’ can be extremely misleading un-
less the response is assessed before treatment allocation (which
it can sometimes be, if all patients have a ‘run-​in’ period on ac-
tive treatment before randomization, partly to assess the response
to treatment and partly to exclude those who seem during this
prerandomization run-​in unlikely to participate wholeheartedly
in the main post-​randomization study).
No bias in patient management or in outcome
assessment
An additional difficulty, in both randomized and non​randomized
comparisons of various treatments, is that there might be system-
atic differences in the use of other treatments (including general
supportive care) or in the assessment of major outcomes. A non-​
randomized comparison may well suffer from moderate biases due
to such systematic differences in ancillary care or assessment, par-
ticularly if it merely involves the retrospective review of medical
records. In the context of a randomized comparison, however, it is
generally possible to devise ways to keep any such biases small. For
example, placebo tablets may be given to control-​allocated patients
and certain subjective assessments may be ‘blinded’ (although this is
less important in studies assessing mortality).
‘Intention-​to-​treat’ analyses with no
post-​randomization exclusions
Even in a properly randomized trial, unnecessary biases may be
introduced by inappropriate statistical analysis. One of the most
important sources of bias in the analysis is undue concentration on
just part of the evidence; that is to say, on ‘data-​derived subgroup
Table 2.4.1  Requirements for reliable assessment of moderate
effects: negligible biases and small random errors
NEGLIGIBLE BIASES
(i.e. guaranteed avoidance of moderate biases)
Proper randomization
(non​randomized methods might suffer moderate biases)
Analysis by allocated treatment
(including all randomized patients: ‘intention-​to-​treat’ analysis)
Chief emphasis on overall results
(no unduly data-​dependent emphasis on particular subgroups)
Systematic overview of all relevant randomized trials
(no unduly data-​dependent emphasis on particular studies)
SMALL RANDOM ERRORS
(i.e. guaranteed avoidance of moderate chance fluctuations)
Large numbers in any new trials
(to be really large, trials should be ‘streamlined’)
Systematic overview of all relevant randomized trials
(which yields the largest possible total numbers)
54
SECTION 2  Background to medicine
analyses’ (see next). Another easily avoided bias is caused by the
post-​randomization exclusion of patients, particularly if the type
(and hence prognosis) of those excluded from one treatment
group differs from those excluded from another. Therefore, one
of the fundamental statistical analyses of a trial that should be
made available is an analysis that compares all those originally
allocated one treatment (even though some of them may not
have actually received it) with all those allocated the other treat-
ment. This is sometimes referred to as an ‘intention-​to-​treat’ ana-
lysis. Additional analyses can also be reported; for example, in
describing the frequency of some very specific side effect it may
well be preferable to record its incidence only among those who
actually received the treatment. (This is because strictly random-
ized comparisons may not be needed to assess extreme relative
risks.) However, in assessing moderate effects on the main out-
come of interest such ‘on-​treatment’ analyses can be misleading,
and ‘intention-​to-​treat’ analyses are generally a more trustworthy
guide as to whether there is any real difference between the trial
treatments in their effects on long-​term outcome.
No unduly data-​dependent emphasis on results
in particular subgroups
Treatment that is appropriate for one patient may be inappropriate
for another. Ideally, therefore, what is wanted is not only an answer
to the question ‘Is this treatment helpful on average for a wide range
of patients?’, but also an answer to the question ‘For which recog-
nizable categories of patient is this treatment particularly helpful?’
However, this ideal is difficult to attain directly because the direct
use of clinical trial results to assess proportional risk reductions
in particular subgroups of patients is surprisingly unreliable. Of
course, patients who already have a very good prognosis anyway and
are at low absolute risk cannot have a large absolute benefit (for even
if a small risk is halved the absolute benefit is small). Classification
of patients as being at low (or high) risk of an adverse disease out-
come is often a reliable guide as to which patients can expect little
absolute gain even if the trial treatment works as expected (and as
to which patients might expect a worthwhile gain). This low-​risk/​
high-​risk split may not require support from formal subgroup
analyses—​indeed, it could even be damaged by such analyses. For,
even if the proportional effects of treatment in specific subgroups are
importantly different, standard subgroup analyses are so insensitive
that they may well fail to demonstrate these differences. Moreover,
even if there are highly significant differences between the propor-
tional risk reductions produced by the trial treatment in different
subgroups, and the results seem to suggest that the treatment works
in some subgroups but not in others (thereby giving the appearance
of a ‘qualitative interaction’), this may still not be good evidence for
subgroup-​specific treatment preferences. The play of chance often
produces qualitatively wrong answers in particular subgroups in
trials (or in meta-​analyses of trials) that could, if interpreted incau-
tiously, lead to millions of people being treated inappropriately, or
untreated inappropriately.
Questions about such ‘interactions’ between patient character-
istics and the effects of treatment are easy to ask, but are surpris-
ingly difficult to answer reliably. Apparent interactions can often be
produced by the play of chance and, in particular subgroups, can
mimic or obscure some of the moderate treatment effects that might
realistically be expected. To demonstrate this, a subgroup analysis
was performed based on the astrological birth signs of patients ran-
domized in the very large Second International Study of Infarct
Survival (ISIS-​2) trial of the treatment of acute myocardial infarc-
tion. Overall in this trial, the 1-​month survival advantage produced
by aspirin was conclusively demonstrated (804 vascular deaths
among 8587 patients allocated aspirin versus 1016 among 8600 al-
located no aspirin; 23% proportional reduction, two-​sided p value
<0.000001). However, when these analyses were subdivided into 12
subgroups by the patients’ astrological birth signs (in mediaeval as-
trology, the ‘birth sign’ is determined by the month of birth; for ex-
ample, ‘Libra’ means born 24 September to 23 October, and ‘Gemini’
means born 22 May to 21 June) to illustrate the unreliability of sub-
group analyses, aspirin appeared totally ineffective for those born
under Libra or Gemini (Table 2.4.2). It would obviously be unwise
to conclude from such a result that patients born under the astro-
logical birth sign of Libra or Gemini should not be given aspirin
if they have a heart attack. However, similar conclusions based on
‘exploratory’ data-​derived subgroup analyses, which from a purely
statistical viewpoint are no more reliable than these, are often re-
ported and believed, with inappropriate effects on worldwide clin-
ical practice.
There are three main remedies for this unavoidable conflict be-
tween the reliable subgroup-​specific conclusions that doctors and
patients want and need, and the statistically unreliable findings that
direct subgroup analyses can usually offer. However, the extent to
which these remedies are helpful in particular instances is one on
which informed judgements differ.
First, where there are good a priori reasons for anticipating
that the proportional effects of treatment might be very different
in different circumstances then a limited number of subgroup
analyses may be prespecified in the study protocol, along with a
prediction of the direction of such proposed interactions. (For
example, it was expected that the benefits of fibrinolytic therapy
for acute myocardial infarction would be greater the earlier such
patients were treated and so some studies prespecified that the
analyses would be subdivided by the number of hours from the
onset of symptoms to treatment:  see later.) These prespecified
Table 2.4.2  False-​negative mortality effect in a subgroup defined
only by the mediaeval astrological birth sign: the ISIS-​2 (1988)
trial of aspirin among over 17 000 patients with acute myocardial
infarction
Astrological
birth sign
No. of 1-​month deaths
(aspirin versus placebo)
Statistical
significance
Libra or Gemini
150 vs. 147
NS
All other signs
654 vs. 869
2p <0.000001
Any birth signa
804 vs. 1016
(9.4%) (11.8%)
2p <0.000001
a Appropriate overall analysis for assessing the true effect in all subgroups. Mediaeval
astrology divides birth dates into 12 ‘birth signs’ (which depend only on the day and
month of birth, not the year of birth). To demonstrate the potential unreliability of other
subgroup analyses, the ISIS-​2 patients were divided into 12 subgroups according to their
astrological birth sign, and the apparent effects of aspirin were calculated separately
in each of these 12 subgroups. Because of the play of chance, the apparent effects
differed from one subgroup to another (ranging from no apparent effect of aspirin in
two subgroups (Libra and Gemini: see text for definition) to aspirin apparently halving
mortality in another (Capricorn)).
2.4  Large-scale randomized evidence: Trials and meta-analyses of trials
55
subgroup-​specific analyses can then be taken somewhat more ser-
iously than other subgroup analyses, but they can still yield im-
portantly wrong answers.
The second approach is to emphasize chiefly the overall results of
a trial (or, better still, of all such trials) for particular outcomes, as
a guide to—​or at least a context for speculation about—​the qualita-
tive results in various specific subgroups of patients, and to give less
weight to the actual results in each separate subgroup. This is clearly
the right way to interpret the astrological ‘findings’ in Table 2.4.2,
but, if used sensibly, it is also likely in many other circumstances to
provide the best assessment of whether one treatment is better than
another in particular subgroups. The proportional effect of treat-
ment as estimated by the overall results may well provide a useful
approximation to the proportional effects of treatment in particular
subgroups.
Of course, any such extrapolation of overall results to particular
subgroups needs to be performed in a sensible way, keeping in touch
with medical realities. In particular, if a treatment has substantial
side effects, it may be obviously inappropriate for low-​risk patients.
(In this case, the side effects in a particular subgroup and the pro-
portional benefit in that subgroup should be estimated separately,
but the estimation for both might be more reliable if based on an
appropriate extrapolation from the overall results rather than on the
results in that one subgroup alone.)
The third approach is to be influenced, in discussing the likely ef-
fects on mortality in specific subgroups, not only by the mortality
analyses in these subgroups but also by the analyses of recurrence-​
free survival or some other major ‘surrogate’ outcome. For, if the
overall results are similar but much more highly significant for
recurrence-​free survival than for mortality, subgroup analyses with
respect to the former may be more stable and may provide a better
guide as to whether there are any major differences between sub-
groups in the effects of treatment.
The appropriate interpretation of apparently different results in
different subgroups of the randomized evidence is still one of the
most difficult matters of judgement in the interpretation of random-
ized evidence; at present, many clinicians and regulatory agencies
pay too much attention to irregularities in apparent effects that are
consistent with chance.
Avoiding moderate random errors
The need for large-​scale randomization
To distinguish reliably between the two alternatives that (a) there is
no worthwhile difference in survival or that (b) treatment confers a
moderate, but worthwhile, benefit (e.g. 10 or 20% fewer deaths), not
only must systematic errors be guaranteed to be small (see earlier)
compared with such a moderate risk reduction, but so too must
any of the purely random errors that are produced just by chance.
Random errors can be reliably avoided only by studying very large
numbers of patients and hence large enough numbers of ‘endpoints’.
However, it is not sufficiently widely appreciated just how large clin-
ical trials need to be in order to detect moderate differences reli-
ably. This can be illustrated by a hypothetical trial that is actually
quite inadequate—​even though by some standards it is moderately
large—​in which a 20% reduction in mortality (from 10 to 8%) is sup-
posed to be detected among 2000 heart attack patients (1000 treated
and 1000 controls). In this case, one might predict about 100 deaths
(10%) in the control group and 80 deaths (8%) in the treated group.
However, if this difference were to be observed it would not be con-
ventionally significant (p = 0.1); indicating that even if there is no
real difference between the effects of the trial treatments, it would
still be relatively easy for a result at least as extreme as this to arise
by chance alone. Although the play of chance might well increase
the difference enough to make it conventionally significant (e.g. 110
deaths vs. 70 deaths, 2p <0.001), it might equally well dilute, oblit-
erate (e.g. 90 deaths vs. 90 deaths), or even reverse it. The situation in
real life is often even worse, as the average trial size may include only
a few dozen events rather than the several hundred (or few thou-
sand) that would ideally be needed to guide the future treatment of
millions.
Mega-​trials: How to randomize large numbers
One of the chief techniques for obtaining appropriately large-​scale
randomized evidence is to make trials extremely simple, and then
to invite hundreds of hospitals to collaborate. The first of these
large streamlined trials (or mega-​trials) were the ISIS and GISSI
studies of heart attack treatment in the 1980s, and many other
mega-​trials have now been successfully undertaken, not only in the
field of cardiology—​where numerous large trials have now been
performed—​but also in other specialties where treatment might be
expected to have only moderate effects on morbidity and mortality
from a common disease or injury. Many such mega-​trials have pro-
duced medically important results that would not otherwise have
been reliably obtained. However, in terms of medically significant
findings, what has been achieved so far is only a fraction of what
would be possible if this research strategy could be more widely
adopted. Any obstacle to simplicity is an obstacle to large size, and
so it is worth making enormous efforts at the design stage to sim-
plify and streamline the process of entering, treating, and assessing
patients. Many trials would be of much greater scientific value if
they collected 10 times less information, both at entry and during
follow-​up, on 10 times more patients. Since those responsible for
entering patients into trials are generally busy people, it is particu-
larly important to simplify the entry of patients, otherwise rapid re-
cruitment may prove difficult (see later). Likewise, when allocating
resources within large-​scale trials, it is important to direct them
to where it chiefly matters, namely the recruitment of large num-
bers of patients and counting how many suffer the main outcomes
of interest, whereas large sums of money may often be wasted on
inappropriate audits or unnecessary or excessively frequent meas-
urements, the analysis of which will not contribute to answering the
main study question.
Simplification of entry procedures for trials:
The ‘uncertainty principle’
For ethical reasons, patients cannot have a commonly available
treatment chosen at random if either they or their doctor are (for
any reasons) already reasonably certain that another treatment is
preferable. Hence, randomization can be offered only if both doctor
and patient feel substantially uncertain as to which of the trial op-
tions is best. The question then arises: ‘Which categories of patients
56
SECTION 2  Background to medicine
about whose treatment there is such uncertainty should be offered
randomization?’ The obvious answer is all of them, welcoming the
heterogeneity that this will produce. (For example, either the treat-
ment of choice will turn out to be the same for men and women, in
which case the trial might as well include both, or it will be different,
in which case it is particularly important to study both.) In appropri-
ately large trials, patient homogeneity is generally a defect while het-
erogeneity is generally a strength. Consider, for example, the trials
of immediate fibrinolytic therapy for acute myocardial infarction.
Some had restrictive entry criteria that allowed inclusion of only
those patients who presented between 0 and 6 h after the onset of
pain, and those trials contributed almost nothing to the key question
of how late such treatment can still be useful. In contrast, the trials
with wider and more heterogeneous entry criteria that included
some patients with somewhat longer delays between pain onset and
hospitalization answered this question prospectively, and were able
to show that fibrinolytic therapy can have definite protective effects
not only when given 0 to 6 but also when given 7 to 12 hours after the
onset of pain (see later).
This approach of randomizing the full range of patients in whom
there is substantial uncertainty as to which treatment option is
best was used in the first Medical Research Council Asymptomatic
Carotid Surgery Trial (ACST-​1). Narrowing of the carotid artery
(which is rapidly detectable by ultrasound) can eventually cause a
stroke, or even a succession of strokes. It can be dealt with surgi-
cally by carotid endarterectomy, but in the 1990s there was much
uncertainty as to whether such surgery, with its inherent periopera-
tive risks, was appropriate for individuals with severe carotid artery
narrowing who were currently asymptomatic (that is, had not had a
stroke in the past few months). The ACST was therefore designed to
compare a policy of immediate carotid endarterectomy with a policy
of ‘watchful waiting’ in asymptomatic patients with substantial ca-
rotid artery narrowing. If a patient was prepared at least to consider
surgery seriously, then the neurologist and surgeon responsible for
that individual’s care considered in their own undefined way what-
ever medical, personal, or other factors seemed to them to be rele-
vant, including, of course, the patient’s own preferences and values.
Eligibility for randomization was defined by the ‘uncertainty prin-
ciple’ (Fig. 2.4.2):
1.	 If they or the patient were reasonably certain, for any reason,
that they did wish to recommend immediate surgery for that
particular patient, the patient was not eligible for entry into
the ACST.
2.	 Conversely, if they or the patient were reasonably certain, for any
reason, that they did not wish to recommend immediate surgery,
the patient was likewise not eligible for entry into the trial.
3.	 If, but only if, the doctor(s) and patient were substantially
uncertain what to recommend, the patient was automatically
eligible for randomization between immediate versus no im-
mediate surgery (with all patients receiving whatever their
doctors judged to be the best available medical care, which
generally included advice to stop smoking, low-​dose aspirin,
and treatment of hypertension; nowadays, it would probably
also include a statin).
In ACST-​1, there were substantial differences between individual
doctors in the types of patients about whom they were uncertain
(in terms of the severity of carotid stenosis—​which was generally
recorded on ultrasound as 70%, 80%, or 90% blockage—​age, gen-
eral health, and various other characteristics). This guaranteed that
no category of patient about which there was widespread uncer-
tainty would be wholly excluded, and hence guaranteed that the trial
would yield at least some direct evidence in a variety of typical pa-
tients. As a result of the wide and simple entry criteria adopted by
ACST-​1, 3120 patients were randomized (which was more than in
any previous vascular surgery trial), so the study was able to provide
some clear answers about who needed carotid endarterectomy. In
asymptomatic patients younger than 75 years of age, with carotid
diameter about 70% or more on ultrasound, immediate carotid end-
arterectomy halved the net 5-​year stroke risk from about 12% to 6%
(even though this 6% included the 3% perioperative hazard). For
patients with only moderate carotid artery stenosis on ultrasound,
the 5-​year risks of carotid stroke (excluding perioperative hazard)
were 2% versus 9%, while among those with tighter stenosis the risks
were 3% versus 10%, suggesting about as much benefit in moderate
as in tight stenosis.
The ‘uncertainty principle’ simultaneously meets the require-
ments of ethicality, heterogeneity, simplicity, and maximal trial
size, and should be widely used. It states that the fundamental
eligibility criterion is that both doctor and patient should be
substantially uncertain about the appropriateness of each of the
trial treatments for that particular patient. With such uncertainty
as the fundamental criterion of eligibility, informed consent can
often be simplified. For, the degree of ‘informed consent’ that is
appropriate in a randomized comparison of two established treat-
ments governed by the ‘uncertainty principle’ should probably
not differ greatly from that which is applied in routine practice
outside trials when treatment is being chosen haphazardly—​or, to
put it another way, ‘double standards’ between trial and non​trial
situations are inappropriate. The haphazard nature of many non-​
randomized treatment choices is reflected in the wide variations
in practice between and within countries. Even when a practice is
similar it may be similarly wrong: for example, before the ISIS-​2
results became available (see later), few doctors routinely used fi-
brinolytic therapy for acute myocardial infarction. Provided that
trials are governed by the ‘uncertainty principle’, there is an ap-
proximate parallel between good science and good ethics. Indeed,
in such circumstances, excessively detailed consent procedures
(which can be distressing and inhumane, and so would not be con-
sidered appropriate in routine non​trial clinical practice) would
not be humane or ethically appropriate in trials. Excessively de-
tailed consent procedures are, unfortunately, quite common, but
their chief purpose is to protect doctors against lawyers rather than
to protect patients against anything.
This ‘uncertainty principle’ is just one of many ways of
simplifying trials and thereby helping them to avoid becoming
enmeshed in a mass of wholly unnecessary traditional complexity.
If randomized trials can be substantially simplified (which, it
must be admitted, requires a reversal of the current trend towards
unnecessary complexity), as has already been achieved for a few
major diseases, and hence made very much larger, then they will
continue to play an appropriately central role in the development
of rational criteria for planning treatment strategies and reducing
death and disability.
2.4  Large-scale randomized evidence: Trials and meta-analyses of trials
57
Minimizing both bias and random
error: Meta-​analyses of randomized trials
Archie Cochrane was one of the first people to emphasize the need
to organize, by specialty, the results from all relevant randomized
trials, and the Cochrane Library brings together in a single place a
large number of systematic reviews (many of which include meta-​
analyses of randomized trials) summarizing the available evidence
about a wide range of therapeutic questions. When several trials
have all addressed much the same question, the traditional pro-
cedure of only a few of them becoming widely known may be a
source of serious bias, since chance fluctuations for or against treat-
ment may affect which trials become well known and widely cited.
To avoid this problem, it is appropriate to base inference chiefly on
a meta-​analysis of all the results from all of the trials that have ad-
dressed a particular type of question (or on an unbiased subset of
such trials), and not on some potentially biased subset of these trials.
Carotid artery stenosis detected by ultrasound,
but, as yet, no clinical evidence of stroke from it.
Should patient be offered immediate carotid surgery?
Doctor(s) or patient
reasonably certain
(no matter why) that immediate
surgery is not appropriate:
Ineligible
Doctor(s) or patient
reasonably certain
(no matter why) that immediate
surgery is appropriate:
Ineligible
Doctor(s) and patient substantially uncertain
whether to risk immediate surgery:
Uncertainty implies eligibility
Telephone to randomize
Group 1: Allocated
No immediate surgery
(unless or until a clear indication is
thought to have arisen)
Group 2: Allocated
Immediate carotid surgery
(unless definite contraindication is
thought to have been discovered, or
patient [or doctor] changes their mind)
Over the next few years
only a small number get
carotid surgery
90% get
carotid surgery
(median delay: 1 month)
Statistical comparisons of various outcomes over 5 to 10 years:
100% of Group 1 vs. 100% of Group 2,
i.e. ‘intention to treat analysis’
Fig. 2.4.2  Example of the ‘uncertainty principle’ to define eligibility for trial entry: the chief
eligibility criterion for the Asymptomatic Carotid Surgery Trial (ACST) was that doctors and patients
should be substantially uncertain whether to risk immediate carotid surgery. Partly because this
criterion was appropriately flexible, ACST-​1 became the largest-​ever trial of vascular surgery,
showing that the long-​term benefits of carotid artery surgery could eventually outweigh the
immediate hazards. ACST-​2 (http://​www.nds.ox.ac.uk/​acst) is now randomizing surgery versus
carotid stenting where the doctor(s) and patient are substantially uncertain which to prefer.
58
SECTION 2  Background to medicine
Such meta-​analyses will also minimize random errors in the assess-
ment of treatment since, in general, far more patients are involved in
a meta-​analysis than in any contributory individual trial.
The separate trials may well be heterogeneous in their entry cri-
teria, their treatment schedules, their follow-​up procedures, their
methods of treating relapse, and so on. In view of this heterogeneity,
at one extreme each trial might be considered in virtual isolation
from all others, while at the opposite extreme the results from all
trials could be combined, largely ignoring any heterogeneity. Both
these extreme views have some merit, and the pursuit of each by
different people may prove more illuminating than too definite an
insistence on any one particular approach. However, the heterogen-
eity of the different trials merely argues for careful interpretation
of any meta-​analyses of different trial results, rather than arguing
against meta-​analyses. Whatever the difficulties in interpreting
meta-​analyses may be, without them it is difficult to avoid moderate
selective biases and substantial random errors, both of which could
obscure any moderate treatment effects, or, conversely, imply an ef-
fect where none exists.
Which meta-​analyses are trustworthy?
Since the 1970s, a rapidly increasing number of meta-​analyses of the
results of randomized trials have been reported, not all of which are
trustworthy. When considering how reliable a given one might be
there are two fundamental questions: What is the potential for bias,
and what is the potential size of purely random errors? To answer the
first question, consideration must be given to whether biases might
exist within individual trials (e.g. because of an unreliable method
of randomization or because of post-​randomization exclusions from
the main analyses), and whether the subset of trials under consid-
eration might be a biased subset of all relevant trials that have been
performed (as might arise, for example, if certain trials were aban-
doned because of unpromising findings, or remained unpublished
for this or any other reason).
The simplest approach to meta-​analysis is merely to have collected
and tabulated the published data from whatever randomized trial
reports are easily accessible in the literature, and sometimes this may
suffice. At the opposite extreme, extensive efforts may have been
made by those organizing the meta-​analysis to locate every poten-
tially relevant randomized trial, including those never published, to
collaborate closely with the trialists to seek individual data on each
patient ever randomized into those trials, and then (after extensive
checks and corrections of such data) to produce, in collaborative re-
analyses with those trialists, agreed analyses and publications. The
results of some of the largest such collaborative re-​analyses will be
described later: the Anti Thrombotic Trialists’ (ATT) Collaborative
Group, the Fibrinolytic Therapy Trialists’ (FTT) Collaborative
Group, and the Early Breast Cancer Trialists’ Collaborative Group
(EBCTCG). Collaboration of the original trialists in the meta-​
analysis process, with collection of detailed data from each indi-
vidual trial participant, can help to avoid or minimize the biases that
could be produced by missing trials (e.g. owing to the greater likeli-
hood of extremely good, or extremely bad, results being particularly
widely known and published), by inappropriate post-​randomization
withdrawals, or by the failure to allocate treatment properly at
random. If randomization was performed properly in the first place,
the post-​randomization withdrawals can often be followed-​up and
restored to the study for an appropriate ‘intention-​to-​treat’ analysis.
Knowledge of the exact methods of treatment allocation (backed up
by checks on whether the main prognostic factors recorded are non-​
randomly distributed between the treatment groups in a particular
trial) may help to identify trials that were so improperly randomized
that they should be excluded from a meta-​analysis of the properly
randomized trials. Meta-​analyses based on individual patient data
may also provide more information about treatment effects than the
more usual overviews of grouped data, for they allow more detailed
analyses—​indeed, if they are really large then they may actually yield
statistically reliable subgroup analyses of the effects of treatment in
particular types of patient.
Conversely, even a perfectly conducted meta-​analysis of an inter-
vention with moderate effects on a major clinical outcome may
not be reliable if the trials were all small. There are two reasons for
this. First, when the true effect of an intervention is only moderate,
most small trials will fail to reach statistical significance, and may
be less likely to be published (or otherwise available) than the few
with results that are misleadingly extreme. Hence, a meta-​analysis
consisting exclusively of small trials is particularly prone to bias.
Secondly, the random errors may be too large to allow reliable in-
terpretation. A meta-​analysis that includes a total of only 100 deaths
will have random errors about as great as a single trial with only
100 deaths. For these reasons small-​scale evidence, whether from a
meta-​analysis or from one trial, is often unreliable and may well be
found in retrospect to have yielded wrong answers. What is needed
is large-​scale randomized evidence; it does not matter much whether
the totality of the evidence comes from a properly conducted meta-​
analysis of several trials or one properly conducted trial with such
clear results that no further trials were done. The practical medical
value of large-​scale randomized evidence will be illustrated by a few
examples.
Examples of information about the efficacy
and safety of widely used drugs that could have
been reliably established only by large-​scale
randomized evidence
Evidence of benefit from a single very large trial
In the ISIS-​2 trial, half of 17 000 patients with suspected acute
myocardial infarction were allocated aspirin tablets (162 mg/​day
for 1 month, which virtually completely inhibits cyclo-​oxygenase-​
dependent platelet activation) and half were allocated placebo tab-
lets. Before 1988, when the ISIS-​2 results were published, aspirin was
not routinely used in the treatment of acute myocardial infarction,
and no other major trial had (or has subsequently) compared aspirin
with an untreated control group in cases of suspected acute myocar-
dial infarction. However, the effects of 1 month of aspirin were so
definite in ISIS-​2 (804/​8587 vascular deaths among those who were
allocated aspirin versus 1016/​8600 among those who were not) that
even the lower 99% confidence limit would have represented a worth-
while benefit from this simple and inexpensive treatment (Fig. 2.4.3).
As a result, worldwide treatment patterns changed sharply when
the ISIS-​2 results emerged in 1988, and aspirin is now routinely used
for the majority of emergency hospital admissions with suspected
acute myocardial infarction not only in Europe and America but
throughout Asia. In the United Kingdom, for example, two British
2.4  Large-scale randomized evidence: Trials and meta-analyses of trials
59
Heart Foundation surveys found cardiologists reporting that rou-
tine aspirin use in acute coronary care had increased from under
10% in 1987 to over 90% in 1989. Worldwide, the annual number of
patients with suspected myocardial infarction who would nowadays
be given such treatment must be several million a year, suggesting
that aspirin is already preventing several tens of thousands of prema-
ture deaths each year in this clinical context alone. However, if the
ISIS-​2 trial had been a factor of 10 smaller (i.e. 1700 instead of 17 000
patients), then exactly the same proportional reduction in mortality
as shown in Fig. 2.4.3 would not have been conventionally signifi-
cant and, therefore, would have been much less likely to influence
medical practice—​indeed, the result might by chance have appeared
exactly flat, greatly damaging future research on aspirin in this con-
text. (In fact, during the early interim monitoring of the ISIS-​2 trial
results by the independent Data Monitoring Committee there was
no apparent difference in mortality based on a few hundred deaths.)
Likewise, if the ISIS-​2 trial had been non​randomized, then it might
well have produced the wrong answer since, in a nonrandomized
study, doctors might tend to give active treatment to patients who
are particularly ill, or who are rather different in various other ways
from those not given active treatment. In addition, even if a non-​
randomized study did happen to produce an unbiased answer, it
would have been impossible to be sure that it had actually done so,
so again a non​randomized study might have had much less influence
on medical practice than ISIS-​2.
In the ISIS-​2 trial, aspirin significantly reduced the 1-​month
mortality, but it also significantly reduced the number of non​fatal
strokes and non​fatal reinfarctions that were recorded in hos-
pital. Combining all these three outcomes into ‘vascular events’
(i.e. stroke, death or reinfarction), 10% of those who were allo-
cated aspirin versus 14% of the controls suffered a vascular event
in the month after randomization (Table 2.4.3)—​an absolute dif-
ference of 40 events per 1000 treated (or, perhaps more relevantly,
40 000 per million). The randomized trials of aspirin, or of other
antiplatelet regimens, in other types of high-​risk patients (e.g. a few
years of aspirin for those who have survived a myocardial infarc-
tion or stroke) were not as large as ISIS-​2, and so, taken separately,
most yielded false-​negative results. However, when the results from
many such trials are combined, statistically definite reductions in
‘vascular events’ are seen (Table 2.4.3). Since such treatments do
not appear to increase non​vascular mortality, all-​cause mortality
Cumulative number of vascular deaths
1000
800
600
400
200
0
Days from randomization
0
7
14
21
28
35
Placebo tablets:
1016 vascular
deaths (11.8%)
Aspirin:
804 vascular
deaths (9.4%)
Fig. 2.4.3  Effect of administration of aspirin for 1 month on 35-​day
mortality in the 1988 ISIS-​2 trial among 17 000 patients with acute
myocardial infarction. (Absolute survival advantage: 24 (SE 5) lives saved
per 1000 patients allocated aspirin, 2p <0.00001. The COMMIT trial in
46 000 such patients has since shown aspirin plus clopidogrel to be
slightly more effective than aspirin alone.)
Table 2.4.3  Summary results of (a) trials of aspirin (or other antiplatelet drugs), and (b) trials of adding a
P2Y12-​receptor antagonist to aspirin
Type of patient (and study)
Mean duration
(total randomized)
Stroke, heart attack, or vascular death
(a) Antiplatelet vs. control
Antiplatelet
Control
Difference
Acute heart attack
(ISIS-​2)
1 month
(20 000)
10%
14%
40 per 1000
(2p <0.00001)
Acute stroke
(CAST)
1 month
(40 000)
9%
10%
10 per 1000
(2p = 0.001)
Previous heart attack
(ATT)
2 years
(20 000)
13%
17%
40 per 1000
(2p <0.00001)
Previous stroke/​TIA
(ATT)
2.5 years
(23 000)
18%
22%
40 per 1000
(2p <0.00001)
Other high-​risk (e.g. angina,
peripheral vascular disease)
1 year
(20 000)
8%
10%
20 per 1000
(2p <0.00001)
(b) Aspirin plus P2Y12-​receptor
antagonist vs. aspirin alone
Aspirin + P2Y12-​receptor
antagonist
Aspirin alone
Acute coronary syndrome
(CURE)
9 months
(13 000)
9%
11%
20 per 1000
(2p <0.001)
Acute heart attack
(CCS2-​COMMIT)
1 month
(46 000)
9%
10%
10 per 1000
(2p = 0.002)
Previous heart attack
(PEGASUS-​TIMI 54)
33 months
(21 000)
8%
9%
10 per 1000
(2p <0.001)
60
SECTION 2  Background to medicine
is also significantly reduced. Subsequently, a combination of aspirin
and a P2Y12-​receptor antagonist such as clopidogrel (which inhibits
platelet activation through different pathways) has been shown to
be more effective than aspirin alone in acute myocardial infarc-
tion or acute coronary syndrome (Table 2.4.3), and this has paved
the way for ‘dual antiplatelet therapy’ to become standard for such
conditions.
The large-​scale randomized evidence on antiplatelet drugs that
is summarized in Table 2.4.3 has changed clinical practice world-
wide, and may already have affected the treatment of hundreds of
millions of patients in ways that, at low cost, have prevented millions
of strokes, heart attacks, or vascular deaths. Small randomized trials
and small meta-​analyses of trials, or non​randomized studies (how-
ever large), could not possibly have provided appropriately reliable
evidence about such moderate risk reductions.
Evidence of hazard from a single large trial
It is not sufficiently appreciated that identical arguments about the
need for large-​scale randomized evidence apply equally when there
is the possibility that a treatment is the cause of a moderate increase
in adverse outcomes. The drug niacin decreases blood low-​density
lipoprotein (LDL) cholesterol and triglycerides, while increasing
high-​density lipoprotein (HDL) cholesterol. It was shown to be ef-
fective for the prevention of coronary events in the Coronary Drug
Project conducted in the 1970s, but a more recent trial (AIM-​HIGH)
had been stopped early because of a lack of efficacy. Many patients
are unable to take niacin because it causes uncomfortable flushing
and other adverse effects.
The second Heart Protection Study—​Treatment of HDL to
Reduce the Incidence of Vascular Events (THRIVE) was designed
to assess the benefits and hazards of extended-​release niacin in
­combination with the drug laropiprant, which had been shown
to reduce flushing in up to two-​thirds of patients and hence, it
was hoped, would improve adherence. The trial randomized over
25 000 patients to niacin-​laropiprant versus placebo and followed
them for about 4 years. There was no significant effect on major vas-
cular events, defined in this trial as a major coronary event (non​fatal
­myocardial infarction or coronary death), stroke of any type, or
coronary or non​coronary revascularization (13.2% vs. 13.7%, rate
ratio 0.96, 95% confidence interval 0.90–​1.03, p = 0.29), but the trial
also provided very important information about several adverse ef-
fects of niacin. Table 2.4.4 summarizes the results from the trial for
selected serious adverse events (which, essentially, means that the
event caused hospitalization or, rarely, death).
While some of these were known adverse effects of niacin (e.g.
diabetes-​related, gastrointestinal, musculoskeletal, and skin-​related
disorders), other adverse events that were in excess in the THRIVE
trial had not been suspected. Allocation to niacin-​laropiprant re-
sulted in moderately raised risks of serious infections (22% rela-
tive increase, corresponding to an absolute excess of 14 cases per
1000 participants treated with niacin-​laropiprant for 4 years) and of
bleeding (38% relative increase, corresponding to an absolute excess
of 7 cases per 1000 participants treated with niacin-​laropiprant for
4 years). A significant excess of serious infections was also seen with
niacin alone in the smaller AIM-​HIGH trial of 3500 patients (8.1%
vs. 5.8, p = 0.008), but there were relatively few serious bleeding
events and the excess observed with niacin alone was not significant
(3.4% vs. 2.9%, p = 0.36), although it is consistent with the THRIVE
result. The discovery of these new adverse effects in THRIVE, to-
gether with the new information about the magnitude of known
adverse effects, has led to a reappraisal of the role of niacin in the
management of blood lipids. For not only is niacin now known to be
ineffective at preventing vascular events (at least in circumstances
similar to those tested in THRIVE), but it is also a much more haz-
ardous drug than had previously been appreciated. This illustrates
the value of large-​scale randomized evidence for studying drug
safety when, as is typical, the hazards requiring detection are of only
moderate magnitude.
Definite result from a very large meta-​analysis
of trials: Benefit from ‘adjuvant’ therapy with
tamoxifen for patients with hormone-​sensitive (ER+)
‘early’ breast cancer
By definition, in ‘early’ breast cancer all detectable deposits of dis-
ease are limited to the breast and the local or regional lymph nodes,
and can be removed surgically. However, experience shows that
undetectably small deposits of breast cancer cells may remain else-
where that eventually cause clinical recurrence at a distant site, per-
haps after a delay of several years, which is then usually followed by
death from disease. If the original tumour was ‘ER positive’ (i.e. if
the tumour cells were still expressing the oestrogen receptor pro-
tein) then the distant deposits of cancer cells that spread from it be-
fore it was removed may also be ER positive, and may be continually
stimulated by circulating hormones. Therefore, among women who
Table 2.4.4  Summary results of the HPS2-​THRIVE trial: effects of niacin-​laropiprant on selected serious adverse events and diabetes
Event type
Niacin-​laropiprant
(n = 12 800)
Placebo
(n = 12 800)
Rate ratio
(95% confidence interval)
Absolute excess
(%) ± SE
Serious adverse event (%)
Gastrointestinal event
4.8%
3.8%
1.28 (1.13–​1.44)
1.0 ± 0.3
Musculoskeletal event
3.7%
3.0%
1.26 (1.10–​1.44)
0.7 ± 0.2
Skin-​related event
0.7%
0.4%
1.67 (1.20–​2.34)
0.3 ± 0.1
Infection event
8.0%
6.6%
1.22 (1.12–​1.34)
1.4 ± 0.3
Bleeding event
2.5%
1.9%
1.38 (1.17–​1.62)
0.7 ± 0.2
New-​onset diabetes
5.7%
4.3%
1.32 (1.16–​1.51)
1.3 ± 0.3
Adapted from The New England Journal of Medicine, The HPS2-​THRIVE Collaborative Group, Effects of Extended-​Release Niacin with Laropiprant in High-​Risk Patients, 371, 210.
Copyright © 2014 Massachusetts Medical Society. Reprinted with permission.
2.4  Large-scale randomized evidence: Trials and meta-analyses of trials
61
have had breast cancer removed by surgery (or by surgery and radio-
therapy), there have been many trials of ‘adjuvant’ daily treatment
with tamoxifen, a drug that blocks the oestrogen receptor. Some in-
volved only 1 to 2 years of treatment, some involved about 5 years,
some compared 5 versus 1 to 2 years and more recent trials have
compared 10 versus 5 years of tamoxifen. In total, more than 100 000
women have been randomized in several dozen such trials, some of
which are still being followed-​up for long-​term outcomes.
Taken separately, most of these tamoxifen trials have been
too small to provide reliable evidence about long-​term survival.
However, if the results of all of them are combined in various ways,
some very definite differences emerge: 1 to 2 years of tamoxifen is
better than nothing, 5 years is better than 1 to 2 years, and 10 years
is better still for delaying or avoiding the recurrence of ER positive
breast cancer. Meta-​analysis has also established that newer endo-
crine therapies (such as aromatase inhibitors) reduce recurrence
rates even further, although, unlike tamoxifen, they can only be used
in postmenopausal women. Fig. 2.4.4a shows the results from the
trials of about 5 years of tamoxifen. Allocation to active treatment
produces a 13% absolute difference in the 15-​year risk of recurrence
(34 vs. 47%), and a 9% absolute difference in survival (25 vs. 34%;
both 2p <0.00001). Most of the effect on recurrence is seen during
the first 5 years, while tamoxifen was still continuing to be given,
but most of the effect on breast cancer mortality comes after this
period (Fig. 2.4.4a). Indeed, the difference in the 15-​year probability
of death from breast cancer is about three times as great as that seen
after 5 years. Reliable assessment of the moderate improvements in
long-​term survival in early breast cancer that are produced by tam-
oxifen (and by radiotherapy and chemotherapy) would have been
impossible without such a meta-​analysis of all trials, with updated
follow-​up data provided periodically, because each of the trials was
too small on its own to answer these questions convincingly.
Collaborative meta-​analyses that involve hundreds of trialists
from all around the world can also foster international acceptance of
the totality of the randomized evidence. In the case of breast cancer
this has, since the mid-​1980s, helped lead to widespread adoption
in many countries of a succession of improvements in treatment
(earlier detection, better local control, progressively better chemo-
therapy and, in ER-​positive disease, progressively better endocrine
therapy). These have, in aggregate, resulted in a sustained fall in na-
tional mortality rates (Fig. 2.4.4b).
Promising meta-​analysis of small trials confirmed
by large trials: benefit from fibrinolytic therapy
in acute myocardial infarction
If a recent thrombus has just blocked a coronary artery, thereby
causing acute myocardial ischaemia or infarction, fibrinolytic drugs
(such as streptokinase or tissue plasminogen activator) can some-
times rapidly dissolve the thrombus, restoring the flow of blood
and reperfusing the heart muscle. These drugs were first introduced
into clinical research in the late 1950s, but the trials of fibrinolytic
therapy for suspected acute myocardial infarction in the 1960s
and 1970s were too small to be statistically reliable (none involved
even 1000 patients). So, by the early 1980s the haemorrhagic side
effects were obvious, the benefits had not been convincingly dem-
onstrated, and such treatments were generally considered to be
definitely dangerous, probably fairly ineffective, and hence inappro-
priate for routine coronary care. Although meta-​analyses published
in the mid-​1980s of the previous small trials (which had involved a
total of only c.6000 patients in 24 trials) indicated a statistically def-
inite benefit, they were not really believed by cardiologists and so
such treatments were still not widely used. The situation was saved
by two large randomized trials, GISSI-​1 and ISIS-​2, which together
involved about 30 000 patients (and by their aggregation with the
seven other randomized trials that each involved more than 1000
patients, yielding a total of 60 000; see next). In ISIS-​2, not only were
patients randomly allocated to receive aspirin or placebo tablets as
described earlier (Fig. 2.4.3), but they were also separately allocated
to receive intravenous streptokinase or a placebo infusion. In this
‘factorial’ design (which allows the separate assessment of more than
one treatment without any material loss in the statistical reliability of
each comparison), one-​quarter of the patients were allocated aspirin
alone, one-​quarter were allocated streptokinase alone, one-​quarter
were allocated both streptokinase and aspirin, and one-​quarter were
allocated neither (that is, they were given placebo tablets and a pla-
cebo infusion). Streptokinase, like aspirin, produced a highly signifi-
cant reduction in mortality, and the combination of streptokinase
and aspirin was highly significantly better than either aspirin alone
or streptokinase alone (Fig. 2.4.5).
The results shown in Fig. 2.4.5 might suggest that there was no
need to collect any more randomized evidence about fibrinolytic
therapy, but this ignores the potential hazards of such treatment
and the heterogeneity of patients. Taken separately, even ISIS-​2
(the largest of these trials) was not large enough for statistically re-
liable subgroup analyses, but when the nine largest trials were all
taken together, they included a total of about 60 000 patients, half
of whom had been randomly allocated fibrinolytic therapy. Those
entering a coronary care unit with a diagnosis of suspected or def-
inite acute myocardial infarction range from patients who are al-
ready in cardiogenic shock with low blood pressure and a fast pulse
(half of whom will die rapidly) to those who have merely had a
history of chest pain and no very definite changes on their electro-
cardiography (ECG) (of whom ‘only’ a small percentage will die be-
fore discharge). Fibrinolytic therapy often causes blood pressure to
fall: should it be used in patients who are already dangerously hypo-
tensive? It occasionally causes serious strokes: Should it be used in
patients who are elderly or hypertensive, and therefore already have
an above-​average risk of cerebral haemorrhage (or who have only
slight changes on their ECG, and therefore have only a low risk of
cardiac death)? Finally, if a coronary artery has been totally occluded
for long enough, the heart muscle that it supplies will have been ir-
reversibly destroyed: How many hours after the heart attack starts is
fibrinolytic treatment still worth risking—​3? 6? 12? 24?
These questions needed to be answered reliably before appropriate
and generally accepted indications for and against such an immedi-
ately hazardous, but potentially effective, therapy could be devised.
To address them, the main fibrinolytic therapy trialists collaborated
in a systematic meta-​analysis of the randomized evidence, based on
individual patient data. On review of the 60 000 patients random-
ized between fibrinolytic therapy and control in trials of more than
1000 patients, some of the therapeutic questions were relatively easy
to answer satisfactorily. For example, it appears that most of those
whose ECG is still normal (or shows a pattern that indicates only a
small immediate risk of death) can be left untreated, leaving open
the option of starting fibrinolytic treatment urgently if their ECG
changes suddenly for the worse over the next few hours or days.
62
SECTION 2  Background to medicine
≈ 5 years tamoxifen vs. Not
RECURRENCE
ER+
(a)
50
Control
47·3%
33·9%
≈ 5 years
tamoxifen
25·0%
≈ 5 years
tamoxifen
Control
34·3%
15–y gain 13·4% (SE 1·1)
Logrank 2p < 0·00001
15–y gain 9·3% (SE 1·1)
Logrank 2p < 0·00001
40
30
20
10
0
5
16·6
26·5
8·9
12·2
26·2
18·6
41·0
29·1
10
15
years
0
5
10
15
years
%
± SE
0
Recurrence
50
40
30
20
10
%
± SE
0
Breast cancer mortality
≈ 5 years tamoxifen vs. Not
BREAST CANCER MORTALITY
ER+
UK and USA, 1950–2014:
Breast cancer mortality at ages 45–54
(b)
Source: WHO mortality &
UN population estimates
Death rate/10 000 women, age standardized*
Further MODERATE effects
are still worthwhile and
achievable
LARGE effect on UK/USA
breast cancer mortality
by combining several
MODERATE effects.
1950
1970
1990
2010
0
2
4
6
UK
USA
0.6%
0.4%
0.2%
0%
10-year
risk
Mean of annual rates in the two
component 5–year age groups
smoothed in 5-year calendar periods
Fig. 2.4.4  (a) Effects of about 5 years of tamoxifen versus not in ER-​positive disease: 15-​year probabilities of recurrence and of breast
cancer mortality (10 000 women in the 2005 worldwide EBCTCG meta-​analysis). (b) Female breast cancer mortality in the United Kingdom
and United States of America at ages 45 to 54 during the period 1950 to 2014. (The UK breast screening programme has little effect on
mortality at these ages.)
2.4  Large-scale randomized evidence: Trials and meta-analyses of trials
63
Conversely, among those who already had ‘high-​risk’ ECG changes
when they were randomized, the absolute benefit of immediate fi-
brinolytic therapy was, if anything, slightly greater than is indicated
by Fig. 2.4.5. Age, sex, blood pressure, heart rate, diabetes, and a pre-
vious history of myocardial infarction could not identify reliably any
subgroup that would not, on average, have their chances of survival
appreciably increased by treatment.
By contrast, the longer that fibrinolytic treatment for such pa-
tients was delayed, the less benefit it seemed to produce. Among
the 45 000 whose ECG showed definite ST-​segment elevation or
bundle-​branch block, the benefit was greatest (about 30 lives saved
per 1000) among those randomized between 0 and 6 h after the
onset of pain (Fig. 2.4.6).
However, the mortality reduction was still substantial and signifi-
cant (about 20 per 1000; 2p <0.003) for the patients whose hospital
admission had been delayed for some hours and who were therefore
randomized 7–​12 h after the onset of pain. Indeed, even if patients
were randomized 13–​18 h after the onset of pain, there still appeared
to be some net reduction in mortality (about 10 per 1000, but not
statistically definite). The regression line in Fig. 2.4.6 reinforces
these separate subgroup analyses in a more reliable way. Yet, before
these large trials it was forcefully, but mistakenly, argued that such
treatments could not possibly be of any worthwhile benefit if given
more than about 3 or 4 h after the onset of symptoms.
Such detailed inferences are difficult enough with large-​scale
properly randomized evidence and would be impossible without it.
Because of their unknowable biases (see earlier), non​randomized
database analyses are simply not a viable alternative to large-​scale
randomized evidence. Nor would randomization of ‘only’ several
thousand patients have been sufficient. The availability of large-​scale
randomized evidence, in this case a meta-​analysis involving about
6000 deaths among 60 000 patients, has been essential in determining
which particular types of patient derive net benefit from fibrinolytic
therapy (or from surgical opening of acutely occluded arteries).
Promising meta-​analysis of small trials refuted by large
trials: Lack of significant benefit from magnesium
infusion in acute myocardial infarction
In animal studies, infusion of a magnesium salt can limit the myo-
cardial damage arising from sudden experimental blockage of a cor-
onary artery. By the early 1990s, there was considerable optimism
that a simple, inexpensive magnesium infusion might prove benefi-
cial after acute myocardial infarction. Twelve small trials, involving
between them a total of only about 2000 patients, had addressed this
question, and their aggregated results indicated a highly statistically
significant—​but implausibly large—​halving of risk (72/​1199 deaths
among those allocated magnesium versus 151/​1191 among the
controls, 2p <0.00001). At this time some argued that such results
constituted proof beyond reasonable doubt that magnesium was of
sufficient value to justify its widespread usage without seeking fur-
ther randomized evidence, but others remained sceptical, arguing
that the apparent results were far too good to be true.
Two trials, one (LIMIT-​2) involving 2000 patients and one (ISIS-​4)
involving 58 000, were then set up to test the possible effects of mag-
nesium more reliably. The first yielded a moderately promising re-
sult (Fig. 2.4.7), indicating avoidance of about one-​quarter of the
early deaths, but with the 99% confidence interval including the pos-
sibility that magnesium had no beneficial effect on early mortality.
The second (which had continued in spite of intense lobbying of its
Data Monitoring Committee to stop the trial), however, yielded a
completely unpromising result, so that the overall evidence, by that
time based on over 60 000 randomized patients, indicated no net
effect on mortality.
Nevertheless, some cardiologists remained hopeful that mag-
nesium might prove to be effective among specific subgroups.
Accordingly, the MAGnesium In Coronaries (MAGIC) trial
Percentage dead
10
5
0
0
1
2
3
4
5
Weeks from starting treatment
Aspirin only
Streptokinase
only
Routine hospital care alone
13% dead (568/4300)
Routine care + combination of
both streptokinase and aspirin
8% dead (343/4292)
Fig. 2.4.5  Effects of a 1-​h streptokinase infusion, and of aspirin for
about one month, on 35-​day mortality in the 1988 ISIS-​2 trial among
17 000 patients with acute myocardial infarction who would not
normally have received either treatment, divided at random into four
similar groups to receive aspirin only, streptokinase only, both or neither.
40
30
20
10
0
6
12
Hours from symptom onset to randomization
18
Loss of benefit
per hour of delay
to randomization:
1.6 SD 0.6 per
1000 patients
3000
14 000
12 000
9000
7000
50
24
0
Lives saved
per 1000
allocated
fibrinolytic
therapy (± SD)
Fig. 2.4.6  Benefit versus delay (0–​1, 2–​3, 4–​6, 7–​12, or 13–​24 h) in
the nine largest randomized trials of fibrinolytic therapy versus control
in patients with acute myocardial infarction. One-​month mortality
results for 45 000 patients with ST elevation or bundle-​branch block
when randomized, showing the definite net benefit even for the 9000
randomized 7–​12 hours after the onset of pain.
64
SECTION 2  Background to medicine
subsequently randomized 6000 patients, all of whom had received
reperfusion therapy within the past few hours, to magnesium versus
placebo but this also found no evidence of any net benefit.
It is interesting to consider what this sequence of magnesium trial
results (Fig. 2.4.7) might mean for those wishing to interpret other
randomized evidence. Our interpretation is that if something seems
too good to be true then it probably is—​or, more formally, that big
benefits are much less plausible than moderate benefits. None of
the 12 small trials had sufficient power to detect a moderate effect
on mortality, and although their aggregated results indicated that
mortality could be reduced by more than half, such an effect is too
extreme to be plausible, and could be misleading even though it is
highly significant. The LIMIT-​2 trial then suggested that magnesium
might reduce mortality by about a quarter, a result that is somewhat
more plausible but not clearly significant. The success of the ISIS-​4
and MAGIC trials in refuting the implausibly large benefit suggested
by the 13 smaller trials reinforces our point that often, when trying
to distinguish between the two medically realistic possibilities of a
moderate effect or no effect, only large-​scale evidence suffices. Even
the LIMIT-​2 trial, which recruited 2000 patients, was in retrospect
too small. (Another important methodological point is that ‘random
effects’ methods for meta-​analysis can produce importantly wrong
answers: applied to the 15 separate trials in Fig. 2.4.7, a standard
‘random effects’ meta-​analysis yields a summary odds ratio of 0.67
[95% CI 0.52–​0.85; 2p <0.001], suggesting—​clearly incorrectly—​
that magnesium reduces mortality by about one-​third!)
Trials in their epidemiological context: Blood pressure,
stroke, and heart disease
Quantitative epidemiological evidence about the effects of long-​term
differences in risk factors such as blood pressure or blood choles-
terol level can help in interpreting the results from trials of the ef-
fects of reducing these risk factors for only a few years. For example,
appropriate meta-​analyses of prospective observational epidemio-
logical studies indicate that, throughout the range of usual systolic
blood pressure in the populations studied (about 115–​180 mm Hg
systolic blood pressure, SBP), a lower value was associated with a
lower risk of ischaemic heart disease, with no apparent ‘threshold’
below which the relationship reversed (Fig. 2.4.8). This analysis
suggests that, in later middle age (60–​69 years), 10 mm Hg lower
Trial name
12 small trials
LIMIT– 2
ISIS–4 (1995)
MAGIC (2002)
Total
99% or
95% confidence intervals
2853/
(8·3%)
34482
2844/
(8·2%)
34487
72/1199
90/1159
2216/29011
475/3113
472/3100
2103/29039
11 8/1157
151/1191
0·44 (SE 0·10)
0·74 (SE 0·13)
1·06 (SE 0·03)
1 ·00 (SE 0·07)
1.00 (SE 0.03)
2p>0·1; NS
Magnesium better
0
0.5
1.0
1.5
2.0
Magnesium worse
Treatment effect 2p > 0·1; NS, adverse
Deaths/Patients
Allocated
magnesium
Allocated
control
Ratio of death rates
Magnesium : Control
Fig. 2.4.7  Effect of a magnesium infusion on 1-​month mortality among patients with acute
myocardial infarction. Ratio of the death rate in the treatment group to that in the control group
is plotted for each trial (as a black square with area proportional to the amount of statistical
information) along with its 99% confidence interval (horizontal line). A stratified overview of the
results of all these trials (and its 95% confidence interval) is represented by an open diamond.
40–49
50–59
60–69
70–79
80–89
Age
at risk
Effect of
10 mm Hg ↓ SBP
18% ↓ risk
27% ↓ risk
31% ↓ risk
120
140
160
180
Usual systolic BP (mm Hg)
Hazard ratio (95% CI)
1
2
4
8
16
32
64
128
Fig. 2.4.8  Stroke and ischaemic heart disease mortality rate in each
decade of age versus usual systolic blood pressure (SBP, mm Hg) at the
start of that decade, in a systematic overview of 61 prospective studies
involving 1 million adults. Hazard ratios are plotted with a group-specific
CI derived only from the variance of the log hazard in that one category
(including the reference category), and each square has area inversely
proportional to the effective variance of the log mortality rate.
2.4  Large-scale randomized evidence: Trials and meta-analyses of trials
65
SBP is associated with about 27% less ischaemic heart disease (IHD)
mortality (and about 35% less stroke mortality: Prospective Studies
Collaboration, data not shown).
By the mid-​1990s, several trials had been conducted to deter-
mine whether a few years of blood pressure reduction in middle
age would reduce the risk of stroke and of coronary heart disease.
Partly because of imperfect compliance, the mean difference in SBP
between the treatment and control groups in these trials was only
about 10 mm Hg. Even if such trial treatments would eventually
produce about 27% less coronary heart disease after many years
(as seen in observational studies), the effects seen within the 2 or
3 years that are available on average between randomization and
death in a 5-​year trial might well be somewhat smaller (perhaps
only about 15%, for example). But, considered separately, none
of the trials recorded enough coronary heart disease events (or
enough vascular deaths) for statistically reliable assessment of a
15% risk reduction.
For stroke, the trials provide direct and highly significant evi-
dence that most, or all, of the risk reduction associated with 10 mm
Hg lower usual SBP appears soon after the blood pressure is lowered
(Fig. 2.4.9). In contrast, the significant reduction in coronary heart
disease seen in the trials (16% SD 4, 95% CI 8–​23%; 2p = 0.00001)
seems to fall somewhat short of the difference of about 27% suggested
by the observational evidence. However, the coronary heart disease
reduction in the trials is still substantial and real (2p = 0.00001).
Taken together, Figs. 2.4.8 and 2.4.9 suggest that antihypertensive
regimens that produce differences of much more than 10 mm Hg
SBP will reduce stroke by more than half and heart disease by more
than a quarter. They also suggest that the proportional risk reduction
produced by a given absolute reduction in SBP will be approximately
independent of the initial SBP.
Results from large anonymous trials are
relevant to real clinical practice
A clinician is used to dealing with individual patients, and may feel
that the results of large trials somehow deny their individuality. This
is almost the opposite of the truth, for one of the main reasons why
trials have to be large is just because patients are so different from one
another. Two apparently similar patients may run entirely different
clinical courses, one remaining stable and the other progressing rap-
idly to severe disability or early death. Consequently, it is only when
really large groups of patients are compared that the proportion of
patients with truly good and bad prognosis in each can be relied on
to be reasonably similar. One commonly hears statements such as: ‘If
the effect of a treatment isn’t obvious in a couple of hundred patients
then it isn’t worth knowing about’. But, the accumulation since 1980
of large-​scale randomized evidence of such moderate effects with
treatments for heart disease, stroke, breast cancer, intestinal cancer
and various other conditions has transformed medical practice, and
may already have avoided millions of catastrophic disabilities, recur-
rences of cancer, and premature deaths.
It is also said that what is really wanted is not a blanket recommen-
dation for everybody, but rather some means of identifying those
few individuals who really stand to benefit from therapy. If any cri-
teria (e.g. a short-​term response to a non​placebo-​controlled course
of some disease-​modifying agent) can be proposed that are likely
to discriminate between people who will and will not benefit, then
these can be recorded prospectively at entry and the eventual trial
result subdivided with respect to them. However, there is a danger
in too detailed an analysis of the apparent responses of small sub-
groups chosen for separate emphasis because of the apparently re-
markable effects of treatment in these subgroups. Even if an agent
brought no benefit, it would have to be acutely poisonous for it not to
appear disproportionately beneficial in one or two such subgroups!
Conversely, if an intervention really avoids an approximately similar
proportion of the risk in each category of patient, it will, by chance
alone, appear not to work in some category or categories of patient.
The surprising extent to which this happens is evident from the ex-
ample in Table 2.4.2. A large anonymous trial will at least still help
to answer the practical questions of whether, on average, a policy of
widespread treatment (except where clearly contraindicated) is pref-
erable to a general policy of no immediate use of the treatment (ex-
cept where clearly indicated). Moreover, without really large trials it
is difficult to see how else many such questions relating to the effects
of treatments on death or disability (or other major outcomes) are
to be resolved reliably. Trials are at least a practical way of making
some solid progress, and it would be unfortunate if desire for the
perfect (that is, knowledge of exactly who will benefit from treat-
ment) were to become the enemy of the possible (that is, knowledge
of the average direction and approximate size of the effects of treat-
ment in many large categories of patient).
Trial
(or group
of trials)
Numbers
of events
treat : control
Odds ratios
& 95% confidence limits
(treat : control)
TREATMENT
←BETTER
TREATMENT
WORSE→
(i) Strokes
(ii) Coronary heart disease (CHD) events
38% SD 4
reduction
2p < 0.00001
16% SD 4
reduction
2p = 0.00001
STROKE
35–40%
LOWER
CHD
20–25%
LOWER
Reductions in risk associated
epidemiologically with a
LONG-TERM difference
of 5–6 mm Hg DBP:
0.5
1.0
HDFP trial
MRC 35–64 trial
SHEP trial
MRC 65–74 trial
13 other trials
ALL TRIALS
102:158
60:109
105:162
101:134
157:272
525:835
275:343
222:234
104:142
128:159
205:226
934:1104
HDFP trial
MRC 35–64 trial
SHEP trial
MRC 65–74 trial
13 other trials
ALL TRIALS
(Heterogeneity χ4
2 = 4.2, NS)
(Heterogeneity χ4
2 = 4.2, NS)
Fig. 2.4.9  Reduction in the odds of stroke and coronary heart disease in
all unconfounded randomized trials of antihypertensive drug treatment
(mean systolic blood pressure differences of about 10 mm Hg for
5 years). Conventions are as for Fig. 2.4.7.
66
SECTION 2  Background to medicine
FURTHER READING
Adjuvant Tamoxifen: Longer Against Shorter (ATLAS) Collaborative
Group (2013). Long-​term effects of continuing adjuvant tamoxifen
to 10 years versus stopping at 5 years after diagnosis of oestrogen
receptor-​positive breast cancer: ATLAS, a randomised trial. Lancet,
381, 805–​16.
Antithrombotic Trialists’ Collaboration (2002). Collaborative meta-​
analysis of randomised trials of antiplatelet therapy for prevention
of death, myocardial infarction, and stroke in high-​risk patients.
BMJ, 324, 71–​86.
Asymptomatic Carotid Surgery Trial (ACST) (2010). 10-​year stroke
prevention after successful carotid endarterectomy for asymptom-
atic stenosis (ACST-​1): a multicentre randomised trial. Lancet, 376,
1074–​84.
Chalmers I (1994). The Cochrane Collaboration: preparing, maintain­
ing and disseminating systematic reviews of the effects of health
care. Ann N Y Acad Sci, 703, 156–​63.
Chen ZM et al. for the COMMIT (ClOpidogrel and Metoprolol in
Myocardial Infarction Trial) collaborative group (2005) Addition
of clopidogrel to aspirin in 45,852 patients with acute myocar-
dial infarction: randomised placebo-​controlled trial. Lancet, 366,
1607–​21.
Cochrane AL (1979). 1931–​1971: a critical review, with particular ref-
erence to the medical profession. In: Teeling-​Smith G, Wells N (eds)
Medicines for the year 2000, pp. 1–​11. Office of Health Economics,
London.
Collins R, MacMahon S (2001). Reliable assessment of the effects of
treatment on mortality and major morbidity I: clinical trials. Lancet,
357, 373–​80.
Collins R, Peto R (1994). Antihypertensive drug therapy: effects on
stroke and coronary heart disease. In: Swales JD (ed) Textbook of
hypertension, pp. 1156–​64. Blackwell Science, Oxford.
Collins R, et al. (1987). Avoidance of large biases and large random
errors in the assessment of moderate treatment effects: the need for
systematic overviews. Stat Med, 6, 245–​50.
Collins R, Doll R, Peto R (1992). Ethics of clinical trials. In: Williams
CJ (ed) Introducing new treatments for cancer: practical, ethical and
legal problems, pp. 49–​65. John Wiley & Sons Ltd, Chichester.
Dowsett M, et  al. (2010). Meta-​analysis of breast cancer outcomes
in adjuvant trials of aromatase inhibitors versus tamoxifen. J Clin
Oncol, 28, 509–​18.
Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) (2005).
Effects of chemotherapy and hormonal therapy for early breast
cancer on recurrence and 15-​year survival: an overview of the ran-
domised trials. Lancet, 365, 1687–​717.
Early Breast Cancer Trialists’ Collaborative Group (EBCTCG)
(2015). Aromatase inhibitors versus tamoxifen in early breast
cancer:  patient-​level meta-​analysis of the randomised trials.
Lancet, 386, 1341–​52.
Fibrinolytic Therapy Trialists’ Collaborative Group (1994). Indications
for fibrinolytic therapy in suspected acute myocardial infarction:
collaborative overview of early mortality and major morbidity re-
sults from all randomised trials of more than 1000 patients. Lancet,
343, 311–​22.
HPS2-​THRIVE Collaborative Group (2014). Effects of extended re-
lease niacin with laropiprant in high-​risk patients. N Engl J Med, 371,
203–​12.
ISIS-​2 (Second International Study of Infarct Survival) Collaborative
Group (1988). Randomised trial of intravenous streptokinase, oral
aspirin, both, or neither among 17,187 cases of suspected acute myo-
cardial infarction: ISIS-​2. Lancet, 332, 349–​60.
ISIS-​4 (Fourth International Study of Infarct Survival) Collaborative
Group (1995). ISIS-​4: a randomised factorial trial assessing early
oral captopril, oral mononitrate, and intravenous magnesium sul-
phate in 58050 patients with suspected acute myocardial infarction.
Lancet, 345, 669–​85.
Law M, Morris J, Wald N (2009). Use of blood pressure lowering drugs
in the prevention of cardiovascular disease: meta-​analysis of 147
randomised trials in the context of expectations from prospective
epidemiological studies. BMJ, 338, b1665.
MacMahon S, Collins R (2001). Reliable assessment of the effects
of treatment on mortality and major morbidity II:  observational
studies. Lancet, 357, 455–​62.
Magnesium in Coronaries (MAGIC) Trial Investigators (2002). Early
administration of intravenous magnesium to high-​risk patients
with acute myocardial infarction in the Magnesium in Coronaries
(MAGIC) Trial: a randomised controlled trial. Lancet, 360, 1189–​96.
MRC Asymptomatic Carotid Surgery Trial (ACST) Collaborative
Group (2004). Prevention of disabling and fatal strokes by successful
carotid endarterectomy in patients without recent neurological
symptoms: randomised controlled trial. Lancet, 363, 1491–​502.
Prospective Studies Collaboration (2002). Age-​specific relevance of
usual blood pressure to vascular mortality: a meta-​analysis of indi-
vidual data for one million adults in 61 prospective studies. Lancet,
360, 1903–​13.
Prospective Studies Collaboration (2007). Blood cholesterol and vas-
cular mortality by age, sex and blood pressure: meta-​analysis of indi-
vidual data from 61 prospective studies with 55,000 vascular deaths.
Lancet, 370, 1829–​39.
Woods KL, et  al. (1992). Intravenous magnesium sulphate in sus-
pected acute myocardial infarction: results of the second Leicester
Intravenous Magnesium Intervention Trial (LIMIT-​2). Lancet, 339,
1553–​58.
Yusuf S, Collins R, Peto R (1984). Why do we need some large, simple
randomized trials? Statistics in Medicine, 3, 409–​20.

2.5 Bioinformatics 67
2.5 Bioinformatics 67
ESSENTIALS
Bioinformatics may be defined as ‘conceptualizing biology in terms
of molecules and applying “informatics techniques” (e.g. applied
mathematics, computer science and statistics) to understand and
organize the information associated with these molecules, on a
large scale’.
Clinical bioinformatics may be defined as ‘the clinical application
of bioinformatics-​associated sciences and technologies to under-
stand molecular mechanisms and potential therapies for human
diseases’.
To achieve these aims: (1) data must be curated to facilitate stand-
ardized access to existing information and to allow the submission
of new entries to data sets; (2) analysis tools should be developed
drawing upon both computational and biological/​clinical expertise;
(3) all analyses must be interpreted in a biologically/​clinically mean-
ingful manner.
Introduction
If clinical bioinformatics is to deliver the integration of molecular
and clinical data and thereby translate research knowledge into ef-
fective ‘personalized’ medicine, then two broad constituencies need
to be supported. Clinicians at the point of care need to understand
and integrate, perhaps via decision support mechanisms, entities
such as genotype/​phenotype correlations, biomarker discovery,
and pharmacogenomics; while researchers require accurate, struc-
tured and (ideally) coded clinical data, as well as biological reference
data sets.
Ever accelerating technological advances and precipitous falls
in the cost of high-​throughput technologies (e.g. whole genome
sequencing, expression profiling, high resolution image processing
and others) means that there is a veritable deluge of available data.
Accompanying falls in the cost of the substantial computational
power and associated data storage needed also mean that there is
opportunity for meaningful analysis. Ever faster turnaround times
(currently measured in hours) mean that it is now feasible to intro-
duce next-​generation sequencing (NGS) into workflows directly
contributing to patient care.
NGS technologies allow the identification of single nucleo-
tide polymorphisms, point mutations, and insertions or deletions
(indels) as well as larger structural changes such as translocations,
rearrangements, inversions, duplications, and copy number vari-
ations. When investigating somatic mutations e.g. in tumours,
comparison with germline samples facilitates variant detection. In
rare diseases the comparison within a given trio, proband and both
parents, serves a similar process. The need for defined metrics to
inform strict criteria-​based quality assurance is crucial if a clinical
bioinformatics pipeline is to be setup.
NGS has additional capabilities to investigate cellular proper-
ties over and above the determination of genomic sequence alone.
Epigenomics deals with the chemical modifications of nucleic acids
(e.g. 5′ methylation, and the consequent effect on gene expres-
sion). NGS offers the potential to identify changes across the entire
genome by capturing epigenetic information from multiple genes
simultaneously. Given that for some tumours epigenetic status re-
flects the overall prognosis, such analyses may provide substantially
enhanced prognostic information.
However, simply aggregating patient-​specific clinical data with
genetic, expression, or other data will not automatically lead to
better clinical outcomes. Clinical data is often unstructured and
incomplete, being spread across multiple paper and electronic sys-
tems, hence clinically meaningful semantic vocabulary standards
are needed (see next). At a cellular level it is clear that biology is
not solely dependent on the genome sequence alone, and in 2012
it was estimated that the biological function of approximately
half of all human genes remained unknown. Projects such as the
Encyclopaedia of DNA Elements (ENCODE), currently building
a comprehensive list of the functional elements in the human
genome, and the Kyoto Encyclopaedia of Genes and Genomes
(KEGG), which supports machine executed models of system-​level
biological pathways, are important. Without these the ability to in-
terpret analyses and to draw relevant biomedical patient-​specific
conclusions will be severely impeded.
2.5
Bioinformatics
Afzal Chaudhry
68
SECTION 2  Background to medicine
Components needed for clinical bioinformatics
Data storage
Flexible, extensible data warehousing is essential to accommodate
the volume of clinical and biomedical information. The ability to
support multiple data sources containing heterogeneous data sets
is crucial, and data structures must also be able to accommodate
sparse data sets as it is unlikely that any individual clinical record
will contain information on all possible concepts.
Typical warehouse designs are built upon a ‘dimensional fact’
model. Here, a fact is a concept relevant to decision-​making
(e.g. an observation made at a specific point in time such as a blood
pressure measurement), while a dimension describes some attri-
bute of that fact (e.g. the blood pressure was measured with the
patient supine).
Use of a common set of semantic terms to support data
aggregation/​interoperability
Control of the metadata dictionary describing all of the facts in the
warehouse is essential. For example, for laboratory tests, normal
ranges, and assay types may change over time and/​or may vary
from one laboratory to another. It is impossible then to meaning-
fully aggregate data over time or from multiple laboratories un-
less the results (facts) are interpreted using the relevant metadata
(dimensions). Examples of such metadata include the struc-
tured vocabulary/​ontologies listed in Table 2.5.1. Hierarchical
ontological terminologies reflecting clinical meaning such as
SNOMED CT are preferred over more epidemiological orien-
tated classification systems such as ICD-​10. Dimensions should
ideally be described using elements from a defined archetype rep-
resented in a definition set such as the openEHR reference model.
When using natural language processing technology to extract
structured standardized data from unstructured information, the
extraction should be ‘supervised’ by the metadata dictionaries to
allow data from text-​based records to be amalgamated with that
from a coded record.
Meaningful analytical tools
There are multiple data sets and tools to support the analysis of bio-
informatics data (Tables 2.5.2 and 2.5.3). Typically, these focus on
assessing similarities between molecular sequences based on align-
ment. Sequence-​based data sets significantly outnumber structural-​
based data sets because of the relative ease by which sequence data
can be obtained. Additional analyses, often using specialized soft-
ware, are needed over and above simple alignment analyses to detect
clinically relevant structural genomic alterations.
Protein orientated databases are often categorized as either pri-
mary, detailing the linear amino-​acid sequence, or secondary,
containing derived information. Secondary-​based analyses may
consider motifs or electrostatic interactions that are contiguous
in three-​dimensional space but not in the linear sequence. Some
macromolecular three-​dimensional structure databases contain a
hierarchical taxonomy to help identify evolutionary relationships.
Ultimately the most value is seen by combining data from multiple
sources—​clinical, sequence, structure, expression, and function (or
as many of these as exist). This may not always be straightforward
due to variations in nomenclature and data formats, although web-​
based gateways supporting the traversal of multiple databases are
becoming more effective.
Examples of clinical/​research areas benefitting
from clinical bioinformatics strategies
See Table 2.5.4.
Oncology research
Oncology research has tended to focus on single gene and single
pathway analysis. However, NGS offers both multiple simultaneous
analyses and extremely high sequence coverage thus substantially
increasing sensitivity. International consortia such as the Cancer
Genome Atlas are sequencing thousands of cancers to generate
data sets across different cancer subtypes. Computational theories
Table 2.5.1  Clinical classifications/​terminologies/​structured vocabularies available in the United Kingdom
Name (acronym)
Full name
Clinical related entities described
URL (accessed November 2018)
ICD-​10
International Classification of Disease 10
Diagnoses
https://www.who.int/classifications/icd/
icdonlineversions/en/
SNOMED CT
Systematized Nomenclature of Medicine—​
Clinical Terms
Symptoms, signs, diagnoses
http://​www.ihtsdo.org/​snomed-​ct
dm+d
NHS Dictionary of Medicines and Devices
Medication
http://​dmd.medicines.org.uk/​
NLMC
National Laboratory Medicine Catalogue
Laboratory investigations
https://​nlmc.x-​labsystems.co.uk/​
LOINC
Logical Observation Identifiers Names and
Codes
Laboratory investigations
https://​loinc.org/​
NICIP
National Interim Clinical Imaging Procedure
code set
Radiological investigations
https://digital.nhs.uk/services/terminology-and-
classifications/national-interim-clinical-imaging-
procedure-nicip-code-set
OPCS 4.7
Office of Population Censuses and
Surveys Classification of Interventions and
Procedures version 4.7
Procedures
https://isd.digital.nhs.uk/trud3/user/guest/group/0/
pack/10​
HPO
Human Phenotype Ontology
Phenotypic abnormalities
encountered in human disease
https://​hpo.jax.org/​app/​
2.5  Bioinformatics
69
including pathway network analysis and graph theory can be used
to model tumour-​related regulatory networks and interactions,
allowing complex interactions to be understood. The predictive
power of multigene biomarker panels, now potentially scaled into
the many thousands of genes analysed simultaneously as opposed
to just tens of genes, can be profoundly enhanced (e.g. in one study
a panel of 2300 genes could discriminate adenocarcinoma of the
lung from healthy tissue with 100% accuracy).
Pharmacogenomics
As the genetic/​molecular basis of the metabolism and mechanism
of action of drugs becomes increasingly understood, so therapy can
be individually tailored to some degree. Recent examples include
trastuzumab for HER2-​positive breast cancer and imatinib for chronic
myeloid leukaemia and conditions associated with tyrosine kinase-​
based mutations. NGS can identify somatic variants which help to
direct therapy (e.g. as resistant tumour clones emerge—​melanoma
with the BRAF mutation V600E is susceptible to vemurafenib while
the p61 BRAF V600E variant is not). In haematology the potential to
stratify an individual’s sensitivity to warfarin (VKORC1 and CYP2C9
gene polymorphisms) will help to guide appropriate dosing and avoid
potentially life-​threatening events.
Infectious diseases
The far smaller size of viral and bacterial genomes makes it possible
to sequence the genome of infecting pathogens. In the case of the
2009 H1N1 influenza pandemic, bioinformatics tools were able to
describe within a few hours of the first identification of a novel muta-
tion a possible mechanistic explanation by which it was able to mani-
fest such a severe phenotype. Computational analysis of genome
sequence and protein structures can help in identifying likely drug
susceptibility (e.g. the enterohaemorrhagic O104:H4 E. coli outbreak
in Germany in 2011), while individual infecting strains can be typed
and traced over both time and geographical distribution so sup-
porting more appropriate and economical public health strategies.
Digital imaging
Even among the most experienced histopathologists there can be
considerable interobserver variation in certain conditions. Objective
algorithms to identify tumour grading and to search for other tissue-​
based measures of disease activity using level sets, fractal analysis,
and machine learning can improve diagnosis. The adaptation of
astronomical algorithms coupled with their application to large an-
notated study cohorts is likely to provide a powerful set of analytical
tools. In dermatology, texture analysis, neural network framework-​
based analyses, data mining of skin images and computer-​based
reconstruction of the skin surface have all been used to support re-
search into reliable diagnostic strategies.
Table 2.5.2  Publicly available databases of biological knowledge
Domain
Database
URL (accessed November 2018)
Nucleotide
DDBJ
https://​www.ddbj.nig.ac.jp
GenBank
https://www.ncbi.nlm.nih.gov/genbank/
Genome
COGs
https://www.ncbi.nlm.nih.gov/COG/
Entrez genome
https://www.ncbi.nlm.nih.gov/genome
GeneCensus
http://bioinfo.mbb.yale.edu/genecensus/
Protein—​primary
NRDB
https://www.ncbi.nlm.nih.gov/protein
OWL
http://130.88.97.239/OWL/index.php
SWISS-​PROT
https://www.uniprot.org/uniprot/?query=reviewed:yes
Protein—​secondary
Pfam
http://pfam.xfam.org/
PRINTS
http://130.88.97.239/PRINTS/index.php
PROSITE
https://prosite.expasy.org/
Protein—​macromolecular
CATH
http://​www.cathdb.info/​
PDBeFold
http://www.ebi.ac.uk/msd-srv/ssm/​
Protein Data Bank
https://www.rcsb.org/
SCOP
http://scop.mrc-lmb.cam.ac.uk/scop/index.html
Functional/​systems biology
ENCODE
https://genome.ucsc.edu/ENCODE/​
KEGG
https://www.genome.jp/kegg/​
Vocabulary
Gene Ontology
http://​geneontology.org/​
Integrated systems/​web gateways
InterPro
https://www.ebi.ac.uk/interpro/
Uniprot
http://​www.uniprot.org/​
Table 2.5.3  Major research institutions providing access to a wide
range of bioinformatics databases and analysis tools
EMBL-​European Bioinformatics
Institute
https://www.ebi.ac.uk/
NCBI
(National Centre for
Biotechnology Information)
http://​www.ncbi.nlm.nih.gov/​
Wellcome Trust Sanger Institute
https://www.sanger.ac.uk/science/tools​
70
SECTION 2  Background to medicine
Conclusions
The need to deliver safe, timely, sustainable, and patient-​centric
care along with the need for evidence-​based strategies means that
any new technology first has to demonstrate clear translational
research benefits before it can be adopted into routine practice.
We now have the means to generate and analyse data sets that can
lead to such breakthroughs, and this continues to develop at a
breakneck pace.
If the most is to be made of these new data sets and analyses
then not only must appropriate clinical data be available for cor-
relation, but there will also be a need for more structured training
programmes and curricula to train clinicians in the analysis,
interpretation, and use of such data. The UK Health Education
England Genomics Education Programme Clinical Bioinformatics
group reported in early 2015 and has recommended a series of
steps and programmes for a range of healthcare staff to address the
long-​term goal of establishing a workforce fit for genomic medicine.
The implementation of such recommendations is awaited.
FURTHER READING
Raza S (2014). Defining the role of a bioinformatician. http://www.
phgfoundation.org/briefing/defining-the-role-of-a-bioinformatician​
Slade I, Burton H (2016). Preparing clinicians for genomic medicine.
Postgraduate Medical Journal, 92, 369–​71.
Table 2.5.4  Examples of clinical/​research areas benefitting from clinical bioinformatics strategies
Clinical/​research area
Example
DNA/​RNA sequencing and
expression profiling
Comprehension of biomolecular pathways underlying malignant transformation
Biomarker identification
Improved classification, early diagnosis, prognostication, and tailoring of therapy
Pharmacogenomics/​proteomics
Tailoring of therapy—​likelihood of therapeutic benefit as well as likely propensity to side effects
Pathogen genome/​protein
sequence/​structure and function
Description of putative mechanisms underpinning phenotypic manifestations
Susceptibility to antimicrobial therapy
Epidemiological analysis to identify environmental factors and support the relevant control mechanisms
Digital imaging
Machine learning based improvements in cellular/​tissue analysis and diagnosis
Image analysis of three-​dimensional geometric/​structural anatomy using both visible (e.g. glaucoma), and non​visible
spectra (e.g. infrared analysis of meibomian gland morphology in dry eye syndromes)

2.6 Principles of clinical pharmacology and drug t
2.6 Principles of clinical pharmacology and drug therapy 71
ESSENTIALS
In its widest sense a drug is any chemical entity that can perturb a
biological system. For the purposes of drug therapy the biological
system is the human body and the perturbation is exploited to aid the
diagnosis, treatment, or even cure of a disease process. Historically,
chemical entities have been plant derived or synthetic small (organic)
molecules (<1000 of Dalton in size). However, increasingly new
drugs are large biomolecules (up to 100 000 Dalton; e.g. monoclonal
antibodies or recombinant proteins). These biologics are expected to
exceed small molecules in the drug pipelines of the pharmaceutical
industry in the coming decades.
The role of clinical pharmacology is to provide the scientific basis
for rational prescribing: ‘patients may recover in spite of drugs . . . or
because of them’ (Gaddum). This sums up the dilemma facing any
doctor who prescribes a drug to a patient: it may be the explicit in-
tention to do the patient some good, but the drug may actually harm
and on rare occasions even kill them.
Principles of clinical pharmacology
Drug therapy can be considered under four headings:  (1)
pharmaceutical—​is the drug getting from the formulation into the
patient?; (2) pharmacokinetic—​how does the drug dose, formulation,
frequency, and route of administration affect the drug concentration
in the body, and the way that this concentration changes with time?;
(3)  pharmacodynamic—​how does a drug produce its pharmaco-
logical effects?; (4) therapeutic—​is the pharmacological effect trans-
lated into a therapeutic effect?
Adverse drug reactions
Definition and causes—​an adverse drug reaction is any unwanted or
harmful reaction experienced following administration of a drug, or
combination of drugs, under normal conditions of use that is sus-
pected of being related to the drug (or combination). If a causal
connection with a drug cannot be established, then the injury is
referred to as an adverse drug event. An adverse drug reaction may
be (1)  dose-​related—​usually due to an exaggeration of a known
pharmacological effect of the drug and to an extent predictable; or
(2) non-​dose-​related—​often caused by immunological or pharma-
cogenetic mechanisms and hence largely unpredictable.
Clinical importance—​adverse drug reactions are responsible for 1
to 4% of acute hospital admissions, affect 5 to 20% of inpatients at
some time during their admission, and are responsible for up to 3%
of inpatient deaths. Pharmacovigilance is the subspecialty of clinical
pharmacology devoted to the detection and evaluation of adverse
drug reactions.
Drug interactions
A drug interaction occurs when the effects of one drug are altered by
the effects of another drug, usually resulting in an adverse drug re-
action. Drugs likely to precipitate interactions typically (1) are highly
protein bound (e.g. aspirin and sulphonamides); or (2) induce drug
metabolism (e.g. phenytoin, carbamazepine, and rifampicin); or
(3) inhibit drug metabolism (e.g. cimetidine, metronidazole, anti-​HIV
protease inhibitors, and triazole antifungals). The drugs most likely to
be affected by drug–​drug interactions are those with a steep dose–​
response curve and a low therapeutic index.
A rational basis for prescribing
To maximize a drug’s therapeutic potential and minimize its side ef-
fects, a series of checks and balances is needed. The following should
be considered before any drug is prescribed: (1) does the patient
need a drug at all?—​do the risks outweigh its benefits?; (2) are the
benefits of the drug well established, preferably by randomized con-
trolled trials?; (3) what drug action is being sought, and what class of
drug can best provide it?; (4) what is the most appropriate drug in
that class, and in what formulation?; (5) what is the appropriate dose
and how long should the drug be prescribed for—​ a single course
or for indefinite use?; (6) will this drug interact with other drugs the
patient is taking?; (7) can this drug replace other drugs the patient
is taking?; (8) what does the patient need to understand about this
drug—​and who will communicate this?; (9) will it be necessary to
review the prescription of this drug—​and if so, when, how, and by
whom?; (10) does the patient need anything else to derive the most
benefit from this drug?
Practical prescribing
Guidelines and formularies—​in practice, the question checklist here
is not completed before every drug is prescribed, as many of the
questions are addressed by using appropriate therapeutic guidelines
2.6
Principles of clinical pharmacology
and drug therapy
Kevin O’Shaughnessy
72
section 2  Background to medicine
and formularies. Guidelines provide prerehearsed decision paths for
many of the issues raised in the checklist, and formularies tackle the
specific question of which drug to prescribe (from within a thera-
peutic class).
The patient’s drug history—​it is essential to obtain a thorough drug
history from the patient before prescribing. Key information that
should be obtained includes details of: (1) all the medicines currently
being prescribed, including their doses; (2)  any previous medical
treatments; (3) any ‘alternative’ treatments (e.g. herbal and homeo-
pathic remedies); (4) any self-​prescribed medicines; (5) any history of
allergy or adverse reactions to drugs.
Prescribing for the individual patient—​guidelines, formularies, and
other prescribing aids are not a substitute for an intelligent clin-
ical approach. The prescriber needs to establish what the patient’s
experience and expectations of drug therapy are, and the patient needs
to know the likely consequences—​both good and bad—​of taking any
drug that is prescribed. This dialogue is important, since it will often de-
cide whether the patient actually takes the drug as prescribed. Patient
compliance (or, more correctly, adherence) is a key variable in the pre-
scribing process, and one over which the doctor often has least control.
Balancing a drug’s therapeutic benefit
with its side effects
The prescriber checklist implies that balancing the expected bene-
fits of a drug against the expected harm is a straightforward process.
It is not. By the time a drug reaches the clinic we know a lot about
the size of its therapeutic effect, its relation to drug dose, and the
proportion of patients likely to show this effect. Measuring harm is
a less precise and much slower process. Terms such as ‘risk:benefit
ratio’ are widely used but can be very misleading. Benefit is often
measured in terms of amount and not its probability or frequency.
It can also be measured accurately by a single therapeutic endpoint;
harm from a drug encompasses a spectrum of side effects that differ
in both their frequency and severity.
Some side effects are predictable from the pharmacology of
the drug (so-​called type A) and are relatively common and dose-​
dependent. Others are rare and unpredictable (so-​called type B or
idiosyncratic). Type B side effects are usually more severe with a
higher burden of mortality and morbidity and, because of their low
frequency, will only come to light after a drug is licensed for clinical
use. It is important to remember that only a few thousand patients
are exposed to a drug before a license is granted (and for drugs with
orphan status, the patient exposure may be very much smaller).
The duration of dosing is also inevitably short when the intention
is to give the drug to patients indefinitely. Hence, crucial important
information about side effects has to be gathered after the drug is
licensed, and it may take many millions of patient-​years of dosing
before some drug side effects emerge.
Perhaps the best way of comparing a drug’s benefit with its harm
is to define them in terms of the number of patients needed to be
treated to observe a certain benefit and cause specific harm. These
are referred to as the number needed to treat (NNT) and number
needed to harm (NNH), respectively. Take, for example, the drug
clopidogrel. It is currently given to patients for 12 months after suf-
fering an acute coronary syndrome (non-​ST elevation myocardial
infarction, NSTEMI) to prevent myocardial infarction. Because of
its antiplatelet action it also causes gastrointestinal bleeding. Used
in this way to prevent non​fatal myocardial infarction, the NNT for
clopidogrel is about 67 and the NNH to cause a major bleeding epi-
sode is 100. So, if 100 patients are treated for 1 year, 1.5 non​fatal
­myocardial infarctions will be prevented at the expense of causing
one major bleeding episode. This is a much more transparent pres-
entation than the relative risk reduction that is often used to high-
light drug benefit in clinical trials. These percentage measures
inevitably boost the psychological impact of a drug’s effect—​the
NNT for clopidogrel equates to a 22% reduction in the frequency of
non​fatal myocardial infarction.
But even with reliable measures of harm and benefit, how do we
decide where the final balance lies? It will depend on other factors
such as the severity of the disease and whether there are safer al-
ternative drugs. Hence, if we want to treat a disease with a high fa-
tality rate using a drug which is highly effective and carries little risk
of harm, the balance is clearly in favour of the drug. But if the dis-
ease itself carries no mortality or morbidity and we propose using a
drug whose effectiveness is low and carries a high risk of harm, the
balance is clearly against using the drug. Most clinical decisions to
use a drug or not will be in the grey area between these two extremes.
Efficacy, effectiveness, and efficiency of drugs
These terms are not synonymous and can be easily confused. Efficacy
is a pharmacological term. It refers to the ability of a drug to bring
about a certain size of an effect at a given concentration or dose.
The effect may not be applicable to clinical practice or is only ac-
curately measured in a clinical trial. Hence the term is best reserved
for the performance of a drug in this setting. Effectiveness refers to
the performance of a drug in everyday clinical use, and is defined as
the likelihood and extent of the therapeutic effect in a given patient.
Efficiency weights drug performance against cost: it is the ratio of
effectiveness to cost. Clearly, it is more efficient to use the cheaper of
two drugs that are equally effective and safe.
The therapeutic index of a drug
The therapeutic index is a term taken from animal pharmacology.
It is the dose needed to harm over the dose needed to produce the
therapeutic response. However, as harm is usually measured by a
drug’s lethality, it is not a useful clinical measure. Instead, the index
is employed clinically in a very loose and entirely qualitative way.
Drugs that produce side effects at doses well outside the clinical
dose range are said to have a high therapeutic index and those where
the ranges are much closer or even overlap are said to have a low
therapeutic index. Hence penicillins have a high therapeutic index;
large doses can be given without the worry of adverse effects unless
the patient is allergic to penicillins. In contrast, digoxin has a low
therapeutic index; the doses causing toxicity overlap with those pro-
ducing therapeutic benefit. Drugs with a low therapeutic index in-
clude: aminoglycoside antibiotics, anticoagulants, anticonvulsants,
antihypertensives, some antiparasitic and antiviral drugs, cardiac
glycosides, and cytotoxic and immunosuppressant drugs.
To increase the margin of safety for drugs with a low index, dosing
can be guided by measuring drug levels in plasma or serum. Such
therapeutic monitoring is mandatory for drugs such as lithium and
aminoglycoside antibiotics because of additional pharmacokin-
etic problems with their use. Anticoagulants are unusual in being
2.6  Principles of clinical pharmacology and drug therapy
73
monitored by their effect on clotting rather than as a plasma drug
concentration.
Formularies
Formularies are lists of medicines for prescribers and pharmacists,
intended to guide the choice and facilitate the dispensing of medi-
cines. Many give details of the formulation and doses of drugs. Each
formulary is produced primarily for a particular group, usually the
prescribers in one country or region or institution, or even one prac-
tice. Most formularies are restrictive (i.e. they make a narrow choice
of medicines from all those available). This is typical for the formu-
lary of a hospital, or of a health maintenance organization. A hospital
formulary lists only the preparations that are stocked in the hospital
pharmacy; a health maintenance organization formulary only those
that the organization will pay for. The British National Formulary
(BNF) is probably the best known and most widely used formulary
of all, but is unusual in including all medicines available for prescrip-
tion in the United Kingdom, whether they are good choices or not.
However, every section of the BNF has concise and critical ‘notes to
facilitate the selection of suitable treatment’ that precede the list of
available agents. This invaluable resource is revised biannually and
available online (https://​www.medicinescomplete.com/​about/​).
The WHO ‘Model List of Essential Drugs’
In many developing countries, having limited health budgets means
that large sections of the population have no access to drugs or
healthcare, and governments cannot afford to provide necessary
drugs. To help them to use their limited funds in the best ways, the
World Health Organization (WHO) has since 1977 published a
regularly updated Model List of Essential Drugs. It is updated every
two years and currently in its 18th edition (http://​www.who.int/​
medicines/​publications/​essentialmedicines/​en/​index.html). Essential
drugs on the list are intended to ‘provide safe, effective treatment
for the majority of communicable and non​communicable diseases’.
The WHO list is a ‘model’ list that can be adapted to meet the needs
of the local health economy. Hence, over 150 countries have an es-
sential list based on the WHO model. It is a salient fact that the first
WHO list contained 208 drugs and, in the intervening 30 years, it
has not doubled in size. There may be some clear lessons here for the
drug lists of developed countries.
Medicines management
Rapidly escalating costs of providing drugs within healthcare sys-
tems and the need to maximize drug safety mean that the concept
of medicines management is now widespread. By bringing together
clinical, pharmacy, and financial skills, drugs that are considered es-
sential can be prescribed in the most cost-​effective and safest way.
In the United Kingdom, drugs are assessed at a national level by the
National Centre for Health and Clinical Excellence (NICE) before
deployment within the National Health Service (NHS). NICE con-
siders in detail the evidence for a drug’s alleged benefit and weights
this against known side effects and economic modelling of its total
cost versus benefit within the NHS.
The principles of clinical pharmacology
Drug therapy can be considered under four headings—​pharma­ceutical,
pharmacokinetic, pharmacodynamic, and therapeutic—​each of which
addresses a pertinent question about drug therapy (Fig. 2.6.1).
The pharmaceutical process
The pharmaceutical step is concerned with the question, ‘Is the drug
getting from the formulation into the patient?’
The route of drug administration is usually a more crucial choice
than how it is formulated. Nevertheless, formulation can greatly af-
fect the rate and extent of drug absorption. Repackaging short-​acting
drugs into ‘sustained’ or ‘modified’ release formulations to slow re-
lease of the drug into the gut allows them to be taken once daily.
Morphine, calcium channel blockers (e.g. diltiazem and nifedipine),
theophylline, and l-​dopa have been widely reformulated in this way.
Other drugs are formulated for specific routes of administration:
glyceryl trinitrate is available for sublingual (as a spray and tablets),
buccal, and transdermal (as a paste or patch) use. To understand the
differences between these various routes, it is necessary to under-
stand the concept of systemic availability.
Systemic availability
Systemic availability, commonly called bioavailability, is the pro-
portion of administered drug that reaches the systemic circulation
and is available for distribution to the site of drug action. If a drug
is given intravenously, it will enter the systemic circulation directly
(i.e. it is said to be 100% bioavailable). The same drug given orally,
or by any other route, must be absorbed first (which may be an in-
complete process), and possibly metabolized, before entering the
systemic circulation.
Metabolism can occur in the gut wall and liver following oral ad-
ministration, although it can occur at any site of drug administra-
tion. The cells lining the gut also express drug efflux pumps (e.g.
P-​glycoprotein) and transporters that can actively reduce the extent
of absorption of a drug. These processes, together with incomplete
absorption, ensure that most drugs have a bioavailability of less than
100% when given orally. In some cases, the reduction is so large that
a drug has zero bioavailability and is clinically ineffective. Oral in-
sulin or benzylpenicillin are good examples of this problem, with
their instability in the stomach preventing significant absorption.
Complete presystemic metabolism also explains why glyceryl tri-
nitrate and buprenorphine are orally effective only if given sublin-
gually or as a buccal patch. Absorption from the mouth allows the
drug to bypass gut wall and liver metabolism.
Special drug formulations
Most drugs are given orally; oral formulations include syrups, or-
dinary (instant release) tablets, capsules, and modified release for-
mulations. However, drugs can be given by other routes, including
sublingually, buccally, rectally, transdermally, by inhalation, and by
injection intravenously, subcutaneously, intramuscularly, or locally.
Modified release formulations
Most conventional instant release tablets are designed to disintegrate
in the stomach or proximal small bowel, so that absorption is com-
plete within a few hours on ingestion. Modified or sustained release
74
section 2  Background to medicine
formulations are oral formulations that allow a drug to be released
over long periods (12–​24 h typically) relative to the half-​life of the
drug. The intention of the prolonged and slowed release is to smooth
out the concentration profile of the drug in the blood and extend its
duration of action. They include formulations of theophylline, ni-
fedipine, diltiazem, morphine, and lithium. Prescriptions of these
drugs should specify the exact formulation, as formulations differ in
systemic availability and may not be interchangeable.
Sublingual, buccal, rectal formulations
Drugs that are absorbed through the oral or rectal mucosa avoid
first-​pass metabolism in the liver by uptake into veins that drain
directly into the systemic circulation. For example, sublingual gly-
ceryl trinitrate is rapidly effective as a sublingual tablet, but if the
tablet is swallowed, the remainder is not bioavailable because of high
presystemic metabolism. Rectal administration achieves a direct ef-
fect on the local bowel wall (e.g. corticosteroids in ulcerative colitis),
but is also useful for achieving high blood levels rapidly when intra-
venous access is difficult (e.g. diazepam for seizure control).
Transdermal formulations
Some lipid-​soluble drugs are well absorbed through the skin,
and their transdermal delivery via ‘patches’ allows controlled re-
lease over many hours or days. Examples are glyceryl trinitrate in
the long-​term treatment of angina pectoris, transdermal hyoscine
for travel sickness, oestradiol as hormone replacement therapy,
buprenorphine for pain control, and nicotine for smoking cessation.
Inhaled formulations
The lung provides a huge surface area (c. 100 m2) for drug absorp-
tion, but to reach the distal airways, a drug for delivery by inhal-
ation must be associated with particles in the 2 to 5 µm range. These
can either be solid particles (dry powder devices) or dissolved in
small droplets (aerosol devices). For hand-​held aerosol inhalers the
drug is dissolved in a volatile hydrocarbon, but in nebulizers the
drug is in an aqueous solution that is aerosolized by a jet of air or
oxygen. Both aerosols and dry powders are widely used to deliver
inhaled corticosteroids and bronchodilators used in the manage-
ment of asthma and chronic obstructive pulmonary disease. Even
peptides can be delivered by this route, as demonstrated by the li-
censing of inhaled insulin. The efficiency of inhalation as a route of
administration also explains the ‘success’ of some drugs of abuse
(e.g. nicotine and crack cocaine).
Subcutaneous, intramuscular, and local injections
The rate of absorption of insulin from the site of subcutaneous in-
jection is controlled by its physical state (e.g. monomer, crystalline,
or non​crystalline), pH, and the presence of zinc ions or protamine
PHARMACEUTICAL PROCESS
‘Is the drug getting into the patient?’
Formulations
Routes of administration
(compliance)
PHARMACOKINETIC PROCESS
‘Is the drug getting to its site of action?’
PHARMACODYNAMIC PROCESS
‘Is the drug producing the required
pharmacological effect?’
THERAPEUTIC PROCESS
‘Is the pharmacological effect being
translated into a therapeutic effect?’
Biliary
excretion
Rectal or
sublingual
administration
Parenteral
administration
Disintegration etc.
Dissolution
‘First pass’
Elimination
Therapeutic/toxic effects
Absorption
Distribution
Plasma proteins
Tissues
Elimination
Hepatic metabolism
Renal excretion
Other
Molecular pharmacology
Cell and tissue pharmacology
Cell and tissue physiology
Organ physiology
Clinical effects
Extracellular fluids
Tissues (sites of action)
d
n
u
o
b
n
U
Drug in solid dosage form
Drug in particulate form
Drug in solution
Metabolism in gut lumen and gut wall
Hepatic metabolism
Pharmacological effects
Protein-
bound
Fig. 2.6.1  The four processes of clinical pharmacology in relation to drug therapy.
2.6  Principles of clinical pharmacology and drug therapy
75
(isophane) in the buffer in which it is suspended. It is also affected
by altering the amino acid sequence of insulin; allowing recom-
binant insulins, which are rapidly (lispro and aspart) or slowly re-
leased (Glargine or Detemir) after injection. Soluble insulins have a
rapid onset (15–​30 min) and short duration of action (4–​6 h), so they
are usually given together with intermediate or long-​acting insulin.
Long-​acting insulins act for more than 24 h and can provide the in-
sulin background as a once daily injection. The isophane insulins
have an intermediate duration of action and are usually given twice
daily mixed with soluble insulin.
Absorption of a drug after subcutaneous injection is affected by
blood flow. Hence, the duration of action of local anaesthetics can be
prolonged by vasoconstriction. Adrenaline or felypressin is added to
the subcutaneous formulations for this purpose. Smoking also re-
duces subcutaneous insulin absorption by causing cutaneous vaso-
constriction. Reduced absorption also explains why subcutaneous
adrenaline is not advised for the treatment of anaphylaxis.
The intramuscular route is a popular parenteral route but can be
erratic. Hence, phenytoin and diazepam should not be given by this
route for emergency use. Absorption following intramuscular in-
jection may be retarded by esterifying a drug to a large lipid mol-
ecule. This gives oily formulations that provide long-​lived drug
depots in muscle, which are used in the treatment of male hypo-
gonadism (testosterone enanthate or undecanoate) and schizo-
phrenia (fluphenazine or flupentixol decanoates).
Combination formulations in oral therapy
Combination products are attractive and may aid compliance (see
following paragraphs), but should be used only when at least two
criteria are met:
•	The frequency of administration of the two drugs is the same.
•	The fixed doses in the combination product are therapeutically
and optimally effective in most cases (i.e. it is not necessary to alter
the dose of one drug independently of the other).
Acceptable combination products include:
•	Aspirin plus codeine (co-​codaprin) or paracetamol plus dihydro­
codeine (co-​dydramol), pairs of drugs that have different analgesic ac-
tions (which synergize) and different adverse effects (which do not).
•	l-​Dopa plus a peripherally-​acting dopa decarboxylase inhibitor
(benserazide or carbidopa); the peripheral action of the decarb-
oxylase inhibitor blocks peripheral metabolism of l-​dopa, which
is free to enter the central nervous system, where it is converted to
dopamine, producing the therapeutic effect in Parkinson’s disease.
•	Combined oral contraceptives, which contain an oestrogen and a
progestogen.
•	Ferrous sulphate plus folic acid, used to prevent anaemia in
pregnancy.
•	Co-​amoxiclav (amoxicillin plus clavulanic acid); the β-​lactamase
inhibitor, clavulanic acid, prevents the breakdown of amoxicillin
by bacterial penicillinase, so broadening its spectrum.
The patient’s use of a drug: Compliance
and concordance
Compliance is the extent to which a patient follows a prescribed drug
regimen. Some prefer the term ‘concordance’ or ‘adherence’, to make
it clear that therapeutic decisions are best arrived at jointly between
prescriber and patient and avoiding a patronizing relationship. Rates
of adherence are difficult to measure and can be surprisingly low,
even in conditions where the drug has a very obvious benefit, such
as epilepsy and asthma. Possibly as few as 1 in 6 patients take a drug
exactly as prescribed and 1 in 6 take none at all; the remainder take
a different dose and/​or a different frequency from that prescribed.
Poor adherence is still not well recognized by doctors as a cause of
therapeutic failure.
The effect of the prescribing regimen
Apart from the financial cost to the patient (which in some countries
is substantial), the two factors that determine adherence to drug
therapy are the complexity of the regimen and the likelihood of side
effects. The complexity of the prescribed regimen reflects frequency
of administration and the number of drugs prescribed. Generally
speaking, the more frequently a drug is prescribed per day, and the
more drugs in total are prescribed, the lower is the rate of compli-
ance. For most chronic diseases, drugs that can be taken once daily
are preferred. Obviously, using several drugs (polypharmacy), where
one may be adequate, should be avoided wherever possible.
Side effects that a patient attributes to a drug may lead them to
stop taking the drug completely. It may be necessary then to persuade
them to persevere or switch to an alternative drug. However, giving
another drug to simply reduce the side effects of the first drug has
to be thought through carefully. For example, giving furosemide to
a patient who develops ankle oedema on nifedipine, causing in turn
gout or hypokalaemia, is poor practice. But giving omeprazole to a
rheumatoid patient who develops acid reflux on their non​steroidal
anti-​inflammatory (NSAID) is appropriate.
The effect of the illness
People with severe mental health problems (e.g. patients with
schizophrenia or manic-​depressive psychosis), often take medicines
unreliably.
Physical disability may cause difficulty even in patients who want
to take their medicine. For example, patients with rheumatoid arth-
ritis who cannot reach the tablets, or cannot remove the top of a
childproof container, cannot take them without help.
Sometimes, a good response to treatment leads patients to stop.
For example, patients with tuberculosis need long courses of several
drugs to eradicate the infection; once the symptoms have resolved,
motivation to continue treatment may decline, risking reactivation
and the emergence of resistant tuberculosis.
Some diseases may promote compliance. Patients with insulin-​
dependent diabetes easily become very ill quite quickly if they forget
to take their insulin, and that is likely to make them comply, al-
though they may not use it precisely as advised. Patients in whom
a β-​blocker or vasodilator has prevented anginal attacks will also be
more likely to have good compliance.
The patient’s behaviour
People tend to forget to take drugs, or cannot be bothered; they may
feel no need for treatment (e.g. in asymptomatic hypertension); they
may be unclear about the prescribing instructions; they may not
want to feel dependent or be thought to be dependent on ‘drugs’.
There may be social or physical reasons why they cannot reach a
pharmacist, financial difficulties, or everyday inconveniences in
carrying and taking the medication.
76
section 2  Background to medicine
The doctor’s behaviour
The enthusiasm and confidence with which a drug is prescribed, and
the extent to which these attitudes are transmitted to the patient,
may influence not only compliance but also the response to therapy.
This is related to the placebo effect of drug taking.
Methods of assessing adherence
It is important to assess adherence both in everyday practice and
in clinical trials. The most obvious and usually the easiest approach
is to ask the patient whether he or she has been taking the drugs,
and whether there have been any problems. If the doctor is non-​
judgemental and indicates that difficulties are common, the patient
is encouraged to be open.
Less directly, one can ask to see the patient’s tablets: this at least
confirms that the prescription has been filled. Counting the tablets
left in the bottle is a guide to how many have been taken, but some
may have been thrown away. Recording devices fitted in the caps of
medication containers can record the frequency and exact timing of
the opening of the container, and are useful in research.
If a patient is vague or untrustworthy, measurement of the drug
level in plasma or urine may give some reassurance—​at least on the
day they visit the doctor. The list of drugs for which assays are rou-
tinely available is small, but a surrogate marker can be used (e.g.
the level of thyroid stimulating hormone (TSH) to ensure compli-
ance with thyroxine or the international normalized ratio (INR)
for patients on warfarin). Alternatively, a pharmacological effect
such as heart rate can be measured directly (to assess β-​blocker
compliance).
Methods of improving adherence
Adherence can be improved by supervised administration of the
drug, by removing barriers, by simplifying the therapeutic regimen,
and by educating the patient on the need to take the medicine, with
reinforcement whenever possible.
Supervised administration
Administration of a drug by the doctor or nurse ensures compli-
ance. This is possible in hospital or when occasional administration
is required (e.g. intramuscular injections of vitamin B12, long-​acting
intramuscular depot injections of neuroleptics, and supervised
twice-​weekly antituberculosis therapy). A relative or other carer can
give the drug at home and this may be aided by using dosimeter
boxes prefilled by a pharmacist.
Removing barriers to adherence
Adherence may be encouraged by prescribing pleasant-​tasting
syrups rather than tablets for children and older people, and by
using a drug or formulation that minimizes side effects.
Simplification of the therapeutic regimen  The therapeutic regimen
can be simplified by reducing both the number of drugs a patient has
to take and the frequency of administration. Modified release or com-
bination formulations may be useful here.
Education and reminders  Educating the patient about why treat-
ment is necessary (e.g. treating hypertension or diabetes reduces the
risk of serious complications) is time-​consuming but improves com-
pliance. In the treatment of certain infections (e.g. tuberculosis) and
in typhoid carriers, the wider importance to the community should
also be explained.
Even in the well-​motivated, reminders to take the treatment may
improve compliance. Many drugs for long-​term use are also now
dispensed in a ‘calendar pack’ to help compliance.
The pharmacokinetic process
Pharmacokinetics concerns the complicated question ‘How does the
drug dose, formulation, frequency and route of administration affect
the drug concentration in the body, and the way that this concentra-
tion changes with time?’ The answer will depend on the absorption,
distribution, metabolism, and excretion of that drug (otherwise
known as ADME). This naturally lends itself to complex mathemat-
ical modelling, but the key points can be understood without being
overwhelmed by the mathematics.
The size and chemistry of biologics poses special problems for
both the measurement and understanding of their ADME. Generally
they are metabolized to small peptide fragments or amino acids
that can be renally cleared or recycled through cellular protein syn-
thesis. This process can be very rapid for small peptides and proteins
(c.hours), but very slow for large proteins such as IgG that can have a
half-​life of 7–​28 days. The uptake of biologics from the subcutaneous
injection sites frequently used for biologics is also slow and affected
by lymphatic drainage and binding of any Fc-​domains to FcRn re-
ceptors on endothelial cells. Small proteins may also be deliberately
modified by, for example, PEGylation, glycosylation, or coupling to
larger proteins such as albumin to reduce their degradation.
Basic pharmacokinetic terms and concepts
Before considering the various components of ADME, a few pharma-
cokinetic terms need to be explained. Consider a time after admin-
istering a drug when its absorption and distribution throughout the
various compartments of the body are complete. At this time, there
are no concentration gradients across compartments and any fur-
ther decline in drug concentration reflects drug elimination, that is,
excretion and metabolism of the drug. Several parameters charac-
terize this elimination phase of a drug:
Volume of distribution (Vd)
This can be a confusing term, since it is really a mathematical device
and not a true physical volume. Hence, it rarely relates to any par-
ticular water compartment (e.g. plasma, 3 litres; extracellular water,
16 litres; total body water, 42 litres). It simply allows the amount of
drug remaining in the body to be inferred from the drug concen-
tration (concentration = amount/​volume). Drugs have a wide range
of values for their Vd and sometimes they are much greater than
the total water space. This is typical for drugs that bind to tissues
or partition into lipid. Amiodarone is an extreme example of this
behaviour with a Vd of approximately 5000 litres. Biologics in sharp
contrast usually have a small Vd (e.g. 2–​4 litres) reflecting their con-
finement to the blood space.
Clearance
In the same way that the volume of distribution is a proportion-
ality term, so clearance is used to relate the rate of drug excretion
2.6  Principles of clinical pharmacology and drug therapy
77
to the concentration of the drug in the body. It is expressed in
units of flow (volume per unit time) and is effectively the volume
of the Vd cleared of drug per unit time. A drug can be elimin-
ated (or cleared) from the plasma by one or more mechanisms
(e.g. through the kidney by filtration and the liver by metabolism).
The total clearance of the drug is simply the sum of these different
organ-​based clearances.
Plasma half-​life
During the elimination phase, the drug concentration falls in a
predictable way: in a fixed time called the half-​life it falls by 50%,
and by a further 50% after another half-​life and so on. Thus, it takes
between four and five half-​lives for 95% of the drug to be elimin-
ated (see Fig. 2.6.2). For most drugs, the half-​life is constant (except
if the kinetics becomes nonlinear—​see following paragraphs). The
half-​life is not affected by the starting concentration or amount of
the drug, but it is directly affected by the volume of distribution and
clearance of a drug (actually proportional to their ratio).
Repeated dosing and the ‘plateau principle’
A drug can be given once but it is more common for a drug to be
given repeatedly (even indefinitely). After the first dose, the plasma
level will rise to its maximum (Cmax) and fall as shown in Fig. 2.6.2.
If the dose is not repeated until all the drug has been eliminated
(which means the dose interval must be long, e.g. 10 half-​lives),
the drug level will oscillate between the Cmax after a single dose and
a trough value of zero. The problem with this regimen is that to
give a plasma level that is therapeutic (but not toxic), a large part
of the dosing interval is spent with a non​therapeutic drug level.
To get around this, the drug is given at smaller doses separated by
short dose intervals compared to the drug’s half-​life. Because the
dose interval is similar to the drug’s half-​life, the first dose of drug
is not completely eliminated before the second dose is given. The
third dose is given before either the second dose, or the remaining
fraction of the first dose is eliminated and so on. Hence, the drug
plasma level accumulates and eventually it comes to a steady state
where it oscillates between a new peak and trough (substantially
greater than zero); ideally, the peak is below the toxic threshold
and the trough above the therapeutic threshold throughout the
dose interval. This is what is known as the plateau principle and the
therapeutic and toxic threshold levels for the drug define its ‘thera-
peutic window’.
Loading doses
The half-​life also dictates the time it takes for a drug to reach its
plateau with repeated dosing. In the same way that it takes almost
5 half-​lives to eliminate a drug, it takes almost 5 half-​lives to reach
plateau. The delay imposed by the drug’s half-​life in reaching steady
state can be unacceptable clinically. Consider, for example, the use
of digoxin to treat fast atrial fibrillation. Because of the long half-​
life of digoxin (40 h), it takes over a week to reach plateau, and over
twice this if the patient has renal failure because of the increased
half-​life. The solution is to give a loading dose of a drug. The loading
dose is intended to take the blood concentration rapidly into the
0
6
12
18
[mg/litre]
12
11
10
9
8
7
6
5
4
3
2
1
0
24
30
36
42
48
54
60
Time after dosing (h)
Drug concentration in the plasma
Therapeutic
threshold
Toxicity threshold
Drug elimination with a t½ of 12 h
Cmax
66
72
78
84
90
96
102 108 114 120
Therapeutic
window
Fig. 2.6.2  Plateau principle. Plasma concentrations of a hypothetical drug after a single oral dose (black
line) versus repeated dosing (red line). A smaller dose is used for repeated dosing, but the half-​life is the
same. Both approaches give levels within the ‘therapeutic window’, but only with repeat dosing are the
levels within the window throughout the dose interval.
78
section 2  Background to medicine
therapeutic window. Dosing then resorts to the usual maintenance
dose, which is chosen to replace the drug eliminated in each subse-
quent dose interval.
Absorption and systemic availability
Systemic availability, introduced earlier in this chapter, is meas-
ured by plotting the blood concentration at various times after
dosing. Typical curves are shown in Fig. 2.6.3 for three different
formulations (containing the same dose) of a hypothetical drug.
The area under each of these curves (abbreviated to AUC) reflects
the bioavailability of the formulation and it is actually the same in
this simulation. However, it is clear that they produce very different
profiles in terms of the maximum concentration (Cmax) and the
time to reach Cmax after dosing (tmax). This reflects the decreasing
rate of absorption from left to right for the three formulations, and
hence the duration over which absorption occurs. Hence the fastest
absorption gives the highest Cmax, but this would cause transient
toxicity as crosses into the toxic range for this drug. The middle
curve takes the drug level above the therapeutic threshold for as
long as the first curve but remains below the toxicity threshold. The
slowest rate of absorption here gives a Cmax that is too low because
the drug concentration is subtherapeutic throughout.
Choosing the optimum concentration profile, and hence the for-
mulation, depends on the therapeutic effect being sought. Hence, a
rapid (instant release) formulation of nifedipine would be needed
to treat an episode of angina and is typified by the left-​hand curve
in Fig. 2.6.3. The transient flushing and headache this would
cause would be an acceptable side effect. However, when drugs are
intended to be given repeatedly, and the therapeutic effect is related
to a steady state concentration, the flatter profile of the slowest ab-
sorption formulations are preferred (right-​hand curve in Fig. 2.6.3).
So, for chronic angina prophylaxis, a slow release formulation of ni-
fedipine would be preferred to achieve the smoothest profile of drug
level in the blood after dosing.
The rate of absorption
Gastrointestinal motility
Drugs are absorbed mainly in the upper small intestine, so altered
gastric emptying can affect absorption. For example, in migraine,
the rate of absorption of analgesics is reduced because of reduced
gastric motility, delaying the response to an oral analgesic. In fact,
the shift in the drug concentration–​time profile is similar to the
rightward shift produced by reformulation discussed in Fig. 2.6.3.
This delay can be reduced by giving metoclopramide to accel-
erate gastric emptying. Erythromycin can have a similar effect on
gastric emptying. The converse of delayed emptying occurs with
antimuscarinics or older antihistamines (e.g. promethazine and di-
phenhydramine) or tricyclic antidepressants (e.g. amitriptyline) that
have substantial antimuscarinic actions. This can be particularly im-
portant in drug overdoses involving these drugs.
If a drug dissolves more slowly than the stomach empties, in-
creased gastrointestinal motility reduces both the rate and extent of
absorption. This effect is exaggerated for sustained release formula-
tions, such that in severe diarrhoea enteric-​coated formulations can
pass through the gut intact. Proximal ileostomies are also a problem
for these formulations.
1
0
10
9
8
7
6
5
4
3
2
Drug concentration in the plasma
0
2
4
6
8
10
12
Time after dosing (h)
Therapeutic threshold
Toxicity threshold
14
16
18
20
22
24
[h]
[mg/litre]
Cmax
Cmax
Cmax
Fig. 2.6.3  Effect of slowed absorption on drug kinetics. Plasma concentration profiles for a hypothetical
drug formulated to progressively slow absorption (slows left to right). As the absorption is slowed, the
peak concentration (Cmax) falls and is delayed.
2.6  Principles of clinical pharmacology and drug therapy
79
Malabsorption
Drug absorption is often impaired in patients with malabsorption,
but not always. For example, the absorption of propranolol and
some antibiotics (co-​trimoxazole and cefalexin) is increased in pa-
tients with coeliac disease. Untreated coeliac disease patients have a
higher intraluminal pH that can affect the ionization of a drug, while
the villous atrophy removes CYP 3A4 that can significantly reduce
the first-​pass metabolism of some drugs (e.g. simvastatin). Digoxin,
however, is less well absorbed from tablets in patients with coeliac
disease, radiation-​induced enteritis, and other gastrointestinal dis-
ease, and thyroxine absorption is impaired in coeliac disease.
Food
Food alters the rate and extent of absorption of many drugs. For ex-
ample, eggs impair iron absorption, and milk (or any calcium, alu-
minium, magnesium, or ferrous salt) impairs tetracycline absorption
by the formation of an insoluble chelate. Such effects are rarely im-
portant clinically, unless the drug has a very limited bioavailability,
when the effect can be highly significant. Hence, bisphosphonates
are only bioavailable in the complete absence of food. Grapefruit
juice, and some other citrus juices such as Seville orange and pomelo,
can markedly increase the bioavailability of some drugs (Fig. 2.6.4).
These include antihistamines (terfenadine), simvastatin, calcium
channel blockers (e.g. felodipine, nifedipine, and verapamil), im-
munosuppressive drugs (ciclosporin A, sirolimus, and tacrolimus),
and PDE5 inhibitors (e.g. sildenafil and vardenafil).
Coadministered drugs
Drugs affecting gastric emptying can affect absorption of coad­
ministered drugs as mentioned earlier. Anion exchange resins (bile
acid sequestrants) used for lipid management avidly bind certain
drugs. The resulting reduction in drug absorption is clinically sig-
nificant for warfarin, thyroxine, thiazides, and digoxin when taken
with cholestyramine. The newer resin, colesevalam, also affects
ciclosporin and sulphonylurea absorption. Hence it is usually re-
commended these resins are taken at least an hour after the drugs
or the interacting drugs taken at least 4 hours after the resin. This
strategy can of course be put to good effect in drug overdose, where
absorption can often be prevented by adsorbing the drug to swal-
lowed activated charcoal (see Chapter 10.4.1).
First-​pass metabolism
First-​pass metabolism is metabolism that occurs before the drug en-
ters the systemic circulation. This may happen in the gut lumen itself
(e.g. with benzylpenicillin or insulin), the gut wall (tyramine, chlor-
promazine), the lungs (various amines and inhaled glucocorticoids),
and the liver, which is the most important site.
Many drugs undergo first-​pass metabolism in the liver. It is
substantial for certain drugs, including cocaine, desipramine, lig-
nocaine, pethidine, morphine, nicotine, nitroglycerin (and other
organic nitrates), propranolol, and verapamil.
When first-​pass metabolism results in the formation of com-
pounds with less pharmacological activity than the parent com-
pound, the drug’s efficacy is lower after oral than intravenous
administration. In some cases, metabolism is so extensive that oral
therapy is impossible. However, such a drug given sublingually, rec-
tally, or transdermally, can bypass the liver (see earlier paragraphs).
Distribution
Protein binding
Many drugs are bound to circulating proteins, usually albumin
(acidic drugs), but also globulins (hormones), lipoproteins, and
α1-​acid glycoprotein (basic drugs). Only free drugs (i.e. not protein
bound) can bind to cell surface receptors, cross cell membranes to
access intracellular drug targets, or distribute to other tissues where
they may be metabolized and excreted (e.g. by the kidney). Thus,
changes in protein binding can sometimes cause changes in drug
distribution. However, such changes are only important if the drug
is extensively bound to plasma proteins (>90%) and not widely dis-
tributed to other tissues. Phenytoin and warfarin are the two drugs
most frequently affected.
The binding of drugs to albumin may be changed in renal im-
pairment (the explanation for this is unknown), hypoalbuminaemia
(drug binding is reduced when the plasma albumin concentration
falls below 25 g/​litre), the last trimester of pregnancy (during which
protein binding is reduced partly because of hypoalbuminaemia),
and displacement by other drugs.
Because α1-​acid glycoprotein is an acute-​phase protein, its levels
are affected by trauma, surgery, inflammatory diseases (e.g. Crohn’s
disease and rheumatoid arthritis), and infections. Hence, the
binding of quinine is increased in malaria.
Tissue distribution
The extent of drug distribution to the tissues of the body varies
widely. The drugs with the widest distributions will have the lar-
gest volumes of distribution (see earlier paragraphs). Some water-​
soluble drugs are limited to one or more of the water compartments
(vascular and extra-​ and intracellular), while others are bound
30
25
20
15
10
5
0
0
2
4
6
8
10
12
24
Time (h)
Simvastatin (ng/ml)
Fig. 2.6.4  Effect of grapefruit juice (closed symbols) or water (open
symbols) taken daily for 3 days on the plasma levels after a single 40 mg
dose simvastatin. The effect relies on furanocoumarin components of
grapefruit juice that directly reduce P450 protein expression (CYP 3A4
principally) in the gut wall.
From Lilja JL, Neuvonen M, Neuvonen PJ (2004). Effects of regular consumption
of grapefruit juice on the pharmacokinetics of simvastatin. Br J Clin Pharmacol, 58,
56–​60, Copyright © 2004, John Wiley and Sons.
80
section 2  Background to medicine
extensively in tissues. The distribution of drugs to different tissues
is influenced by plasma-​protein binding, specific receptor sites in
tissues (e.g. the binding of cardiac glycosides to the Na+,K+-​ATPase
in cell membranes throughout the body), regional blood flow (well-​
perfused organs, such as the heart, kidneys, and liver accumulate
drugs more than poorly perfused organs, such as fat and bone),
lipid solubility (lipid-​soluble drugs enter tissues more readily than
charged compounds with poor lipid solubility), and active transport
across cell membranes (adrenergic neuron-​blocking drugs such as
guanethidine accumulate in noradrenergic nerve terminals through
the noradrenaline transporter, uptake-​1) or active transport across
epithelia.
The importance of epithelial transport has been appreciated with
the discovery of specific transporters. These fall into two families: ei-
ther solute carriers (SLC) or ATP-​binding cassette (ABC) trans-
porters. P-​glycoprotein is the best understood ABC transporter and
was first identified in tumour cells where it confers multidrug resist-
ance (MDR) by promoting the extrusion of many structurally unre-
lated anticancer drugs. But this drug efflux pump is widely expressed
in normal tissues such as small intestine (brush border membrane
of enterocytes), kidney (brush border membrane of proximal tubule
cells), liver (canalicular membrane of the hepatocytes), and at the
blood–​brain barrier (capillary endothelial cells). The latter explains,
for example, why drugs such as ivermectin, digoxin, loperamide,
and domperidone are rapidly pumped out of the cerebrospinal fluid,
so effectively excluding them from the brain.
In some diseases, drug distribution is altered for unknown
reasons. In renal failure, for example, the distribution of drugs such
as insulin and digoxin is decreased as well as their protein binding.
In cardiac failure, the distribution of some antiarrhythmic drugs,
such as disopyramide, is also reduced. Obesity and malnutrition
influence the distribution of drugs that are highly fat soluble (e.g.
inhalational anaesthetics).
Metabolism
Most drugs are metabolized in the liver. Examples of other sites are
suxamethonium in the plasma; insulin and vitamin D in the kid-
neys; and acetylcholine and catecholamines at their corresponding
synapses and nerve endings.
Drug metabolism occurs in two phases:
•	Phase I  metabolism involves chemical alteration of the basic
structure of the drug (e.g. by oxidation, reduction, or hydrolysis).
This results in free groups such as –​OH or –​NH2 that are con-
jugated in phase II. Many phase I reactions are catalysed by en-
zymes from the P450 family (see following paragraphs). Examples
of phase I reactions include the N-​demethylation of diazepam to
desmethyldiazepam, an active metabolite with a long duration of
action, and the oxidation of ethanol to acetaldehyde.
•	Phase II metabolism involves conjugation (e.g. by sulphation,
glucuronidation, methylation, or acetylation). Some drugs are
conjugated without prior phase I  transformation, while others
undergo phase I metabolism before conjugation can occur. The
products of conjugation are more water soluble and therefore
more easily cleared by the kidney or biliary system. They are
usually, although not always, pharmacologically inactive (e.g.
morphine-​6-​glucuronide is an active metabolite of morphine
that accumulates in renal failure). Examples of phase II reactions
are the glucuronidation of paracetamol and the N-​acetylation of
hydralazine and procainamide. A conjugated product may some-
times be further metabolized. For example, oestrogens are ex-
creted via the bile, deconjugated in the gut by bacteria, and then
reabsorbed (so-​called enterohepatic recycling).
The end result of drug metabolism is inactivation, but during
the process, compounds with pharmacological activity are often
formed. Alternatively, inactive drugs (prodrugs) may be metabol-
ized to active ones. The antiparkinsonian drug l-​dopa, for example,
is a typical prodrug. It has no action on dopamine receptors until it
enters the central nervous system and is metabolized to dopamine.
Other examples include diamorphine and codeine (which are rap-
idly metabolized to diacetylmorphine and morphine, respectively)
and many angiotensin-​converting enzyme (ACE) inhibitors (e.g.
enalapril is converted to its active form enalaprilat).
Some drugs also have toxic metabolites. For example, norpethidine
is a metabolite of pethidine that can cause fits, and acrolein is a
bladder irritant formed from cyclophosphamide. The normally
minor metabolic pathway for paracetamol forming N-​acetyl-​p-​
benzoquinone imine (NABQI) is important in overdose because
of the saturation of other detoxification pathways; this metabolite
causes the hepatotoxicity that follows paracetamol overdose.
Most of the phase I reactions in the liver are carried out by en-
zymes of the large cytochrome P450 superfamily of proteins. They
are all prefixed by ‘CYP’ and tens of different isoforms are recog-
nized in the human liver. The commonest is the CYP3A4 isoform
through which over 50% of marketed drugs are metabolized. Others
such as CYP2D6 have special importance because they are poly-
morphic, that is, they are encoded by CYP genes that have common
variants that affect enzyme activity. The 10 major isoforms and some
of the commoner drugs that are substrates for them are shown in
Table 2.6.1.
Certain drugs also selectively block CYP isoforms, which re-
duce the elimination of drugs through these enzymes. Examples
of these are cimetidine (1A2 and 2D6), amiodarone (2C9, 2D6,
and 3A4), fluoxetine (2D6), ketoconazole (3A4 and 2C19), and
ritonavir (3A4). Conversely, drug elimination through some CYP
pathways is enhanced by drugs that act as inducers that increase
enzyme activity by directly activating transcription of the corres-
ponding CYP genes. The classical inducers include phenobarbitone
(2B6, 3A4), rifampicin (2B6, 2C9, and 3A4), and antiepileptics such
as phenytoin and carbamazepine (both 3A4). Tobacco and ethanol
are also inducers, albeit for minor CYP isoforms (1A2 and 2E1,
respectively). A full list of CYP inhibitors and inducers (so-called
Flockhart table) is available at this link: https://drug-interactions.
medicine.iu.edu/Main-Table.aspx
Other factors that affect hepatic drug metabolism are hep-
atic blood flow (for drugs that are rapidly cleared), liver disease
(only important in extensive liver disease or when there is ar-
teriovenous shunting), and age. The metabolism of some drugs
is impaired in old people and in babies younger than about
6 months, particularly premature babies. In both cases, this is
due to reduced activity of the microsomal enzymes that includes
CYP and non​CYP enzymes. For example, in neonates uridine
diphosphoglucuronosyl transferase (EC 2.4.1.7), which conju-
gates chloramphenicol, is relatively inactive; neonates eliminate
chloramphenicol slowly, and may suffer peripheral circulatory
2.6  Principles of clinical pharmacology and drug therapy
81
collapse (the ‘grey syndrome’) when given the drug in weight-​
related doses that do not harm adults.
Excretion
The kidney is a major route of drug excretion. Other, usually minor,
routes include the lungs (important for paraldehyde and gaseous
anaesthetic gases), breast milk, sweat, tears, and genital secretions
(alarming if the orange-​red discoloration caused by rifampicin is
not mentioned), saliva, and bile. Excretion in bile can be prominent
for some drugs and this can lead to the reabsorption of some of the
excreted compounds from the gut—​a process called enterohepatic
recycling. This recycling affects drugs such as oestradiol, rifampicin,
and tetracyclines, and can substantially prolong their duration in
the body. Drugs excreted in bile as water-​soluble conjugates (after
phase II metabolism—​see earlier) may also be deconjugated by gut
bacteria to release parent drug so facilitating enterohepatic recycling
(e.g. chloramphenicol, digitoxin, indomethacin, and valproic acid).
Renal excretion of drugs involves three separate processes:
glomerular filtration, passive tubular reabsorption, and active
tubular secretion (often through an SLC transporter). Thus: Total
Table 2.6.1  The main cytochrome P450 isoforms (in bold) and their common substrate drugs
1A2
2C9
2D6
3A4,5,7
Clozapine
NSAIDs
β-​Blockers
Macrolide antibiotics
Imipramine
Diclofenac
Metoprolol
Clarithromycin
Mexiletine
Ibuprofen
Propafenone
Erythromycin
Naproxen
Naproxen
Timolol
NOT azithromycin
Theophylline
Celecoxib
Antidepressants
Antiarrhythmics
2B6
Sulphonylureas
Amitriptyline
Quinidine
Bupropion
Glibenclamide
Clomipramine
Benzodiazepines
Cyclophosphamide
Glipizide
Desipramine
Diazepam
Efavirenz
Tolbutamide
Duloxetine
Midazolam
Ifosfamide
Angiotensin II blockers
Fluoxetine
Triazolam
Methadone
Irbesartan
Imipramine
Immune modulators
2E1
Losartan
Paroxetine
Ciclosporin
Anaesthetics
NOT valsartan
Venlafaxine
Tacrolimus
Enflurane
NOT candesartan
Antipsychotics
HIV protease inhibitors
Halothane
Phenytoin
Haloperidol
Indinavir
Isoflurane
Sulfamethoxazole
Risperidone
Ritonavir
Methoxyflurane
Tamoxifen
Opiates
Saquinavir
Sevoflurane
Warfarin
Codeine
Antihistamines
2C19
oxycodone
Astemizole
Paracetamol
Proton pump inhibitors
Tramadol
Chlorpheniramine
Ethanol
Omeprazole
Dextromethorphan
Calcium channel blockers
2C8
lansoprazole
Flecainide
Amlodipine
Paclitaxel
Pantoprazole
Mexiletine
Diltiazem
Torsemide
Antiepileptics
Ondansetron
Felodipine
Amodiaquine
Diazepam
Tamoxifen
Nifedipine
Repaglinide
Phenytoin
Verapamil
Phenobarbitone
Statins
Atorvastatin
Simvastatin
NOT pravastatin
Others
Amitriptyline
Citalopram
Clomipramine
Clopidogrel
Cyclophosphamide proguanil
NOT rosuvastatin
Imatinib
Methadone
Pimozide
Quinine
Sildenafil
Source: http://​www.medicine.iupui.edu/​CLINPHARM/​ddis/​clinical-​table
82
section 2  Background to medicine
renal clearance = Clearance by filtration + Clearance by secretion −
Retention by reabsorption.
If a drug is mainly metabolized to inactive compounds, renal func-
tion will have little impact on its elimination. However, if the drug
itself or an active metabolite is excreted unchanged via the kidneys,
changes in renal function will substantially affect its elimination.
Glomerular filtration
All drugs are filtered at the renal glomerulus, although molecules
larger than 2 kDa are not freely filtered (e.g. insulin concentration
in the ultrafiltrate is 0.89 × plasma) and proteins as large as albumin
(69 kDa) are not filtered at all by a normal glomerulus. The extent
of filtration is directly proportional to the glomerular filtration rate
(GFR = 120 ml/​min) and to the fraction of unbound drug in the
plasma (fu), that is, rate of clearance by filtration = fu × GFR.
If the total renal clearance of a drug is equal to fu × GFR, then it is
cleared principally by filtration (it could be affected by the other two
mechanisms, secretion and reabsorption, but they would have to
cancel each other out). Examples of drugs whose clearance is similar
to the GFR (after correction for protein binding) are digoxin, gen-
tamicin, vancomycin, methotrexate, and ethambutol. As creatinine
is cleared mainly by filtration, the creatinine clearance is useful in
estimating the clearance rates of these drugs; although there is some
secretion, which explains its tendency to overestimate GFR.
Passive tubular reabsorption
Drugs are passively reabsorbed by the renal tubules. The elimination
of drugs with very low rates of renal clearance (i.e. approaching
urine flow rate, or about 1–​2 ml/​min) will be significantly affected
by changes in urine flow rate (because a doubling of flow rate will in-
crease their rate of clearance by 1–​2 ml/​min, i.e. twofold). However,
for weak acids and weak bases that are non​polar in their union-
ized form, the main factor affecting passive reabsorption is the pH
of the renal tubular fluid (which can vary between 4.5 and 8), be-
cause the extent of their ionization (and therefore of their passive
reabsorption) depends on the urine pH in relation to the pKa of the
drug. For example, in an alkaline urine, weak acids with a pKa below
7.5, such as salicylate (pKa 3), are more highly ionized, and there-
fore less well reabsorbed. The reverse is true for weak bases with a
pKa greater than 7.5, such as methamphetamine (pKa 10), whose re-
absorption is reduced, and whose clearance is therefore enhanced,
by an acid urine. These principles are put to good use in the treat-
ment of overdose (see Chapter 10.4.1). Renal failure alters passive
reabsorption indirectly, by changing both urine flow rate and pH.
Active tubular secretion
If the renal clearance of a drug exceeds GFR then there must be
active secretion of the drug into the renal tubule. The active trans-
port of organic anions and cations is dependent on specific trans-
porter proteins (SLC and ABC transporters) in the cells lining
the proximal tubule (detailed in https://www.ncbi.nlm.nih.gov/
pubmed/28210973). There are broadly two functional groups: or-
ganic anion and organic cation transporters. The organic anion
transporters are responsible for the secretion of many β-​lactam
antibiotics, NSAIDs, antivirals, and ACE inhibitors, as well as
acidic glutathione and glucuronide-​conjugated drug metabolites.
Probenecid is a generic inhibitor of anion transport and anionic
transport capacity can be measured by p-​aminohippuric acid (PAH)
excretion. Substrates for the organic cation transporters include en-
dogenous cations (e.g. guanidine, choline, N-​methylnicotinamide,
and monoamine neurotransmitters), cationic toxins (e.g. 1-​methyl-​
4-​phenylpyridinium, MPP+), and cationic drugs (e.g. cimetidine,
procainamide, quinidine, vecuronium, cardiac glycosides). Both ci-
metidine and trimethoprim can affect cationic transport. Blockade
of kidney drug transporters explains several drug–​drug interactions
(see following paragraphs).
Nonlinear kinetics
Most drugs show linear kinetics (i.e. an increase in dose will cause
a similar-​fold increase in plasma concentration). There are excep-
tions, however, and when a drug shows nonlinear behaviour (e.g. a
twofold increase in dose produces a tenfold increase in plasma level)
it implies that some aspect of the drug’s elimination pathway has
saturated. Some tricyclic antidepressants and the selective serotonin
reuptake inhibitor (SSRI) paroxetine show this behaviour. But the
best-​known example is phenytoin (see Fig. 2.6.5), whose oxidation
through CYP 2C19 in the liver is saturated within the therapeutic
range. Inspection of the curves in Fig. 2.6.5 shows how small in-
crements in the dose of phenytoin can produce large increments in
its plasma level and hence neurotoxicity. In fact, many drugs taken
in overdose saturate their elimination pathways, and this is clearly
shown by drugs such as ethanol, salicylate, and paracetamol.
The pharmacodynamic process
Pharmacodynamics addresses the question ‘How does a drug pro-
duce its pharmacological effects?’ Of course, this concerns both ad-
verse as well as therapeutic effects.
Drugs have many different mechanisms of action but the
overwhelming majority are mediated through receptors, which they
either block or activate.
0
0
100
200
300
Daily dose of PHENYTOIN (mg)
400
500
600
25
50
75
Serum [PHENYTOIN]  (μmol/litre)
100
125
150
175
200
Fig. 2.6.5  Non​linear kinetics. The plasma level achieved after repeated
phenytoin dosing is shown versus the corresponding phenytoin dose for
three hypothetical patients. The plots are not straight lines but rise steeply
within the therapeutic range (between the dashed lines) as the CYP2C19
metabolic pathway for phenytoin saturates.
2.6  Principles of clinical pharmacology and drug therapy
83
Actions via direct effects on receptors
Receptors are proteins situated either in cell membranes or within
the cellular cytoplasm. For each receptor type there is usually an en-
dogenous molecule (or ligand) that binds to the receptor. There are
broadly four types of receptors:  G protein-​coupled receptors, ion
channels, nuclear receptors, and tyrosine kinase receptors. Drug mol-
ecules target the same binding site on the receptor as the endogenous
ligand and may activate the receptor, block the binding of the en-
dogenous ligand, or a have mixture of both effects. These actions are
referred to as agonism, antagonism, and partial agonism, respectively.
This terminology is made clearer by considering these actions in
terms of the µ opioid receptor. Morphine acts as an agonist at µ opioid
receptors to cause analgesia. This µ receptor activation mimics the ac-
tion of endogenous analgesic peptides such as enkephalin and endor-
phin. Compare this with naloxone, which behaves as an antagonist
at the same receptor and blocks its activation by either morphine or
other opiates. Importantly, in the absence of an agonist, a pure antag-
onist such as naloxone has no pharmacological action through the µ
opiate receptor. This contrasts with buprenorphine that behaves as a
partial agonist at the µ receptor. Hence, in the absence of an agonist
such as morphine, buprenorphine is able to activate the µ receptor, but
it is not able to produce an analgesic effect as large as morphine (this
explains the ‘ceiling’ effect seen clinically). Yet, in the presence of a full
agonist such as morphine, buprenorphine can block µ receptor activa-
tion (this explains why buprenorphine can trigger opiate withdrawal).
Receptor subtypes
In many cases, a receptor will have subtypes, and drugs may have sub-
type selectivity. For example, there at least three subtypes of opioid
receptors: µ, κ, and δ. The µ and δ are both involved in analgesia,
gastrointestinal motility, and respiratory depression, and κ is involved
in analgesia, sedation, and miosis. These receptors are variably distrib-
uted in the nervous system. Most opiates act at µ receptors, but none is
completely selective and they may act at other subtypes.
Long-​term effects of drugs at receptors
During long-​term therapy, the effects of a drug may be altered
by adaptive responses, usually accompanied by either increases
(‘upregulation’) or decreases (‘downregulation’) in receptor num-
bers. Such changes may explain both the therapeutic and adverse
effects of drugs. Examples include:
•	The therapeutic response to antidepressants (e.g. an SSRI) that may
involve changes in receptors within the central nervous system
secondary to the increased synaptic levels of neurotransmitter
caused by these drugs. This probably explains why the therapeutic
response to antidepressants takes a few weeks to emerge.
•	The way in which the response to l-​dopa in Parkinson’s disease
changes during long-​term administration (e.g. producing the
‘on–​off’ effect).
•	 Withdrawal syndromes that may occur because long-​term changes
become unopposed when the drug is withdrawn (e.g. after the long-​
term use of opiates or benzodiazepines).
Actions via direct effects on second messengers
When an agonist stimulates a membrane-​bound receptor, its
effect is usually signalled in one of two ways:  either through a
so-​called second messenger (e.g. cAMP, diacylglycerol, or inositol
trisphosphate) or by changing the activity of an ion channel linked
to the receptor. Some drugs may act by affecting second messen-
gers directly. For example, some drugs block phosphodiesterases
that normally metabolize the second messengers cAMP and cGMP.
Several types of phosphodiesterases regulate cAMP levels and theo-
phylline and caffeine work by inhibiting them non​specifically. In
contrast, type 5 phosphodiesterase (EC 3.1.4.17) specifically de-
grades cGMP in smooth muscle and is selectively inhibited by drugs
such as sildenafil and vardenafil.
Actions via indirect alterations of the effects
of endogenous receptor ligands
Drugs may oppose the physiological effects of the endogenous
ligand. For example, glucagon is a physiological antagonist of the
actions of insulin; hence its use to treat hypoglycaemia. Other drugs
act indirectly on a receptor by altering the levels of the endogenous
ligand for that receptor. They can do this in a number of ways:
Increase in endogenous release
Some drugs enhance release of the endogenous ligand. For example,
amphetamine and tyramine increase the release of dopamine and
noradrenaline, respectively, from nerve terminals.
Prevention of endogenous release or synthesis
The release of many neurotransmitters from nerve terminals is
regulated by inhibitory receptors activated by the neurotransmitter
itself. Hence, α2-​receptors on noradrenergic nerve terminals re-
duce noradrenaline release into the synapse. Clonidine and α-​
methyldopa reduce the release of noradrenaline by activating these
receptors (α-​methyldopa actually works through its metabolite
α-​methylnoradrenaline) and is the basis of their antihypertensive
action.
Drugs can also reduce the production of the endogenous ligand
rather than affect release of a preformed store of the ligand. For ex-
ample, angiotensin II (which activates the angiotensin II receptor
to cause vasoconstriction and aldosterone release) is produced
enzymatically from angiotensin I. ACE inhibitors block the con-
verting enzyme (cleaving angiotensin I to angiotensin II) and renin
inhibitors (e.g. aliskiren) block the enzymatic cleavage of angio-
tensin I from angiotensinogen.
Inhibition of endogenous reuptake
Many neurotransmitters are pumped back into their corresponding
nerve terminals. The transporters that carry out this reuptake are
targeted by certain psychoactive drugs. Many antidepressants, for
example, block the reuptake of noradrenaline (such as amitrip-
tyline), or 5-​hydroxytryptamine (serotonin, 5HT) (such as fluox-
etine), or both (such as venlafaxine). Reuptake of γ-​aminobutyric
acid (GABA) is blocked by the anticonvulsant tiagabine.
Inhibition of endogenous metabolism
Drugs can also increase the effect of an endogenous ligand by
blocking its metabolism. Examples of these include: cholinesterase
inhibitors for acetylcholine (e.g. neostigmine), vigabatrin for GABA,
and monoamine oxidase inhibitors (MAO) inhibitors for catechol-
amines and 5HT (e.g. tranylcypromine).
84
section 2  Background to medicine
The MAO inhibitors have some selectivity for the two forms of
MAO (types A and B), which explains their use both as antidepres-
sants and antiparkinsonian agents. For example, selegiline select-
ively inhibits monoamine oxidase type B. This results in blockade of
dopamine metabolism in the central nervous system and enhances
the action of l-​dopa in parkinsonism. However, because MAO in
the gut and liver is predominantly type A, selegiline does not pro-
duce the ‘cheese reaction’ (due to tyramine and other amines, see
following paragraphs) seen with non​selective MAO inhibitors (e.g.
tranylcypromine). In contrast, the antidepressant moclobemide se-
lectively inhibits MAO A. Moclobemide is also unusual in being a
reversible inhibitor of MAO. All other MAO inhibitors block MAO
irreversibly, so that recovery requires synthesis of new enzyme.
This explains the slow offset of their action (7–​10 days) and why
moclobemide does not produce a ‘cheese reaction’.
Actions by inhibiting the movement of ions
Since cations (such as sodium, potassium, and calcium) and anions
(such as chloride and iodide) have so many important roles in the
maintenance of normal cellular function, inhibition of their trans-
port is an important mechanism of drug action. This movement of
cations and anions across membranes can be either through trans-
porters or channels.
Diuretics
Most diuretics reduce sodium reabsorption in the renal tubules by
targeting specific transporters or ion channels. Hence, loop diuretics
(e.g. furosemide and bumetanide) block the Na–​K–​Cl transporter
(NKCC2) in the ascending limb of Henle’s loop. Thiazide diuretics
inhibit Na–​Cl cotransporter (NCC) in the distal convoluted tubule
and potassium-​sparing diuretics (amiloride and triamterene) block
the epithelial sodium channels in the collecting ducts (ENaC).
Calcium channel antagonists
The calcium antagonists, such as verapamil, diltiazem, and the
dihydropyridines (e.g. nifedipine), inhibit the transport of calcium
via voltage-​operated calcium channels (of the l-​type). They are able
to exert different effects in different tissues (e.g. verapamil slows atrio-
ventricular nodal conduction in the heart but nifedipine does not)
because of separate binding sites within the l-​type calcium channel.
Drugs acting on potassium channels
The potassium permeability of cell membranes affects their mem-
brane potential and is controlled by potassium (K) channels. They
are a very large and diverse group of ion channels and drugs may
either open or close (or block them).
Drugs that open potassium channels include vascular smooth
muscle relaxants, such as minoxidil and nicorandil (targeting KATP
channels). The same KATP channels are closed by sulphonylureas
such as gliclazide. Of the many potassium channels in the heart,
the iKr channel plays a key role as it is involved in repolarization of
the myocyte. Blockade of the iKr channels explains why many drugs
cause prolongation of the QT interval and are hence arrhythmogenic
(e.g. terfenadine).
Drugs acting on chloride channels
Intracellular chloride plays a key role in neuronal excitability. Hence,
the inhibitory effect of the neurotransmitter GABA on neurons
is due to chloride entry into neurons through GABA-​activated
chloride channels. Progabide behaves as a GABA agonist and
GABA-​activated chloride currents are increased by benzodiazepines
and barbiturates. The antiseizure activity of these drugs is directly
attributed the neuronal hyperpolarization that results.
Some anaesthetic agents also probably rely on effects on chloride
currents for their general anaesthetic effect. They form a diverse
group of agents, such as the halogenated hydrocarbons (halothane,
trichloroethylene), and non​halogenated agents (nitrous oxide,
ether, cyclopropane), and were previously thought to influence
neuronal excitability by non-​specific effects on the lipid phase of
the cell membrane, changing its biophysical properties and hence
the kinetics of ion channels. More recent evidence has shown that
they specifically affect currents through GABA/​glycine-​coupled ion
channels. Others (such as ketamine and nitrous oxide) antagonize
the excitatory NMDA-​coupled ion channel.
Actions via enzyme inhibition
Drugs often act by directly inhibiting enzymes. Inhibitors of cholin-
esterase and MAO (see earlier paragraphs), for example, have been
in therapeutic use for more than half a century.
The metabolism of purines involves the oxidation of xanthine and
hypoxanthine to uric acid by xanthine oxidase. Hence, blockade of
this enzyme with allopurinol prevents excessive uric acid produc-
tion during tumour lysis and reduces the frequency of gout.
The cardiac glycosides act by inhibiting the Na/​K pump. This
changes the distribution of sodium across excitable membranes es-
pecially in the heart: and the secondary change in intracellular cal-
cium causes the cells’ contractility to increase.
Other drugs that act via enzyme inhibition include warfarin
(vitamin K epoxide reductase), aspirin, and other NSAIDs, targeting
the cyclo-​oxygenase (COX) enzymes involved in prostaglandin
synthesis, ACE inhibitors, disulfiram (alcohol dehydrogenase),
some anticancer drugs such as cytarabine (DNA polymerase), and
imatinib (chronic myelogenous leukaemia-​specific tyrosine kinase),
and some anti-​infective drugs (bacterial or viral enzymes; e.g. tri-
methoprim inhibits bacterial dihydrofolate reductase, the quin-
olones inhibit bacterial DNA gyrase, and zidovudine and didanosine
inhibit the reverse transcriptase of HIV).
Danazol and stanozolol are examples of drugs that inhibit an en-
zyme indirectly—​they stimulate the production of an inhibitor of
C1 esterase and are used to treat hereditary angio-​oedema, in which
there is reduced plasma activity of the inhibitor.
Actions via enzyme activation or direct
enzymatic activity
Some drugs either activate enzymes or are enzymes themselves.
The clotting and fibrinolytic factors are enzymes, and certain drugs
that act on clotting and fibrinolysis do so by increasing their activity.
Heparin acts by activating antithrombin III. The thrombolytic drugs
streptokinase, alteplase, and tenecteplase activate plasminogen.
Deficiencies of clotting factors can be treated by replacing defi-
cient enzymes of the clotting pathway (e.g. factor VIII in patients
with haemophilia and fresh frozen plasma in warfarin toxicity).
Pancreatic enzymes are used in treating malabsorption in patients
with chronic pancreatic insufficiency.
Rasburicase is used to prevent the hyperuricaemia that accom-
panies the tumour lysis syndrome during the treatment of acute
2.6  Principles of clinical pharmacology and drug therapy
85
leukaemias and lymphomas. It is recombinant urate acid oxidase
that directly catalyses the breakdown of uric acid to the more soluble
allantoin.
Actions via other miscellaneous effects
Chelating agents
Drugs that chelate metals can be used to hasten their removal from
the body (see Chapter  10.4.1). Calcium sodium edetate chelates
many divalent and trivalent metals and is used to treat poisoning,
particularly with lead. Dimercaprol chelates some heavy metals
and is used to treat mercury and gold poisoning. Desferrioxamine
chelates iron and is used in treating iron poisoning and the iron
overload that occurs with repeated blood transfusion (as in thal-
assaemia). Penicillamine chelates copper and is used in treating
hepatolenticular degeneration (Wilson’s disease); it is also used to
complex cystine and thus prevent renal damage in cystinuria.
Osmotic diuretics
Mannitol is freely filtered at the glomerulus but the renal tubules
reabsorb relatively small amounts. It therefore increases the concen-
tration of osmotically active solute in the tubular fluid, the subse-
quent influx of water massively augments urine flow rates.
Replacement of vitamins and minerals
Some drugs are used simply to replace deficiencies (e.g. ferrous salts
in iron deficiency anaemia and hydroxocobalamin (vitamin B12) in
vitamin B12 deficiency).
Stereoisomerism and drug action
Stereoisomerism (chirality) of organic compounds is due to asym-
metry in one or more of their atoms (usually carbon), resulting
in two or more three-​dimensional structures (enantiomers) that
cannot be superimposed on each other.
Several different terminologies are used to describe the two chiral
partners: R and S (from the Latin rectus = right and sinister = left),
(+) -​d and (–​)-​l and d and l (from the Latin dexter = right and
laevus = left). Examples of drug enantiomers are R-​warfarin and
S-​warfarin, d-​glucose (dextrose) and l-​glucose (laevulose), and d-​
propranolol and l-​propranolol.
Of all synthetic drugs used in clinical practice, almost 50% are
chiral and about 90% of those are marketed in their racemic form (i.e.
as an equal mixture of the two enantiomers). Examples include d,l-​
propranolol and R,S-​warfarin. Naproxen is one of the few examples
of a synthetic compound that is marketed as a single enantiomer
(hence an enantiopure drug). In contrast, naturally occurring and
semisynthetic compounds are almost all chiral and almost all are
marketed as a single isomer. Examples include d-​glucose (dextrose)
and the naturally occurring amino acids (e.g. l-​dopa), l-​thyroxine,
and l-​noradrenaline.
Enantiomers often have different pharmacological actions. For
example, l-​propranolol is a β-​blocker, whereas d-​propranolol has
membrane-​stabilizing activity like that of local anaesthetics; l-​
sotalol is a β-​blocker, whereas d-​sotalol has antiarrhythmic effects
like those of amiodarone.
Sometimes these differences between the pharmacology of en-
antiomers can separate the therapeutic and adverse effects of a ra-
cemate. For example, R-​thalidomide is responsible for the sedative
effects of R,S-​thalidomide, while the teratogenic effect resides in the
S enantiomer. However, this cannot be exploited clinically to limit
thalidomide toxicity, because the two enantiomers spontaneously
interconvert in the body. This so-​called ‘chiral inversion’ also occurs
in some NSAIDs such as ibuprofen.
The interaction of a drug with its target (such as a receptor or en-
zyme) is often stereo-​selective, so enantiomers may show marked
differences in potency. For example, S-​warfarin is some five times
more potent an anticoagulant than R-​warfarin, and S-​citalopram
is 30 times more potent as an SSRI than R-​citalopram. In some
cases, one enantiomer is completely inactive therapeutically (e.g.
levofloxacin is the active antimicrobial enantiomer in ofloxacin with
dextrofloxacin being completely inactive).
The enantiomers may have different pharmacokinetics. For ex-
ample, the half-​lives of S-​warfarin and R-​warfarin average around
30 h and 50 h and they are metabolized to 7-​hydroxywarfarin and
warfarin alcohols, respectively. This is important in some drug
interactions with warfarin, because drugs inhibiting warfarin me-
tabolism (such as metronidazole) primarily affect the more potent
enantiomer, S-​warfarin. Omeprazole provides a further example.
The single enantiomer of omeprazole, S-​omeprazole, has almost
twice the bioavailability of R-​omeprazole. Hence, esomeprazole
(S-​omeprazole) gives plasma levels of omeprazole 70 to 90%
higher than racemic omeprazole. However, both enantiomers are
converted to the same active intermediate and since this lacks a
chiral centre the enantiomers are equiactive if they achieve the
same plasma level.
Because of these caveats, despite the appeal of prescribing single-​
drug enantiomers, in most instances the single enantiomer has
failed to show a substantial clinical advantage over the racemate.
When comparing racemates to single enantiomers in clinical trials,
it is critical they are compared at comparable doses of the active
enantiomer.
The therapeutic process
The question associated with the therapeutic process is, ‘Is the
pharmacological effect being translated into a therapeutic effect?’
Translation of pharmacological effect into therapeutic
effect during short-​term therapy
The short-​term therapeutic and toxic effects of drugs occur as a re-
sult of the pharmacological actions discussed earlier. However, the
translation of molecular and cellular pharmacological effects into
the therapeutic or toxic effect is not a simple process, and involves
several translational stages at different pharmacological and physio-
logical levels.
Take, for example, the action of salbutamol, a β2-​adrenoceptor
agonist, in the treatment of asthma. Salbutamol stimulates bronchial
β2-​adrenoceptors, and so increases the activity of adenylate cyclase;
this is its pharmacological effect at the molecular level. The increase
in adenylate cyclase activity raises the intracellular concentration of
cAMP that leads to relaxation of the bronchial smooth muscle cells;
this is the cellular effect. This leads to dilatation of the bronchioles,
reduces the resistance to air flow in the bronchial tree, and improves
gas exchange; this is the effect on tissue and whole organ function.
86
section 2  Background to medicine
Finally, the patient feels less breathless and oxygen saturation may
improve; this is the desired clinical effect.
This analysis of the short-​term effects of a drug teaches us sev-
eral things about drug action: how drug action may be modified;
how therapeutic and adverse effects may be mediated via different
pharmacological effects; the relation between the pharmacological
effects of a drug and the rate of onset or duration of its therapeutic
action; and drug–​disease interactions.
How to modify drug action
It is often possible to modify drug action positively or negatively. For
example, a methylxanthine derivative, such as theophylline, which
blocks cAMP breakdown by its inhibition of phosphodiesterase,
should potentiate the action of salbutamol. This turns out to be both
a beneficial and an adverse interaction—​beneficial because theo-
phylline enhances the therapeutic action of salbutamol, adverse
because it enhances the hypokalaemia (by stimulating Na+/​K+-​
ATPase) and tachycardia (by activating cardiac β-​adrenoceptors)
that salbutamol causes.
Different pharmacological actions may mediate therapeutic
and adverse or other effects
Some drugs have more than one molecular mechanism of action,
and different therapeutic effects of a drug may result from different
actions. For example, tetracycline acts against bacteria by interfering
with their protein synthesis, but in acne, it helps by interfering with
sebum production in sebaceous glands.
Similarly, a therapeutic effect may be brought about by one
pharmacological action and an adverse effect by another. For ex-
ample, the antibacterial action of erythromycin is due to inhibition
of bacterial ribosomal function, but its gastrointestinal side effects
are due to activation of motilin receptors in the human gut. Other
macrolides such as azithromycin are much weaker motilin mimetics
and hence produce much less vomiting and diarrhoea.
A drug may also produce therapeutic and adverse effects through
the same molecular mechanism but in different tissues. For example,
the inhibition of β2-​adrenoceptors within muscle spindles by pro-
pranolol reduces benign essential tremor, but blocking the same
receptors in the lung causes bronchoconstriction in susceptible in-
dividuals and impairs glycogenolysis in the liver (which can delay a
diabetic’s recovery from hypoglycaemia).
The relation between the pharmacological actions of a drug
and the rate of onset and duration of its effects
The rate of onset of a drug’s effects depends not only on its pharma-
cokinetics (i.e. the time it takes for a therapeutic concentration of
drug to build up at its site of action), but also on how long it takes
for the full pharmacodynamic sequence of events to unroll. In the
case of salbutamol, the time between β2-​adrenoceptor stimulation
and bronchodilatation is of the order of a few minutes. However,
for other drugs, the sequence of events takes much longer. For ex-
ample, corticosteroids bind to an intracellular receptor protein in
the cytoplasm of target cells to form a steroid–​receptor complex.
This complex translocates to the nucleus, where it binds to regu-
latory sequences on target genes to cause RNA transcription. The
induction of de novo protein synthesis by RNA transcription and
translation takes several hours, explaining the slow onset of cortico-
steroid effects.
Similarly, the duration of action of a drug is related not only to
the time it takes for the drug to be cleared from the body, but also
to the duration of its pharmacological effects. For example, aspirin
inhibits COX-​1 by acetylating a serine moiety at the active site of
the enzyme. As platelets cannot synthesize new protein, the re-
covery from aspirin requires the appearance of new platelets from
the marrow. This process can take 7 to 10 days to restore peripheral
platelet function.
Drug–​disease interactions
Because of the complex links between the pharmacological effects of
a drug and its therapeutic or adverse effects, the pathophysiology of
the disease being treated, or of other coincidental diseases, can vari-
ously impact the way in which the pharmacological effect is trans-
lated into a therapeutic effect.
The use of digoxin in cardiac failure exemplifies this. Digoxin
inhibits the activity of the membrane-​bound Na+,K+-​ATPase.
Pump inhibition increases the intracellular concentration of so-
dium, which secondarily raises intracellular calcium to produce
a positive inotropic effect. The various steps in this process are af-
fected by drug–​disease interactions. Hypokalaemia, for example, is
a common side effect of diuretic use to manage fluid overload in
heart failure. Low extracellular potassium increases pump block by
raising the affinity of digoxin for the sodium pump. This risks cal-
cium overload and hence digoxin toxicity. Coincident diseases also
affect digoxin’s therapeutic effect. In hyperthyroidism, the Na+,K+-​
ATPase is upregulated, resulting in resistance to the inhibitory ef-
fects of digoxin. So, increasing the dose may cause digoxin toxicity
without ever producing a therapeutic effect. In patients with chronic
cor pulmonale, tissue hypoxia may also lead to Na/​K pump inhib-
ition and cardiac arrhythmias without increasing myocardial con-
tractility. In patients with hypertrophic obstructive cardiomyopathy,
although digoxin increases the rate of myocardial contractility, a rise
in cardiac output is prevented because left ventricular outflow re-
mains obstructed.
Thus even when it can be shown that a drug is having its ex-
pected action at a particular pharmacological or physiological
level, it cannot automatically be assumed that it will have a thera-
peutic effect.
Interactions with circadian rhythms (chronopharmacology)
Most physiological functions follow a circadian rhythm, so some
drug effects are liable to differ at different phases of the rhythm. In
some instances, the difference is dramatic. The timing of cortico-
steroid dosing is a good example. Peak ACTH release from the pi-
tuitary occurs at night leading to plasma cortisol peaks at around
08.00 a.m. when ACTH levels are at trough. So, exogenous cortisol
in the morning will have no impact on ACTH release. But given in
the evening it will completely inhibit night-​time ACTH. For this
reason, cortisol given once daily in the morning causes much less pi-
tuitary inhibition than the same dose given in the evening, or spread
throughout the day.
The cholesterol-​lowering effect of some statins is also affected by
the time of administration because cholesterol synthesis has a cir-
cadian rhythm with most synthesis occurring at night. Short-​acting
statins are hence more effective if given at night (e.g. simvastatin,
half-​life c.3 h). For longer-​acting statins there is no discernible dif-
ference, as the drug level will still be high enough during the night
2.6  Principles of clinical pharmacology and drug therapy
87
even with morning dosing (e.g. atorvastatin half-​life c.20 h with even
longer-​lived active metabolites).
Translation of pharmacological effect into therapeutic
effect: Long-​term therapy
During prolonged therapy, adaptation may develop to the short-​
term pharmacological effects of the drug with several consequences.
Therapeutic effects through adaptation
In immunization, by adapting to an initial immunological challenge,
the immune system develops the ability to respond to a subsequent
similar challenge (e.g. tetanus immunization).
Although tricyclic antidepressants rapidly inhibit reuptake of
noradrenaline and 5-​HT in the brain, the therapeutic effect of these
drugs takes 1 to 2 weeks to become evident. In certain brain regions,
there is adaptation to the increased concentrations of neurotransmit-
ters in the synaptic cleft with reduction in numbers of postsynaptic
receptors. Part of this adaptive downregulation probably explains
their therapeutic effects.
Tolerance: Increasing ineffectiveness of therapy
Tolerance is a state of decreased responsiveness to a drug, resulting
from previous exposure, either to the same drug or to one with
similar short-​term effects.
For example, it can develop to the vasoconstricting effects of
ephedrine nose drops, used to treat vasomotor rhinitis: as ephedrine
acts by releasing noradrenaline from sympathetic nerve endings,
noradrenaline depletion will reduce the effectiveness of ephedrine.
Patients taking long-​term glyceryl trinitrate, particularly from
transdermal patches, become tolerant to its acute effects. To avoid
this, a patch should be applied for no longer than 18 h. This effect
probably reflects depletion of tissue sulphydryl groups by oxida-
tion to disulphide groups. Some oral preparations of isosorbide
mononitrate are formulated to release their contents over 18 h for
the same reason.
Physiological tolerance by homoeostatic mechanisms
Secondary hyperaldosteronism is a physiological response to so-
dium loss produced by loop or thiazide diuretics. The enhanced
potassium excretion that it causes may be reduced by using a
potassium-​sparing diuretic (e.g. amiloride) or the aldosterone an-
tagonist spironolactone.
Another type of physiological tolerance occurs in patients given
the carbonic anhydrase inhibitor acetazolamide. This causes both
a diuresis and a kaluresis. However, these effects are only sustained
for a matter of days, because the large amounts of bicarbonate lost
from the kidney causes a metabolic acidosis. Interestingly, topical
carbonic anhydrase inhibitors do not show the same tolerance when
they are used in chronic glaucoma.
Metabolic tolerance
Metabolic tolerance results from faster metabolic clearance of
the drug. The commonest cause is induction of hepatic P450 en-
zymes by drugs such as barbiturates, phenytoin, carbamazepine,
or rifampicin. Occasionally, drugs may induce their own metabolic
clearance—​a phenomenon called ‘autoinduction’ (e.g. carbamaze-
pine and artemisinin).
Withdrawal syndromes
A common, but not inevitable, outcome of an adaptive response to
long-​term drug use is a withdrawal response either when the drug is
withdrawn or when an antagonist is given.
A withdrawal syndrome occurs in opiate users when the opiate is
withdrawn or an antagonist, such as naloxone, is given. The symp-
toms consist of yawning, rhinorrhoea, and sweating, followed by
shivering and goose flesh (‘cold turkey’); later, nausea, vomiting,
diarrhoea, and hypertension may occur. The acute syndrome sub-
sides within a week, but the anxious and disturbed sleep patterns
may last for several weeks or months. This syndrome can be avoided
by introducing increasing doses of methadone as the opiate is with-
drawn; methadone has a longer half-​life than opiates such as heroin
and causes much less withdrawal when it is eventually discontinued.
Delirium tremens is a feature of alcohol withdrawal in chronic
alcohol abusers. This syndrome consists of disorientation and visual
hallucinations. Withdrawal of benzodiazepines after long-​term
therapy may result in a disturbance of sleep pattern (rebound in-
somnia associated with abnormal sleep patterns), agitation, restless-
ness, and occasionally epileptic convulsions.
The risk of angina pectoris, myocardial infarction, and ar-
rhythmias is increased in patients with ischaemic heart disease
when β-​adrenoceptor antagonists are withdrawn after long-​term
use. This may be due to upregulation in the number of cardiac
β-​adrenoceptors, with increased sensitivity to the β-​adrenergic
effects of sympathetic stimulation.
Long-​term therapy with corticosteroids suppresses pituitary
secretion of ACTH, leading to adrenal cortical atrophy. When
treatment is suddenly withdrawn, ACTH secretion may take sev-
eral weeks or months to recover. During this time patients risk an
Addisonian crisis if stressed (e.g. if they have a myocardial infarction
or are operated on).
Adverse effects directly due to adaptation
Patients taking a neuroleptic drug (e.g. fluphenazine or haloperidol)
continuously for a period of years commonly develop abnormal
movements (known collectively as tardive dyskinesia). The face,
mouth, and tongue are often affected, with stereotyped sucking and
smacking of the lips, lateral jaw movements, and darting move-
ments of the tongue. Occasionally more widespread dyskinesia may
resemble choreoathetosis. The long-​term blockade of central dopa-
mine receptors is thought to lead to increased central sensitivity to
the effects of dopamine; this partly reflects increases in the number
of dopamine receptors. The risk of tardive dyskinesia is much lower
in the newer atypical agents such as quetiapine and risperidone; the
atypical clozapine has even been suggested to ameliorate established
tardive dyskinesia.
Adverse drug reactions
An adverse drug reaction can be defined as ‘an unwanted or harmful
reaction experienced following administration of a drug, or com-
bination of drugs, under normal conditions of use and is suspected
as being related to the drug (or combination)’. Sometimes, the term
is broadened to include all adverse reactions whether the drug is
dosed appropriately or not. Hence the term ‘adverse drug event’,
88
section 2  Background to medicine
which includes drug prescription and dispensing errors and failures
of patient compliance, is also used.
Incidence
The scale of the problem is probably still underestimated. Data
suggests that:
•	1 to 4% of acute hospital admissions are due to an adverse drug
reaction
•	5 to 20% of inpatients suffer an adverse drug reaction at some
point in their admission
•	up to 3% of deaths in hospital inpatients are due to an adverse drug
reaction
In addition to the morbidity and mortality, adverse drug reactions
are hugely expensive for healthcare systems. In the United Kingdom
the cost probably exceeds several billion pounds annually, and in the
United States of America some estimates have exceeded 100 billion
dollars annually.
Classification
Dose-​related adverse reactions
Dose-​related adverse reactions are usually due to an exaggeration of
a known pharmacological effect of the drug. The pharmacological
effect that produces the adverse reaction may be responsible for the
therapeutic effect (e.g. hypoglycaemia following insulin administra-
tion), or be a parallel effect (e.g. the anticholinergic action of tri-
cyclic antidepressants, producing a dry mouth or urinary retention).
Dose-​related adverse reactions may occur because of variations
in the pharmaceutical, pharmacokinetic, or pharmacodynamic
properties of a drug, often due to a drug–​disease interaction or a
pharmacogenetic characteristic of the patient. These mechanisms
are illustrated in the following paragraphs.
Pharmaceutical problem
Adverse reactions can be caused by a contaminant (e.g. pyrogens
or even bacteria in intravenous formulations). This is obviously a
hazard for illicit drugs that are used intravenously: not only are they
dissolved under non​sterile conditions, they may also be ‘cut’ with
other drugs (e.g. quinine, caffeine, and procaine).
Febrile reactions can occur routinely with some manu-
factured drugs given intravenously (e.g. amphotericin B and
bisphosphonates), but otherwise fever should be treated very sus-
piciously and the drug and giving set should be sent for microbio-
logical screening.
Out-​of-​date formulations may sometimes cause adverse reac-
tions because of degradation products. For example, outdated tetra-
cycline may cause Fanconi’s syndrome, because it is degraded to
anhydrotetracycline and epiandrotetracycline. The omission of the
preservative citric acid from tetracycline formulations has reduced
the risk of this effect, but has not removed it completely.
Very occasionally a drug has been incorrectly labelled by the manu-
facturer. Of more concern is the rise of counterfeit medicines (see
Chapter 2.10). They are thought to account for 15% of drug sales world-
wide and in parts of Asia and Africa the figure probably exceeds 50%
of sales. Counterfeit agents frequently contain none of the active drug,
or subtherapeutic doses, and may also contain additional chemicals or
drugs that are harmful. For example, ethylene glycol has been used in
the manufacture of fake paracetamol syrups (for its sweetness and vis-
cosity) and caused a number of deaths, especially in children.
Pharmacokinetic variation
There is often enormous variation in rate of drug elimination be-
tween individuals. This variation is greatest for drugs cleared by hep-
atic metabolism and is determined by several factors, which may be
genetic, environmental (diet, smoking, alcohol), or hepatic (blood
flow and intrinsic drug-​metabolizing capacity). On top of this vari-
ability, pharmacogenetic or hepatic abnormalities may be associated
with specific adverse reactions. In addition, renal and cardiac dis-
ease can change drug pharmacokinetics. The impact of pharmaco-
genetics is discussed in the following paragraphs.
The reserve of the liver parenchyma is large, so adverse reactions
due to impaired hepatic metabolism are uncommon. Nevertheless,
in patients with severe liver disease caution is needed when a drug
has a low therapeutic index or is subject to extensive first-​pass me-
tabolism. For example, severely impaired hepatocellular function
can reduce the clearance of drugs such as phenytoin, theophylline,
and warfarin. The portosystemic shunting seen in advanced cirrhosis
can also dramatically increase the bioavailability of drugs normally
cleared rapidly by the liver (e.g. morphine and other narcotic anal-
gesics, propranolol, and chlormethiazole). Drugs that the kidneys
excrete unchanged, or whose active metabolites are excreted, will
accumulate in renal failure. Important examples include digoxin,
atenolol, lithium, aminoglycoside antibiotics, and vancomycin.
Pharmacodynamic variation
The variability in pharmacodynamic response to a drug may be
compounded by concomitant disease. The patient with cirrhosis is
a good example: impaired hepatocellular function can reduce the
synthesis of clotting factors; the presence of oesophageal and gastric
varices imposes a further risk of upper gastrointestinal bleeding and
patients with alcoholic liver disease may have additional thrombo-
cytopenia and impaired platelet function. Judging the response to an
anticoagulant, antithrombotic, or antiplatelet drug in this setting is
very difficult, and the risk of haemorrhage is high.
A cirrhotic patient is also at risk of exaggerated sedation and en-
cephalopathy from opiates or long-​acting benzodiazepines that
are cleared by the liver. The hypokalaemic effects of diuretics or
amphotericin carry a similar risk. Patients with cirrhosis also have
inappropriate salt and water retention that can be worsened by drugs
such as NSAIDs, corticosteroids, and carbamazepine.
The pharmacodynamic effects of some drugs may be altered by
changes in fluid and electrolyte balance. For example, both hypo-
kalaemia and hypercalcaemia potentiate the toxic effects of di-
goxin. Hypocalcaemia prolongs the action of muscle relaxants such
as tubocurarine. Fluid depletion and hypovolaemia enhances the
hypotensive effects of antihypertensive drugs.
Non-​dose-​related adverse reactions
Non-​dose-​related adverse drug reactions are caused by immuno-
logical and pharmacogenetic mechanisms.
Allergic drug reactions are unrelated to the usual pharmacological
effects of the drug, and frequently show a delay between the first ex-
posure to the drug and the subsequent adverse reaction. Very small
doses of the drug may elicit the reaction once allergy is established.
2.6  Principles of clinical pharmacology and drug therapy
89
The reaction disappears on withdrawal; and the illness is often rec-
ognizable as a form of immunological reaction (e.g. rash, serum
sickness, anaphylaxis, asthma, urticaria, angio-​oedema).
Factors associated with an increased risk of allergic drug reac-
tions include a history of allergic disorders (patients with a history
of atopic disease and those with hereditary angio-​oedema) and
HLA status (e.g. the risk of severe skin reactions to carbamazepine
and allopurinol is strongly associated with specific alleles of the
HLA B locus).
Drug allergy and its manifestations are classifiable according
to the classification of hypersensitivity reactions (i.e. into four
types, I–​IV).
Type I reactions (anaphylaxis; immediate hypersensitivity)
In type I reactions, the drug or metabolite interacts with IgE mol-
ecules fixed to cells, particularly tissue mast cells and basophil
leucocytes. This triggers a process that leads to the release of pharma-
cological mediators (a cocktail of histamine, 5-​HT, kinins, and ara-
chidonic acid derivatives including leukotrienes), which cause the
allergic response.
Clinically, type I reactions manifest as urticaria, rhinitis, bron-
chial asthma, angio-​oedema, and anaphylactic shock. Drugs likely
to cause anaphylactic shock include penicillins, local anaesthetics,
and iodide-​containing radiographic contrast media.
Type II reactions (cytotoxic reactions)
In type II reactions, a circulating antibody of the IgG, IgM, or IgA
class interacts with a hapten (drug) combined with a cell membrane
constituent (protein), to form a hapten–​protein/​antigen–​antibody
complex. Complement is then activated and cell lysis occurs. Most
examples are haematological: thrombocytopenia from quinidine or
its enantiomer quinine (‘gin and tonic purpura’), and occasionally
rifampicin; ‘immune’ neutropenia, which can be difficult to distin-
guish from neutropenia occurring as a direct toxic effect on the bone
marrow—​phenylbutazone, carbimazole, tolbutamide, anticonvul-
sants, chlorpropamide, and metronidazole have all been incrimin-
ated; and the haemolytic anaemias that penicillins, cephalosporins,
rifampicin, and quinidine can also produce by this mechanism.
Type III reactions (immune-​complex reactions)
In type III reactions, antibody (IgG) combines with antigen to form
immune complexes that deposit in tissues; complement is then ac-
tivated, causing capillary endothelial damage. Serum sickness, with
fever, arthritis, enlarged lymph nodes, urticaria, and maculopapular
rashes, is the typical drug reaction of this type. Penicillins,
sulphonamides, and antithyroid drugs may cause it. Another type
III reaction is the acute interstitial nephritis caused by penicillins,
some NSAIDs, and some diuretics.
Type IV reactions (cell-​mediated or delayed hypersensitivity
reactions)
In type IV reactions, T lymphocytes are sensitized by a hapten–​
protein antigenic complex. When the lymphocytes meet the antigen,
an inflammatory response ensues. Examples are the contact derma-
titis caused by topical local anaesthetics and antihistamines, and
topical antibiotics and antifungal drugs. Rashes caused by a type
IV mechanism in response to sulphonamides and thiacetazone are
more common in people infected with HIV.
Anaphylactoid and pseudoallergic reactions
Anaphylactoid reactions resemble type I allergic reactions clinic-
ally, but the mast cell and basophil activation is not IgE-​dependent.
Instead, the cells are triggered directly, and drugs capable of doing
this include:  succinylcholine, morphine, d-​tubocurarine, vanco-
mycin (hence ‘red man’ syndrome on rapid intravenous admin-
istration), and N-​acetylcysteine. They are generally less severe
than allergen-​mediated anaphylaxis, and emergency treatment is
the same.
Aspirin can also trigger an attack of asthma by a non​immune
mechanism. The inhibition of airway COX-​1 enzyme by as-
pirin is thought to remove the inhibitory effect of prostaglandin
E2 and divert arachidonic acid towards production of cysteinyl-​
leukotrienes (especially LTC4). This is a very powerful constrictor
of airway smooth muscle (c. 1000-​fold the potency of histamine),
and this mechanism explains why aspirin-​sensitive asthmatics are
often sensitive to other NSAIDs (although COX-​2 selective drugs
may be relatively safe). It does not appear to explain why half of
aspirin-​sensitive asthmatics are also sensitive to the yellow food dye
tartrazine (E102).
In some patients, ampicillin, and its derivative amoxicillin, causes
a maculopapular erythematous rash resembling the toxic erythema
that can occur in penicillin hypersensitivity. However, this ampi-
cillin rash is not immunological in origin. It can be distinguished
from true penicillin hypersensitivity by its later onset after the first
dose (typically 10–​14 days compared with 7–​10 days in penicillin
hypersensitivity, though they overlap) and non​recurrence if re-​
exposed to ampicillin. Unlike penicillin hypersensitivity, it carries
no increased risk of anaphylaxis to penicillin. An ampicillin rash oc-
curs in about 1% of the normal population, but at a much higher fre-
quency in some groups of patients: it occurs almost invariably (and
can be a useful diagnostic pointer) in patients with some viral in-
fections (e.g. infectious mononucleosis, cytomegalovirus infection,
measles), lymphomas, and leukaemias. The risk is also increased in
patients taking allopurinol.
Other manifestations of allergic reactions
Drugs may cause other adverse reactions that do not fit clearly into
the earlier hypersensitivity classification, but where there is a strong
suspicion of an immune basis.
Drug fever as an isolated phenomenon can occur with anti-
biotics (penicillins, cephalosporins, isoniazid, sulphonamides, and
vancomycin), anticonvulsants (phenytoin and carbamazepine), α-​
methyldopa, hydralazine, and quinidine. The height or periodicity
of the fever is not a useful clue that it is drug-​induced, but all drug-​
induced fevers defervesce rapidly on drug withdrawal (c.24–​48 h).
Fever is also a manifestation of neuroleptic malignant syndrome,
a rare and serious idiosyncratic adverse reaction to neuroleptic
therapy (either initiation or dose-​escalation). In neuroleptic malig-
nant syndrome, the fever is accompanied by rigidity, reduced con-
sciousness, and autonomic disturbance. It resembles another rare
syndrome, malignant hyperpyrexia, although here the fever and ri-
gidity follow sensitization to volatile halogenated anaesthetic gases
(e.g. halothane) or suxamethonium. The aetiology of neuroleptic
malignant syndrome is still unclear, but patients with malignant
hyperthermia have mutations either in the ryanodine receptor or
skeletal muscle l-​type calcium channel.
90
section 2  Background to medicine
A syndrome mimicking systemic lupus erythematosus, often
involving joints and skin but generally sparing the kidneys, may
follow long-​term treatment with hydralazine, procainamide, pheny-
toin, or ethosuximide. Drug-​induced lupus is partly dose-​related
and, in the case of procainamide and hydralazine, is affected by the
acetylator status of the patient. Both drugs are metabolized by N-​
acetylation which is controlled by the polymorphic enzyme NAT2
(see following paragraphs). The two variants cause either slow or fast
clearance of the drug and slow acetylators are at increased risk of
drug-​induced lupus.
Asthma occurring as a pseudoallergic reaction to NSAIDs and
tartrazine is noted in the earlier paragraph. Other adverse drug reac-
tions in the lung include pneumonitis associated with drug-​induced
lupus (see earlier paragraph), pulmonary eosinophilia, and fibrosing
alveolitis. Eosinophilic granulomatosis with polyangiitis has been
associated with the use of cysteinyl-​leukotriene receptor antagonists
(e.g. montelukast), but it now seems unlikely that the syndrome is
actually caused by them. Rather their introduction was frequently
accompanied by withdrawal of corticosteroids that probably un-
covered a pre-​existing disease.
Jaundice may occur as an allergic response to some drugs through
either cholestasis (e.g. with phenothiazines, erythromycin, and
chlorpropamide) or generalized liver damage (e.g. with halothane,
isoniazid, and monoamine oxidase inhibitors).
Long-​term effects causing adverse drug reactions
Some adverse effects during long-​term therapy are related to both
the duration of treatment and the dose.
Adaptive changes
These are the basis of some adverse reactions such as the develop-
ment of tolerance and physical dependence to opiates, and tardive
dyskinesia in patients receiving long-​term neuroleptic therapy for
schizophrenia.
Rebound phenomena
When adaptive changes occur during long-​term therapy, sudden
withdrawal of the drug may result in rebound reactions. Examples
include the typical syndromes that occur after the sudden withdrawal
of opiates or of alcohol (delirium tremens). Sudden withdrawal of
barbiturates may result in restlessness, confusion, and convulsions,
and a similar syndrome in which anxiety features predominate may
occur after the sudden withdrawal of benzodiazepines. Similarly, pa-
tients who abruptly stop an SSRI may complain of a constellation of
symptoms 3 to 4 days later, which include headache, insomnia, diz-
ziness, paraesthesia, sweating, and flu-​like symptoms.
The withdrawal of some antihypertensive drugs may result in
rebound hypertension, but is especially common with clonidine.
Sudden withdrawal of β-​adrenoceptor antagonists may result in re-
bound tachycardia and arrhythmia, sometimes precipitating myo-
cardial ischaemia.
Acute adrenal insufficiency can occur when corticosteroids are
stopped abruptly. The risk depends on the potency and duration
of corticosteroids used, but not the route of administration—​it
has even been reported after stopping high-​dose topical or inhaled
glucocorticoids.
Reversal of the effects of unfractionated heparin with protamine
sulphate may be associated with rebound hypercoagulability and an
increased risk of thromboembolism. However, this risk may be justi-
fied if heparin overdosage has caused life-​threatening haemorrhage.
Importantly, protamine sulphate will not reverse the effect of low-​
molecular-​weight heparin and stopping or reversing oral anticoagu-
lants (such as warfarin) does not lead to rebound hypercoagulability.
Other long-​term effects
Chloroquine may accumulate in the corneal epithelium (causing a
keratopathy) and in the retina (causing a pigmentary retinopathy
and blindness). The former occurs in most patients on long-​term
therapy; the latter is less common but more serious. The risk in-
creases with daily doses of more than 2.5 mg/​kg (as the free base)
and chloroquine should only be used in inflammatory arthropathies
where the safer hydroxychloroquine has failed.
Amiodarone also accumulates extensively in tissues. Almost all
patients on long-​term amiodarone develop photosensitization from
skin deposition. They also develop microdeposits in the cornea, al-
though these are rarely symptomatic. It also accumulates in other
tissues but pulmonary alveolitis, neuropathy, and hepatocellular im-
pairment are relatively uncommon.
Delayed effects causing long-​term adverse drug reactions
Carcinogenesis
The long-​term effects of oestrogen therapy on cancer risk are
complex and depend on whether they are administered to pre-​ or
postmenopausal women. Long-​term oestrogen exposure through
the oral contraceptive pill probably increases the risk of breast cancer
but the effect is not consistent across all studies. The administration
of hormone replacement therapy (HRT) to postmenopausal woman
also increases their risk of breast cancer, although it appears that the
breast cancer risk from combined oestrogen–​progesterone HRT is
greater than from oestrogen-​only HRT (used in women after hyster-
ectomy). The incidence of endometrial carcinoma is also increased
in women taking oestrogen HRT for menopausal symptoms. In
contrast, the oral contraceptive pill protects against endometrial
cancer and the effect persists for many years after taking it. The
risk of colon cancer may also be reduced by HRT but not the oral
contraceptive pill.
Anabolic steroids carry a risk of both benign and malignant hep-
atic tumours on long-​term administration. The risk is greatest for
the 17-​alkylated derivatives such as oxymetholone. The latter is now
largely restricted for palliative use in cachectic states such as HIV
wasting syndrome, but is still used illicitly by bodybuilders.
Various anticancer drugs increase the risk of secondary solid tu-
mours and haematological malignancy. For example, cyclophos-
phamide containing chemotherapy regimens increase the risk of
bladder cancer. The risk of secondary acute myeloid leukaemia and
myelodysplastic syndromes is substantially increased in patients re-
ceiving chemotherapy for Hodgkin’s or non-​Hodgkin’s lymphoma
and testicular cancers. Alkylating agents, especially older regimens
using the mustard mechlorethamine, carry the highest risk from
lymphoma chemotherapy, and etoposide is the greatest risk from
testicular cancer chemotherapy.
Immunosuppressive drug regimens are also widely associated with
an increased risk of lymphoma and solid tumours in organ trans-
plant recipients. These regimens are typically based on calcineurin
inhibitors, such as ciclosporin. Recent data suggests that regimens
2.6  Principles of clinical pharmacology and drug therapy
91
based on a non​calcineurin inhibitor, sirolimus, may reduce the risk
of lymphoma and solid tumours in these patients.
Adverse drug reactions associated with reproduction
Some drugs impair fertility. For example, cytotoxic drugs can
cause permanent ovarian failure with amenorrhoea. Sperm pro-
duction may be reversibly impaired by sulphasalazine (espe-
cially in slow metabolizers), gonadotropin hormone antagonists,
methotrexate, and androgens. In fact, the reversible azoospermia
achieved with depot formulations of testosterone esters and pro-
gestogens is being developed for male contraception. Cytotoxic
drugs (especially alkylating agents) can reversibly or irreversibly
affect sperm production depending on the age of administration
(the prepubertal testis is relatively insensitive), the doses used, and
duration of exposure.
Teratogenesis  Teratogenesis occurs when a drug taken early in
pregnancy causes a developmental abnormality in a fetus. Exposure
to a teratogen in the first trimester of pregnancy, and particularly
the period of organogenesis (weeks 2–​8 of gestation), is most likely
to cause structural abnormalities. The central nervous system is vul-
nerable throughout pregnancy.
For a drug to be teratogenic it must first cross the placenta. As
a general rule, the drugs that do this have a low molecular weight,
are poorly ionized at physiological pH, and are very lipophilic.
Hence, heparin (even low-​molecular-​weight heparin) is a large,
highly charged molecule and d-​tubocurarine is a small, ionized, and
hydrophilic molecule; neither crosses the placenta. The placenta
also expresses large numbers of transporter proteins including the
ABC efflux pump P-​glycoprotein (MDR-​1). The efflux pumps are
expressed predominantly on the maternal-​facing surface of the pla-
cental villi and form part of the maternal–​fetal barrier. So, whether
a drug crosses the placenta depends not just on physiochemical
properties, but also its affinity for the efflux pump. The existence
of functional polymorphisms in these pumps probably explains
the wide variability observed in fetal drug concentrations, inci-
dence of teratogenesis, and drug failure in pregnancies exposed to
therapeutic drugs.
Since most drugs that are proven teratogens in animals will not
enter human development, it is unclear how well animal teratogen-
icity testing predicts human teratogenicity. A drug will have been
tested for teratogenicity in rodents and one other non​primate
species before it can be registered. However, negative results from
animal testing should not reassure anyone that the drug is safe in
human pregnancy. Indeed, thalidomide itself is not teratogenic in
rodents and the New Zealand rabbit is the only common labora-
tory mammal that shows similar sensitivity to humans. Women
of childbearing age are excluded from preregistration testing, so
the teratogenic risk for humans of a drug at the time of its regis-
tration is usually unknown. This explains the comments that exist
in package inserts and other literature on new drugs, discouraging
their use in pregnancy. It also explains why obstetricians often em-
ploy drugs that seem to other clinicians to be obsolete. For example,
in pregnancy-​related hypertension and pre-​eclampsia, hydralazine
and α-​methyldopa are still first-​line drugs. In non​pregnant hyper-
tension, they would not even be fourth-​line choices. Both drugs
have, however, been in obstetric use for a half century without asso-
ciated teratogenicity.
Adverse drug reactions on the fetus during the later stages of preg-
nancy  Some drugs that are not teratogenic may have adverse effects
on the fetus if given later in pregnancy. Table 2.6.2 lists some im-
portant drugs that should be avoided or used with care during later
pregnancy (some throughout the whole duration of pregnancy).
Given the uncertainty about teratogenic risk, what should be done
if a woman of childbearing potential is given a drug, and then finds
out days or weeks later that she is pregnant? First, it is important
to identify the drug and the exact time of exposure to it. If it is a
known or a likely teratogen, the relation between the time of ex-
posure and the likely time of conception should be determined.
Even if the precise date of conception is known, dating the preg-
nancy by ultrasound is advisable if a suspected teratogen has been
taken. If exposure to a known teratogen has occurred during the
first 8 weeks of pregnancy, detailed ultrasound examination of the
fetus may detect structural abnormalities, and serum and amniotic
α-​fetoprotein concentrations measured to screen for neural tube de-
fects. Any advice on termination of a pregnancy should be based on
a consideration of the risk of fetal abnormality from both published
information and investigation of the individual case.
Adverse reactions to drugs in breast milk  Some drugs cause ad-
verse reactions in babies after ingestion by the mother and excretion
in her breast milk. Excretion of drugs into breast milk is important
because 90% of women take at least one prescribed drug during the
first week after delivery. This may discourage many women from
breastfeeding unless they can be reassured that the benefits for the
baby of being breastfed far outweigh any risks for the overwhelming
majority of drugs (see Table 2.6.3).
The factors affecting drug excretion into breast milk are the same as
those that govern drug transfer across the placenta. The only important
difference is that breast milk has a lower pH and higher lipid content
than plasma. Like the placenta, this passive transfer only explains a
fraction of the total transport capacity. Active drug transport also oc-
curs through transporters and efflux pumps expressed in the breast
epithelium, which explains why some drugs are present in breast milk
at levels higher than expected (e.g. atenolol and benzylpenicillin).
Table 2.6.2  Drugs with proven or very high teratogenic risk
in human pregnancy
Drug
Associated risk
ACE inhibitorsa
Fetal renal toxicity/oligohydramnios
Alcohol
Fetal alcohol syndrome
Carbamazepine
Neural tube defects?/​Features of hydantoin syndrome
Diethylstilboestrol
Vaginal adenocarcinoma in female offspring
Isotretinoin
Craniofacial and cardiac anomalies
Lithium
Cardiac defects including Ebstein’s anomaly
Methotrexate
Fetal aminopterin syndrome
Misoprostol
Moebius syndrome
Phenytoin
Fetal hydantoin syndrome
Tetracycline
Decidual teeth staining
Thalidomide
Phocomelia
Valproic acid
Neural tube defects
Warfarin
Stippled epiphyses/​nasal hypoplasia
a ATII receptor antagonists carry similar risk.
92
section 2  Background to medicine
Neonates and young babies may be at risk of adverse drug
reactions because clearance pathways for a drug are immature.
The GFR is only 25% of the adult value at birth (based on body
weight) and only reaches the adult range at 3–​6 months. Tubular
secretion measured by PAH clearance is also impaired (10% of
adult capacity until 6 months), which explains the slow elimin-
ation of frusemide in neonates (it is secreted by the PAH trans-
porter). Phase I metabolism through CYP enzymes are isoform
dependent: 3A7 is the predominant isoform in utero; it is replaced
by 3A4 postpartum, but it takes more than 1 month to reach adult
levels; CYP 1A2 does not appear until 3 months; CYP2D6 and E1
appear after birth but rapidly achieve adult levels within hours of
delivery.
Drugs that should be given to breastfeeding mothers with caution
are included among the drugs in Table 2.6.3. A more extensive list
can be found in the British National Formulary and the Physician’s
Desk Reference. If the safety of a drug is in serious doubt and it is not
possible to identify an alternative, breastfeeding can be temporarily
suspended while the drug is given.
Detecting adverse drug reactions: Pharmacovigilance
The importance of adverse reactions for drug regulatory bodies
and the pharmaceutical industry has led to the evolution of a
subspecialty of clinical pharmacology devoted to the detection and
evaluation of adverse drug reactions. This specialization is called
pharmacovigilance.
Because drugs reach the market based on the experience of dosing
just a few thousand patients, capturing as many adverse drug reac-
tions as possible afterwards is of crucial importance. This activity
is called postmarketing surveillance, which usually relies on self-​
reporting by health professionals and in some countries, the gen-
eral public as well. In the United Kingdom, reporting by healthcare
workers operates through the yellow card scheme run by the
Medicines and Healthcare Products Regulatory Agency (MHRA)
and the Commission on Human Medicines (CHM). It is a spon-
taneous reporting scheme in the sense that it is voluntary and used
to rely on the ‘yellow card’ in the back of every BNF being filled
out by the healthcare worker. This can now be done online (https://​
yellowcard.mhra.gov.uk/​). The scheme aims to collect (1) serious or
fatal adverse drug reactions from all drugs; (2) all adverse drug re-
actions from newly licensed drugs (designated with a black triangle
in the BNF); and (3) all adverse drug reactions in children under 18.
Spontaneous reporting schemes rely on goodwill and are prone to
considerable underreporting. Rare or unusual side effects are easily
detected with them (e.g. the withdrawal of cerivastatin because of
rhabdomyolysis and pergolide because of valvular heart disease). On
the other hand, serious adverse reactions involving effects that occur
frequently in the population receiving the drug are not so easily de-
tected. The thrombotic risk of COX-​2 selective NSAIDs (coxibs), es-
pecially rofecoxib (Vioxx), is a timely reminder of this serious flaw
in spontaneous reporting schemes. If used over long periods, coxibs
cause an excess of stroke and myocardial infarction, but both of these
are common events in patients receiving NSAIDs. Most physicians,
for example, would not immediately conclude that a 65-​year-​old pa-
tient who died from a stroke or myocardial infarction while on a coxib
was the victim of an adverse drug reaction. It is a sobering reminder
that rofecoxib was only withdrawn after some 20 million people were
exposed to it and many thousands had probably died from taking it.
Despite the limitations of spontaneous reporting, it is a cru-
cial part of the World Health Organization’s efforts to coordinate
pharmacovigilance on a global scale as opposed to local efforts such
as the yellow card scheme. Its database (http://​www.who-​umc.org/​)
currently contains some 11 million reports (May 2015), and 122
countries had signed up to the scheme by September 2015. There is
also a Europe-​wide (or rather European Union-​wide) pharmacovigi­
lance programme run by the European Medicines Agency, called
EudraVigilance (https://www.ema.europa.eu/en/human-regulatory/
research-development/pharmacovigilance/eudravigilance). The dif-
ference in this scheme is that adverse drug reaction notification
by the pharmaceutical companies operating within the European
Union is not voluntary, by a statutory requirement of their
marketing authorization. It is too early to say how the performance
of these two systems compares in terms of the early detection of ad-
verse drug reactions.
The probity of the pharmaceutical industry itself has also been
seriously questioned in recent years in the process of alerting the
regulatory authorities about suspected adverse drug reactions. Large
Table 2.6.3  Effects of common maternal drugs on breastfed infants
or milk production
Acebutolol/​atenolol
Cyanosis, hypotension, bradycardia
Amoxicillin
None
Aspirin
Metabolic acidosisa
Bendrofluazide
May reduce milk production
Bromocriptine
Suppresses lactation
Caffeine
Irritability if >2–​3 cups of coffee/​day
Carbamazepine
None
Carbimazole
Goitre
Ciprofloxacin
None
Digoxin
None
Diltiazem
None
Fluconazole
None
Isoniazid
None
Labetalol
None
Levothyroxine
None
Lithium
Plasma level up to 50% mother
Nalidixic acid/​nitrofurantoin
Haemolysis in G6PD deficient infanta
Morphine
None
Oestrogen (in oral contraceptive pill)
May reduce milk production
Pethidine
None
Phenobarbitone/​primidone
Sedation
Phenytoin
Methaemoglobinaemiaa
Prednisolone
None
Propylthiouracil
None
Valproic acid
None
Verapamil
None
Warfarin
None
a Single case reports.
From American Academy of Pediatrics (2001). Policy statement. (http://​aappolicy.
aappublications.org/​cgi/​content/​full/​pediatrics;108/​3/​776).
2.6  Principles of clinical pharmacology and drug therapy
93
observational studies commissioned by the pharmaceutical com-
panies to capture adverse reactions have been deliberately hidden
(by Bayer in the case of aprotinin) or methodologically questioned
(by Merck in the case of Vioxx) when they confirmed suspected ad-
verse reactions. These and other cases have highlighted the need for
regulatory agencies to be able to compel pharmaceutical companies
to: (1) carry out the necessary controlled trials to investigate the
safety of a new drug; and (2) make disclosure of these studies man-
datory. Alternatively, healthcare systems should probably commis-
sion the necessary studies. These would certainly be cost-​effective
considering how much they will have been spent (e.g. managing the
excess of strokes and myocardial infarction that followed the intro-
duction of coxibs).
Prevention of adverse drug reactions: The role
of the patient
Most people who take either a prescribed drug or one purchased
over the counter do not usually expect unwanted effects from it, al-
though they may have been alerted to potential adverse reactions by
the doctor, pharmacist, or packet insert. The information from any
of these sources should warn about all potential adverse drug reac-
tions, so that the patient can:
•	assess the potential disadvantages of the drug, before deciding to
take it;
•	connect an adverse reaction with the taking of the drug and take
appropriate action.
However, what is usually lacking is advice about how to minimize
or avoid adverse drug reactions. This advice is as relevant for the
patient as it is for the prescriber. The prescriber in particular should
consider the following:
1.	 Are there dose-​dependent effects? If there are, they can be pre-
vented or minimized by keeping the drug dosage as low as
possible. It is always worth considering whether dosing below
the pharmaceutical company’s recommended range may be ap-
propriate for individual patients. This is especially the case in
older people or for non​white ethnic groups (who may metab-
olize the drug differently) as both are often underrepresented in
preregistration drug trials. If adverse reactions are more likely
with continued use of the drug, then the duration of use should
be limited (e.g. with neuroleptic drugs).
2.	 Are drug interactions likely? If there are known interactions, they
may be prevented by ensuring interacting drugs are avoided. This
list should include agents that could be taken by the patient but
are not prescribed. For example, it may be necessary to avoid con-
sumption of grapefruit juice or herbal remedies such as St John’s
wort if the drugs are substrates for CYP 3A4 or P-​glycoprotein
(and the patient is taking simvastatin bought over the counter).
3.	 There may be serious adverse reactions that are unknown and/​
or undetected during preregistration trials. Drugs that have
been prescribed over many years have an established margin of
safety and should be used in preference to newer drugs. If there
is a compelling reason to use a new drug, the patient should be
aware that it is a new drug and there should be closer monitoring
for adverse reactions.
4.	 Adverse reactions are more likely when a prescriber is using a
drug they are not familiar with. The prescriber might consider
asking for a second opinion about the appropriateness of the
drug and problems with its use if they are not sure.
5.	 Some individuals are predisposed to adverse drug reactions. This
is usually a genetic susceptibility. They may, for example, be poor
drug metabolizers, porphyriacs, or have G6PD deficiency. If this
is known, patients should be aware of drugs they should avoid,
and they should tell other prescribers of the problem before they
take any new drug.
Drug interactions
A drug interaction occurs when the effects of one drug are altered
by the effects of another drug. Usually this results in an adverse reac-
tion, but in a few cases, it may actually prove beneficial. Interactions
form up to 10% of all adverse drug reactions, but crucially, among
patients who die from an adverse drug reaction, about one-​third of
deaths are due to interactions.
Drugs likely to precipitate interactions often have one or more of
the following properties:
•	They are highly protein bound (e.g. aspirin and sulphonamides).
•	They induce drug metabolism (e.g. phenytoin, carbamazepine,
and rifampicin).
•	They inhibit drug metabolism (e.g. cimetidine, metronidazole,
and triazole antifungals).
The drugs most likely to be affected by drug–​drug interactions are
also those with a steep dose–​response curve and a low therapeutic
index, which includes aminoglycoside antibiotics, warfarin, and
other coumarins, anticonvulsants, antihypertensive drugs, cardiac
glycosides, cytotoxic and immunosuppressant drugs, oestrogen-​
containing oral contraceptives, and some centrally acting drugs. The
most frequent interactions occur with warfarin and other coumarins.
Drug interactions fall into three basic types (see Table 2.6.4).
Pharmaceutical interactions
These involve physiochemical effects between a drug and the solu-
tion it is mixed with or between two drugs when they are mixed
together (usually for injection). The interaction results in the drug
either precipitating from solution or being inactivated in some other
way. There are numerous examples, but some general principles can
help to avoid many of them: give intravenous drugs by bolus injec-
tion if possible or via an infusion pump; do not add drugs to infu-
sion solutions other than dextrose or saline; only mix drugs in the
same infusion solution if they are known to be safe (e.g. potassium
chloride with insulin).
Pharmacokinetic interactions
Pharmacokinetic interactions occur when a drug alters the ab-
sorption, distribution, or elimination (metabolism or excretion) of
another drug.
Absorption interactions
One drug substantially altering the absorption of another is rela-
tively uncommon but there are important examples. Anion ex-
change resins, for example, bind warfarin, digoxin, and thyroxine
very avidly (see earlier paragraphs). Other examples are given in
Table 2.6.4.
94
section 2  Background to medicine
This type of interaction may occasionally be beneficial. Hence,
metoclopramide increases gastric emptying and speeds the absorp-
tion of analgesics (such as ibuprofen and paracetamol) used to treat
acute migraine. Activated charcoal also binds many drugs in the gut
lumen, so preventing their absorption or enterohepatic recycling.
This is widely exploited in drug overdoses (see earlier paragraphs
and Chapter 10.4.1). It is also used to accelerate excretion of the
antirheumatoid drug leflunomide if it causes a serious adverse re-
action because of the extremely long half-​life (2 weeks) of its active
metabolite.
Protein-​binding displacement interactions
Displacement of one drug by another from its binding sites on
plasma proteins will cause an increase in the circulating concentra-
tion of unbound drug. This is only important if the displaced drug is
highly protein bound (>90%) and has a small volume of distribution
(that will exaggerate the rise in free drug concentration). The drugs
concerned are warfarin, phenytoin, and tolbutamide.
The most common precipitant drugs in protein-​binding dis-
placement interactions are sulphonamides, salicylates, and chloral
hydrate and some of its congeners (because of their metabolite,
trichloroacetic acid). In addition, valproate specifically displaces
phenytoin.
However, displacement interactions are generally relatively un-
important. This is because the rise in free drug concentration in-
creases the drug’s clearance, so that the total concentration actually
falls if the displacing drug is given chronically. Displacement is only
a problem if the initial rise in free drug concentration itself causes
toxicity; fortunately, if this is not serious it will be transient.
Interactions through induction of metabolism
Induction of P450 enzymes is a major source of drug inter-
actions. Drugs that induce drug metabolism include barbiturates,
carbamazepine, griseofulvin, phenytoin, and rifampicin (see
Table 2.6.4). The herbal preparation of St John’s wort also induces
P450 isoforms.
Table 2.6.4  Mechanisms of drug interactions, with important examples
Mechanism
Example
Outcome
Physiochemical
Calcium gluconate plus sodium bicarbonate
Precipitation of calcium carbonate in
infusion solution
Pharmacokinetic
Altered
absorption
Reabsorption of oestrogens reduced by antibiotics
Contraceptive failure
Gastric emptying increased by metoclopramide
Increased rate of absorption of simple
analgesics
Fluoroquinolones (e.g. ciprofloxacin) and agents containing
divalent/​trivalent cations (e.g. antacids or Fe salts) or sucralfate
Reduced absorption
Altered protein
binding
Displacement of phenytoin by aspirin (not low-​dose aspirin)
Phenytoin toxicity (transient)
Increased
metabolism
Oestrogen metabolism increased by carbamazepine, phenytoin,
rifampicin
Contraceptive failure
Reduced
metabolism
Warfarin metabolism inhibited by amiodarone, metronidazole,
cimetidine
Warfarin toxicity
Theophylline metabolism inhibited by erythromycin or
fluoroquinolones (e.g. ciprofloxacin)
Theophylline toxicity
Phenytoin/​carbamazepine metabolism inhibited by cimetidine,
erythromycin, fluconazole, or isoniazid
Phenytoin/​carbamazepine toxicity
Amine metabolism inhibited by monoamine oxidase inhibitors
(includes tyramine in foods and phenyl-​propanolamine in cold cures)
Acute severe hypertension
Reduced renal
elimination
Penicillin/​cephalosporin excretion reduced by probenecid
Prolonged duration of antibiotic action
Lithium excretion reduced by diuretics or NSAIDs
Lithium toxicity
Digoxin excretion reduced by amiodarone, quinidine, verapamil
Digoxin toxicity
Pharmacodynamic
Shared
mechanisms
Vitamin K competes with warfarin for epoxide reductase
Reversal of the effect of warfarin
Naloxone displaces opioids from opioid receptors
Reversal of opioid toxicity
PDE5 inhibitors (e.g. sildenafil) and nitrates
Profound hypotension
Statin and niacin, fibrates (especially gemfibrozil)
Risk of rhabdomyolysis/​renal failure
SSRIs and tramadol, triptans (e.g. sumatriptan) or monoamine oxidase
inhibitors (e.g. selegiline)
Risk of serotonin syndrome
Parallel
mechanisms
Psychoactive drugs and alcohol
Increased sedation
Cytotoxic drugs acting at different stages of the cell cycle
Therapeutic potentiation in cancer
chemotherapy
Warfarin and antiplatelet drugs (aspirin and other NSAIDs or
clopidogrel)
Increased risk of bleeding and impaired
haemostasis
2.6  Principles of clinical pharmacology and drug therapy
95
Interactions through inhibition of metabolism
Most of the interactions under this heading involve drugs that in-
hibit specific P450 pathways, for example, through CYP3A4 or 2D6.
Some examples are shown in Table 2.6.4.
Metabolism through non-​CYP metabolic pathways can also be af-
fected. Hence, the interaction of allopurinol with azathioprine and
6-​mercaptopurine results from its inhibition of xanthine oxidase.
Both 6-​mercaptopurine and azathioprine (which is metabolized to
6-​mercaptopurine) are metabolized by xanthine oxidase.
The interaction of MAO inhibitors with dietary tyramine is an-
other example of a non-​CYP pathway. Tyramine in the diet usu-
ally has very low bioavailability because of degradation by MAO in
the gut wall and liver. MAO inhibitors inactivate the enzyme both
here and within sympathetic nerve endings, which increase their
noradrenaline content. Hence, tyramine ingested after an MAO in-
hibitor has substantially increased bioavailability, and the resulting
plasma levels are sufficient to displace the enhanced noradrenaline
stores from sympathetic nerve terminals. The hypertensive crisis
this causes is referred to as the ‘cheese reaction’.
These interactions may again be sometimes useful. The protease
inhibitor ritonavir, for example, is a potent CYP3A4 inhibitor and
is added to antiretroviral drug regimens containing other protease
inhibitors to specifically inhibit their metabolism and so increase
bioavailability (which may otherwise be low). This is referred to as
PK-​enhanced drug formulation.
Excretion interactions
Most interactions involving drug excretion occur in the kidneys,
although secretion into the gut through P-​glycoprotein is im-
portant for some drugs (e.g. digoxin, cyclosporin). The latter ex-
plains drug interactions with substances such as St John’s wort and
grapefruit juice.
Interactions due to the alteration of active renal drug secretion
are shown in Table 2.6.4. Some drugs that are weak acids or bases
are passively reabsorbed along the nephron. The extent of reabsorp-
tion is pH-​dependent and can be exploited following overdose to
increase drug elimination in the urine (see earlier paragraphs).
Pharmacodynamic interactions
Pharmacodynamic interactions are very common and occur when
one drug alters the effect of another. This can arise because the two
drugs act on the same drug target or pathway. Alternatively, the
interacting drugs may produce the same effect but by entirely sep-
arate mechanisms. Some examples are shown in Table 2.6.4.
Interactions through a shared mechanism
When there is a shared drug target (such as a receptor) the inter-
action may arise because of antagonism or combined agonism at
this site. Many antagonistic interactions are therapeutically benefi-
cial. Examples are the reversal of the effects of benzodiazepines with
flumazenil or warfarin with vitamin K. In contrast, synergistic inter-
actions are often adverse. Hence, nitroglycerin and sildenafil are
useful drugs to treatment angina and erectile failure, respectively;
nitroglycerin being metabolized to nitric oxide (NO) and sildenafil
augmenting the effect of endogenous NO. Combining the two, how-
ever, causes devastating hypotension.
Interactions through parallel mechanisms
Two drugs may have similar pharmacological or toxic effects but
they are achieved through different mechanisms. For example, the
bleeding risk of warfarin is increased by drugs that affect platelet
function (e.g. NSAIDs, clopidogrel, or glycoprotein IIb/​IIIa antag-
onists), reduce platelet number (drug-​induced thrombocytopenia)
or cause gastrointestinal ulceration (e.g. NSAIDs).
Exploiting parallel pathways can also be beneficial. Many chemo-
therapy regimens exploit pharmacodynamic interactions of their
component drugs. And combinations of antibiotics are used rou-
tinely even when a single organism is being targeted (e.g. penicillin
plus an aminoglycoside in bacterial endocarditis and multiple-​drug
regimens for tuberculosis).
Pharmacogenetics and pharmacogenomics
The variability between individuals in terms of both the pharmaco-
kinetics and pharmacodynamics of a drug is partly determined by
genetic variation. The variability can occur in a single gene (hence
pharmacogenetics) or arise from the interplay of a variety of genes
(pharmacogenomics). The realization that the human genome is an
important player has fuelled expectation that drug prescription may
in future be ‘personalized’ based on the patient’s genotype.
Pharmacokinetic variability
Several important pathways of drug metabolism can affect drug
pharmacokinetics through polymorphic variation in the enzymes
involved. These include oxidation, acetylation, S-​methylation, and
suxamethonium hydrolysis (Table 2.6.5).
Oxidation
About 80% of oxidative drug metabolism occurs through the cyto-
chrome P450 or CYP enzyme family of enzymes (see earlier para-
graphs). There are some 57 genes in humans encoding these enzymes
and they are named with a number and letter for the family and sub-
family to which they belong (e.g. 3A4 or 2D6). Many of these genes
are polymorphic, with the common wild-​type gene being mutated to
produce a change in gene function; usually a reduction. The alleles
for these different polymorphisms or gene variants are designated by
a number after an asterisk (e.g. 2D6*4 for the commonest variant of
2D6 that reduces enzyme function).
Drugs that are metabolized through polymorphic CYP enzymes
show drug metabolism in the population that has a bimodal (or
occasionally trimodal) distribution. An example of this is shown
in Fig. 2.6.6, where the metabolic ratios for debrisoquine are
shown in the urine of over 1000 patients (debrisoquine is an ob-
solete antihypertensive but is still used as a ‘probe’ drug to define
metabolic phenotype for CYP enzymes). Most are extensive met-
abolizers with low-​to-​intermediate drug:metabolite ratios in their
urine, but a significant minority are poor metabolizers with high
ratios. In this case, the poor-​metabolizer status is due to 2D6 gene
variants that reduce 2D6 enzyme activity. Genetically, the defect
is very heterogenous with around 100-​point mutations, promoter
variants, and deletions now recognized (see http://​www.imm.ki.se/​
cypalleles). The frequency of the poor-​metabolizer phenotype
96
section 2  Background to medicine
varies across different ethnic groups:  it is around 5 to 10% in
northern Europeans but only approximately 1% in the Chinese.
Drugs that are commonly metabolized through 2D6 include co-
deine, dextromethorphan, flecainide, metoprolol, nortriptyline,
propafenone, timolol, and tramadol (see Table 2.6.1). The dose-​
related adverse reactions of these drugs (e.g. central nervous
system toxicity with nortriptyline or bradycardia and hypo-
tension with the β-​blockers) are more likely in patients with the
poor-​metabolizer phenotype. Quinidine is an inhibitor of the 2D6
enzyme and is able to convert an extensive to a poor metabolizer.
Some patients even have a super-​ or ultrametabolizer pheno-
type. This phenotype occurs through duplication of the 2D6 gene
and shortening of the half-​life of 2D6-​metabolized drugs in these
patients is directly related to the degree of duplication (Fig. 2.6.7).
They are relatively uncommon compared to the poor metabol-
izers (see Fig. 2.6.6), but their status has important clinical conse-
quences. A 2D6 drug will appear to be relatively ineffective in an
ultrametabolizer unless very large doses are used, and they may
even be suspected of poor compliance. In contrast, if the drug is
a prodrug, ultrametabolizers are characteristically very sensitive
to its effects. Hence, they may show opiate intoxication from what
appear to be low therapeutic doses of codeine because of its rapid
and complete metabolism to morphine. Although they are rare in
Europe, 30% of East Africans may be ultrametabolizers.
The other CYP enzymes are not as well studied as 2D6, but poor
metabolizers have been identified for 2C9 (metabolizing losartan,
phenytoin, tolbutamide, and warfarin) and 2C19 (metabolizing
omeprazole). There is a commercial microarray available to geno-
type individuals for the commoner variations in the 2D6 and 2C19
genes (it is called the AmpliChip P450) that has been approved by
the FDA, but the technology in this area is evolving rapidly
Table 2.6.5  Pharmacogenetic variation affecting drug metabolism
Metabolizer phenotype
Enzyme responsible
Common deficiency
gene variant
Frequency in white people
Comments
Suxamethonium apnoea
Butyrylcholinesterase
(CHE)
CHE70 D>G
1:3000
Many other variants encoding CHE with
reduced or no activity
Poor metabolizer
CYP2D6
2D64 (premature
stop codon)
5–​10%
Commonest allele for PM phenotype
Slow acetylator
N-​acetyltransferase (NAT2)
NAT2*5B
c.40% carry it but compound
heterozygotes as likely as *5B
homozygotes
4 is the fast allele and behaves
dominantly
Thiopurine sensitivity
Thiopurine S-​
methyltransferase (TPMT)
TPMT3A (154A>T,
240T>C)
5% are carriers (homozygous
rate c.1:250)
20 other variants but 95% of cases due to
*2 or *3(A–​D)
Irinotecan sensitivity
UDP-​glucuronyl
transferase 1 (UGT1A1)
(TA)7–​TATA
c.40%
Promoter variant reduces UGT1A1
expression
0
0
0.01
0.10
1
10
100
40
Ultrarapid
metabolism
Poor
metabolism
Extensive metabolism
Cutoff
80
120
Debrisoquin:4-Hydroxydebrisoquin metabolic ratio
No. of subjects
Fig. 2.6.6  Pharmacogenetics of CYP2D6. Urinary metabolic ratios of
debrisoquin to its metabolite, 4-​hydroxydebrisoquin, for more than
1000 Swedish subjects. Poor metabolizers with no or reduced CYP2D6
activity are separated by the cut-​off box from extensive metabolizers.
Adapted from Bertilsson L, et al. (1992). Pronounced differences between native
Chinese and Swedish populations in the polymorphic hydroxylations of debrisoquin
and S-​mephenytoin. Clin Pharmacol Ther, 51, 388–​97.
Morphine
Morphine-3-glucuronide
0
0
10
20
30
40
50
60
70
80
90
2
4
6
8
10
12
14
16
18
20
22
24
0
0
1
2
3
4
2
4
6
8
10
12
14
16
18
20
22
24
Concentration [μg/litre]
Concentration [μg/litre]
PM
EM
UM
Fig. 2.6.7  Effect of 2D6 genotype on the plasma concentration of
morphine and its active metabolite following a 30 mg dose of codeine.
Ultrametabolizers with duplication of the 2D6 wild-​type gene show
higher levels compared to extensive metabolizers carrying single
copies. Poor metabolizers for comparison have only loss-​of-​function or
non​functional copies.
Reprinted by permission from Springer Nature: The Pharmacogenomics
Journal. Kirchheiner J, et al. (2007). Pharmacokinetics of codeine and its
metabolite morphine in ultra-​rapid metabolizers due to CYP2D6 duplication.
Pharmacogenomics J, 7, 257–​65, Copyright © 2006, Springer Nature.
2.6  Principles of clinical pharmacology and drug therapy
97
Acetylation
Some drugs are cleared by acetylation. These include dapsone,
hydralazine, isoniazid, procainamide, and some sulphonamides.
The enzyme involved is the hepatic enzyme N-​acetyltransferase
(NAT2, EC 2.3.1.5) and if drug:metabolite ratios for a drug cleared
by acetylation are measured, they distribute bimodally in the
population. Subjects producing the highest ratios are described
as being fast acetylators (cf. extensive metabolizers for the 2D6
pathway). Again, the genetic basis for the biochemical phenotype
is very heterogenous, and the frequency of fast acetylators varies
across different ethnic groups: 40% in northern Europeans, 85% in
Japanese, and 5% among the Inuit.
Slow acetylators require lower doses of drugs that are cleared
by acetylation than fast acetylators. Slow acetylators are also more
likely to develop the lupus erythematosus-​like syndrome caused
by isoniazid, hydralazine, and procainamide, and the peripheral
neuropathy caused by isoniazid (isoniazid actually causes deg-
radation of pyridoxine). The interaction between isoniazid and
phenytoin, in which isoniazid inhibits phenytoin metabolism
causing phenytoin toxicity, is also more frequent among slow
acetylators.
Disease associations with polymorphic metabolism
As some diseases may be related to the effects of environmental
chemicals that have carcinogenic metabolites, it is of interest
that polymorphic acetylation, hydroxylation, and sulphoxidation
have other clinical associations. The evidence linking acetylator
status with the risk of bladder cancer is probably the best-​
established association; slow acetylators having increased risk.
Other weaker associations may exist between debrisoquine
metabolizer status (2D6) and risk of parkinsonism in poor
debrisoquine hydroxylators, of bronchogenic carcinoma in exten-
sive debrisoquine hydroxylators, and of primary biliary cirrhosis
in poor sulphoxidizers.
Glucuronidation
The antitumour agent irinotecan is a prodrug. Its active metabolite,
SN-​38, has about 1000-​fold higher activity than irinotecan itself
and is inactivated by glucuronidation through UGT1A1. Patients
homozygous for a polymorphism in the promoter of UGT1A1 are
much more likely to develop severe neutropenia after irinotecan
because of defective glucuronidation of SN-​38. The same poly-
morphism is also involved in Gilbert’s syndrome.
Methylation
The thiopurines (6-​mercaptopurine and 6-​thioguanine) are me-
tabolized by S-​methylation through the enzyme thiopurine
S-​methyltransferase (TPMT, EC 2.1.1.67). The metabolism of
azathioprine is also affected by this enzyme as azathioprine is re-
duced to 6-​mercaptopurine after dosing (i.e. it is a prodrug). High
levels of TPMT are found in red cells, and the activity is trimodal
in the population with high, intermediate, and low levels of TPMT
detectable. Because low levels of TPMT (present in c.1:300 whites)
reduces metabolic clearance of thiopurines, these subjects are ex-
posed to high levels of 6-​thioguanine nucleotides. This leads to
severe myelosuppression if they are given standard doses of these
drugs. Hence, thiopurines can only be used safely in subjects with
low TPMT if they are given very low doses.
There are some 20 gene variants that affect the activity of TPMT
and carriers for TPMT variants are common—​overall frequencies
are typically 5 to 10%. The commonest variant (TPMT*3A) is ac-
tually a double mutant that switches two amino acids and produces
an unstable enzyme that explains the low level of TPMT in homo-
zygotes. There is considerable ethnic variation in the frequency of
variants (e.g. the *3A is not seen among the Chinese).
Suxamethonium hydrolysis
Suxamethonium (succinylcholine) is metabolized in the plasma
by the non​specific esterase pseudocholinesterase (EC 3.1.1.8;
also called butyrylcholinesterase). Normally this happens quickly,
which explains why neuromuscular blockade with this drug lasts
only a few minutes. Some patients, however, have very slow clear-
ance of suxamethonium due to low plasma pseudocholinesterase
activity. This manifests as prolonged neuromuscular blockade or
apnoea after the use of suxamethonium, which can last several
hours or longer.
Suxamethonium apnoea is usually very rare, but is common in
Inuit (up to 10%). Several different gene defects can cause pseudo-
cholinesterase deficiency: the dibucaine-​resistant, fluoride-​resistant,
and ‘silent’ gene types. Patients with reduced pseudocholinesterase
also show increased sensitivity to the non​depolarizing blocker
mivacurium. Interestingly, pseudocholinesterase deficiency does
not appear to affect the conversion of the prodrug bambuterol
(which is also a substrate for pseudocholinesterase) to its active me-
tabolite terbutaline.
Pharmacodynamic defects
Some biochemical abnormalities make individuals peculiarly sensi-
tive or resistant to the effects of certain drugs. All have a genetic basis.
Red cell enzyme defects (see Chapter 22.6.11)
Adverse drug reactions may affect people whose red cells are de-
ficient in glucose-​6-​phosphate dehydrogenase (G6PD). If their
red cells are exposed to an oxidizing drug, they lose their oxygen-​
carrying capacity as haemoglobin is oxidized to methaemoglobin
and they eventually haemolyse. Drugs commonly implicated
are doxorubicin, nalidixic acid, nitrofurantoin, primaquine, and
sulphamethoxazole. There is a longer list of drugs that may cause
haemolysis in some G6PD-​deficient individuals, depending on their
genotype. This is because the common African variant gives higher
red cell G6PD levels than the Mediterranean form, so mild oxidative
stress is better tolerated. This list includes aspirin (low dose is usually
safe), chloramphenicol, l-​dopa, isoniazid, quinine and related com-
pounds, trimethoprim, and vitamin K.
Porphyria (see Chapter 12.5)
The hepatic porphyrias, acute intermittent porphyria and porphyria
cutanea tarda, are characterized by abnormalities of haem biosyn-
thesis. Certain drugs may precipitate an attack of porphyria espe-
cially cytochrome P450 inducers (e.g. barbiturates, carbamazepine,
and rifampicin). The quality of data as to the safety (or not) or other
drugs is variable. There is a useful web database that rates drug safety
on a five-​point scale (http://​porphyriadrugs.com/​).
98
section 2  Background to medicine
Malignant hyperthermia
This potentially fatal complication of general anaesthesia follows ex-
posure to halogenated anaesthetic gases and suxamethonium (see
earlier paragraphs).
Vitamin D-​resistant rickets
Three varieties of rickets are resistant to the effects of vitamin D
(cholecalciferol):  familial hypophosphataemic rickets, vitamin D
dependency, and Fanconi’s syndrome (see Chapter 21.16).
Warfarin sensitivity
Sensitivity to warfarin varies widely in the general population. This
has been explained on the basis of common gene variants in its
target enzyme vitamin K epoxide reductase (EC 1.1.4.2, encoded by
VKORC1) and the CYP2C9 gene responsible for its metabolism.
Pharmacogenomics and the prospect of ‘personal
prescribing’
The realization that the pharmacokinetics and pharmacodynamics
of a drug may be genetically determined has raised the prospect of
tailoring the drug to the patient. Predefining the generic variants
that a patient has before a drug is given could, in principle, avoid
many adverse drug reactions and interactions. The sequencing
technology to perform the necessary genotyping is now available,
and many drug trials now incorporate this technology to identify
genotypic signatures that both affect the therapeutic response to the
drugs and predict adverse reactions. However, we are some way off
truly ‘personalized prescribing’.
Monitoring drug therapy
Monitoring drug therapy usually involves trying to measure the clin-
ical response directly. If this is difficult, or is not related directly in
time to a dose of the drug, a surrogate measure of the response may
be required. In some cases, it may be necessary to resort to measure-
ment of the plasma concentration of the drug.
Monitoring the therapeutic effects of drugs
Some events can be directly monitored in the individual patient.
Examples of therapeutic events that can be monitored in the indi-
vidual include frequency of seizures during anticonvulsant drug
therapy, muscle power during treatment of myasthenia gravis, the
frequency of attacks of angina pectoris, and body weight during di-
uretic therapy.
Preventive measures in medicine often cannot be monitored in the
individual patient and their impact has to be predicted from popula-
tion studies (usually a clinical trial). Examples include the frequency
of infections after immunization, the reduction in NSAID-​induced
peptic ulceration with a proton pump inhibitor, or the prevention
of the complications of myocardial infarction by the use of thromb-
olysis and aspirin.
Monitoring the pharmacodynamic effects of drugs
In some circumstances, the pharmacological effect of a drug can
be carefully measured, followed sequentially, and used as a guide
to drug therapy even though it may not be correlated precisely with
the therapeutic effect. Examples include the effect of insulin on the
blood glucose concentration in diabetes mellitus, anticoagulants on
the prothrombin time, bronchodilators on FEV1 and peak flow rate
in asthma, and cancer chemotherapy on tumour markers.
Monitoring drug pharmacokinetics (therapeutic
drug monitoring)
This is useful for drugs where:
•	the clinical evidence of therapeutic or toxic effects is difficult to
measure or interpret
•	the relation between dose and plasma concentration is
unpredictable
•	the drug has a low therapeutic index
•	the plasma concentration of the drug is a good predictor of
response
There are only a handful of drugs that meet these requirements (see
Table 2.6.6). For these drugs, monitoring drug levels can be used
to individualize therapy (e.g. at the start of drug dosing, when the
relation between dose and plasma concentration in the individual
is uncertain or rapid changes in renal function alter the relation
between dose and plasma concentration), to monitor toxicity and
to assess compliance (see earlier section, ‘The patient’s use of a
drug: Compliance and concordance’).
Phenytoin
Plasma concentrations of phenytoin in the toxic range are quite well
related to its acute neurotoxic effects, but not to its long-​term adverse
reactions, such as gingival hyperplasia, hirsutism and acne, and folate
and vitamin D deficiencies. At low dosages it takes about 2 weeks of
maintenance therapy to reach steady state after a change in dose, but
because of its non​linear kinetics (Fig. 2.6.5), the half-​life lengthens at
higher plasma concentrations; so it can take up to 3 weeks or longer in
some patients to reach steady state. For this reason the dosage should
not be changed frequently. Provided the sample is not taken too soon
after a dose (i.e. within 1 to 2 h), the time of sampling for phenytoin is
not critical, as peak–​trough fluctuation is small between doses.
Digoxin
Plasma digoxin concentrations correlate well with toxic effects
but not with the therapeutic effect within the therapeutic dosage
range. The time of blood sampling should be at least 6 h after the
last dose. During regular maintenance dosage without a loading
Table 2.6.6  Drugs that commonly require therapeutic
concentration monitoring and their reference ranges
Drug
Concentration below
which a therapeutic
effect is unlikely
Concentration above which a
toxic effect is more likely
Gentamicina
5 µg/​ml (peak)
12 µg/​ml (peak), 2 µg/​ml (trough)
Digoxin
1.0 nmol/​litre
3.8 nmol/​litre
Ciclosporinb
80–​200 nmol/​litre
170–​300 nmol/​litre
Lithium
0.4 mmol/​litre
1.0 mmol/​litre
Phenytoin
40 µmol/​litre
80 µmol/​litre
Theophylline
55 µmol/​litre
100 µmol/​litre
a Conventional dosage regimens.
b Actual range will vary between laboratories.
2.6  Principles of clinical pharmacology and drug therapy
99
dose, steady state is reached after about 7 days (normal renal func-
tion) to more than 14 days (functionally anephric). The pharma-
cological response to a given plasma level of digoxin is dependent
on thyroid function (hyperthyroidism decreases responsiveness
and hypothyroidism increases it) and the plasma potassium (hypo-
kalaemia increases responsiveness and hyperkalaemia reduces it).
Children younger than 6 months have lower plasma digoxin con-
centrations at a given dose than older children and adults, and they
are also more resistant to the pharmacodynamic actions of digitalis;
so plasma digoxin concentrations cannot be clearly interpreted in
this age group.
Lithium
The therapeutic range is 0.4 to 1 mmol/​litre. In the range of 1 to
1.5 mmol/​litre, the incidence of both acute toxicity and long-​term
adverse effects is increased. Concentrations above 1.5 mmol/​litre
should be avoided. Blood samples should be taken at least 12 h after
the last dose. It takes about 3 days for steady state to be reached
during regular maintenance therapy, but patients vary widely. It may
take up to a week.
Monitoring the plasma lithium concentration is necessary for
several reasons. Lithium is nephrotoxic and excreted by the kid-
neys, so if toxicity occurs it is self-​perpetuating. Systemic avail-
ability varies from person to person and is altered by diarrhoea.
It also varies widely between formulations, which cannot be used
interchangeably. Sodium balance also affects renal excretion of
lithium. For example, diuretic-​induced renal sodium loss reduces
renal lithium excretion and can precipitate toxicity.
Aminoglycoside antibiotics
Gentamicin is the most widely used aminoglycoside antibiotic, but
the principles hold for other aminoglycosides. The relation between
the plasma concentration of gentamicin and its therapeutic efficacy
is complicated by the fact that different organisms have different
sensitivities to the antibiotic. Gentamicin is renally excreted, so in
renal impairment it accumulates unless the dose frequency (and
eventually the dose itself) is reduced.
The toxic effects of gentamicin on the inner ear and kidneys
are related to the ‘peak’ concentration (usually taken 1 h after an
intramuscular injection or the start of an intravenous infusion)
and the ‘trough’ concentration (taken just before the next dose).
These should be measured after three or four doses, or sooner
if there is renal impairment. A peak plasma concentration of 5
to 9 mg/​litre is generally considered necessary, although when
gentamicin is used together with benzylpenicillin to treat bac-
terial endocarditis, lower plasma gentamicin concentrations may
be effective. Bacteriological measurement of in vitro inhibitory
concentrations will help to guide therapy. If there is uncertainty,
expert advice on dosing and target plasma concentrations should
be sought.
Theophylline
Plasma theophylline concentrations correlate well with therapeutic
and toxic effects. Measurement is essential in any patient who has
been taking oral theophylline before it is given intravenously. It is
also important in smokers who usually have increased theophylline
clearance and hence require higher maintenance doses.
Ciclosporin
Ciclosporin is generally measured in whole blood, and the result
of the assay may depend on whether the measurement technique
is by immunoassay or high-​performance liquid chromatography.
The time to steady state is about 2 days and samples should be taken
at trough (just before the next dose). The whole-​blood concentra-
tion of ciclosporin can be affected by reduced absorption (due to
diarrhoea or reduced bile-​salt production) or altered bioavailability
(due to liver disease, coadministered drugs such as ketoconazole
and rifampicin, grapefruit juice, or St John’s wort).
FURTHER READING
Clinical pharmacology
Brown MJ, Sharma P, Mir FA, Bennett PN (2019). Clinical pharma-
cology, 12th Edition. Elsevier, North Holland.
Ritter JM, et al. (2008). A textbook of clinical pharmacology, 5th edition.
Arnold, London.
Pharmacological effects of drugs
Brunton LL, et al. (2012). Goodman & Gilman’s the pharmacological
basis of therapeutics, 12th edition. Mc-​Graw Hill, London.
Rang HP, et  al. (2019). Rang & Dale’s pharmacology, 9th edition.
Elsevier, Amsterdam.
Pharmacokinetics
Gibaldi M, Perrier D (1982). Pharmacokinetics, 2nd edition. Marcel
Dekker, New York.
Rowland M, Tozer TN (1995). Clinical pharmacokinetics. Concepts
and applications, 3rd edition. Lea & Febiger, Philadelphia, PA.
Adverse effects of drugs
Aronson JK (ed) (2015). Meyler’s side effects of drugs, 16th edition.
Elsevier, Amsterdam.
Websites
Medicines and Healthcare Products Regulatory Agency (MHRA).
https://​www.gov.uk/​government/​organisations/​medicines-​and-
​healthcare-​products-​regulatory-​agency/​
UL Cochrane Centre. http://​www.cochrane.co.uk/​
World Health Organization (1994). Guide to Good Prescribing. http://​
whqlibdoc.who.int/​hq/​1994/​WHO_​DAP_​94.11.pdf

2.7 Biological therapies for immune, inflammatory,
2.7 Biological therapies for immune, inflammatory, and allergic diseases 100
ESSENTIALS
Therapeutic monoclonal antibodies and related molecules are
increasingly used to treat immune-​mediated and inflamma-
tory diseases. They interact very precisely with a soluble or cell-​
bound ligand to have three predominant effects:  neutralization
of proinflammatory cytokines or growth factors; modulation of
intercellular interactions; or depletion of pathogenic cells. They
deliver significantly enhanced specificity in comparison to trad-
itional synthetic drugs and have delivered substantial improve-
ments in clinical outcomes in many immune, inflammatory, and
allergic diseases. However, there are no biomarkers to help de-
cide which biological class to try first in a particular patient, hence
one of the main challenges for the future is the identification of
‘stratifiers’ to guide therapy.
Introduction
A biological therapy is a medicinal product that is produced from a
biological source, usually a living cell, in contrast to a synthetic, small
molecule drug. Biological therapies include growth factors, hor-
mones, and other recombinant therapeutic proteins, but this chapter
will focus on therapeutic monoclonal antibodies (mAbs) and related
molecules in the context of immune-​mediated and inflammatory
diseases (Table 2.7.1). The overriding concept underpinning these
molecules is that they are ‘targeted therapies’, developed to interact
very precisely with a soluble or cell-​bound ligand. In this way they
have significantly enhanced specificity in comparison to traditional
synthetic drugs.
In the 30 years since the first mAb was licensed for acute trans-
plant rejection the field has expanded rapidly (see Chapter 3.8).
These complex molecules are now a multibillion-​dollar industry
and occupy an enlarging space in the global pharmaceutical
market. Growth has been catalysed by increased knowledge of dis-
ease pathologies but also by advances in protein engineering and
bioprocessing, which have improved the efficiency of therapeutic
protein production as well as the quality of the products.
The development of therapeutic antibodies
and derivatives
In the 1970s Kohler and Milstein developed technology that enabled
the isolation and immortalization of B-​cell clones (hybridomas)
producing a particular type of antibody. The product of each clone
was a mAb of unique and defined specificity, which raised hopes of
modern-​day ‘magic bullets’ to target cancers and other diseases.
The first licensed mAb was OKT3, a murine mAb against the
antigen CD3 on T cells. This reversed steroid-​refractory transplant
rejection with significant success, despite a cytokine release syn-
drome precipitated by T-​cell activation. A further limitation was
immunogenicity, a consequence of the recipient’s immune system
viewing the murine mAb as a foreign protein. The resulting immune
(antiglobulin) response neutralized the mAb and accelerated its
clearance, frustrating attempts to retreat patients and occasionally
causing hypersensitivity reactions.
Concurrent advances in recombinant DNA technology, how-
ever, enabled the cloning, manipulation, and recombination of
antibody genes from hybridomas (Fig. 2.7.1). The next generation
of therapeutic mAbs were chimeric, whereby the murine variable
(V) region, responsible for target-​binding, remained intact but was
linked to a human constant (C) region, with an incremental reduc-
tion in ‘foreignness’. In a further iteration, the complementarity-​
determining regions (CDRs) of the murine mAb (the V-​region
elements that determined specificity) were left intact but embedded
in a human framework, providing ‘humanized’ mAbs. Ultimately,
‘fully human’ mAbs were produced, either from transgenic animals
whose genomes had been manipulated to produce human anti-
bodies upon immunization; or by in vitro techniques such as phage
display technology.
As well as reducing potential immunogenicity these techniques
enabled the engineering of antibody C-​regions, to enhance or reduce
effector function, and thereby the fate of the mAb target. Antibody
fragments were also generated, both intact Fab fragments and smaller
elements that retained target-​binding capacity; other technologies
enabled the generation of antibody-​drug conjugates and bispecific
antibodies of dual specificity. Receptor-​fusion proteins resemble
2.7
Biological therapies for immune,
inflammatory, and allergic diseases
John D. Isaacs and Nishanthi Thalayasingam
2.7  Biological therapies for immune, inflammatory, and allergic diseases
101
Table 2.7.1  Currently licensed biological therapies for immune, inflammatory, and allergic diseases
Drug name
Target
Structure
Licensed indication(s)
Route of
administration
Notes
Abatacept
CD80/CD86
Fusion protein of CTLA4 EC
domain and human IgG1 Fc
RA, JIA
SC (RA); IV (RA, JIA)
Fc is modified to prevent
cytotoxicity
Adalimumab
TNF
Fully human mAb
RA; JIA; CD; UC; AS; PsA; Ps;
Hidradenitis suppurativa, uveitis
SC
Biosimilar available
Alemtuzumab
CD52
Humanized mAb
MS
IV
Also used extensively for lymphoma/
leukaemia and conditioning of solid
organ and bone marrow transplant
recipients
Anakinra
IL-1
Recombinant human IL-1
receptor antagonist
RA; CAPS; Still’s disease
SC
Recombinant form of naturally
occurring IL-1 receptor antagonist
Atacicept
BLyS, APRIL
Fusion protein of TACI EC
domain and human IgG1 Fc
SLE
SC
In development
Belimumab
Blys
Fully human mAb
SLE
IV
Benralizumab
IL-5 receptor
Humanized mAb
Eosinophilic asthma
SC
Brodalumab
IL-17RA
Fully human mAb
Ps
SC
Canakinumab
IL-1β
Fully human mAb
Gout; JIA (systemic); CAPS; Still’s
disease
SC
Certolizumab
Pegol
TNF
Humanized mAb Fab'
fragment linked to PEG
RA; AS; PsA; Ps
SC
Denosumab
RANKL
Fully human mAb
Osteoporosis
SC
Cases of osteonecrosis of the jaw
and atypical femoral fractures have
been reported
Etanercept
TNF
Lymphotoxin
Fusion protein of human p75
TNF receptor EC domain and
human IgG1 Fc
RA; AS; JIA; PsA; Ps
SC
Ineffective in CD
Biosimilar available
Etrolizumab
β7 integrin
Humanized mAb
IBD
SC
In development
Golimumab
TNF
Fully human mAb
RA; AS; PsA; UC
SC
Guselkumab
IL-23
(p19 subunit)
Fully human mAb
Ps
SC
Infliximab
TNF
Chimeric mAb
RA; CD; UC; PsA; AS, Ps
IV
Biosimilar available
Ixekizumab
IL-17A (both
IL-17A and
IL-17A/F)
Humanized mAb
PsA; Ps
SC
Can exacerbate IBD
Mepolizumab
IL-5
Humanized mAb
Eosinophilic asthma
SC
Natalizumab
α4 integrin
Humanized mAb
MS
IV
Increased susceptibility to PML
Ocrelizumab
CD20
Humanized mAb
MS
IV
Omalizumab
IgE
Humanized Mab
Allergic asthma; chronic
spontaneous urticaria
SC
Dose determined by serum IgE and
body weight in allergic asthma
Reslizumab
IL-5
Humanized mAb
Eosinophilic asthma
IV
Rituximab
CD20
Chimeric mAb
RA; GPA; MPA; NHL; CLL
IV
Dosing regimen differs
between rheumatological and
haematological indications
Biosimilar available
Sarilumab
IL-6 receptor
Fully human mAb
RA
SC
Secukinumab
IL-17A
Fully human mAb
Ps, PsA, AS
SC
Can exacerbate IBD
Tildrakizumab
IL-23
(p19 subunit)
Humanized mAb
Ps
SC
Tocilizumab
IL-6 receptor
Humanized mAb
RA; JIA; GCA
IV (RA, JIA); SC (RA)
Ustekinumab
IL-12 & IL-23
(p40 subunit)
Fully human mAb
Ps, PsA; CD
SC
Inhibits Th1 and Th17 pathways
Vedolizumab
α4β7 integrin
Humanized mAb
CD, UC
IV
Abbreviations: CD, cluster of differentiation; CTLA4, cytotoxic T-lymphocyte associated protein 4; EC, extracellular; Ig, immunoglobulin; Fc, fragment crystallisable; RA, rheumatoid
arthritis; JIA juvenile idiopathic arthritis; SC, sub-cutaneous; IV, intravenous; TNF, tumour necrosis factor; mAb, monoclonal antibody; CD, Crohn’s Disease; UC, ulcerative colitis; AS,
ankylosing spondylitis; PsA, psoriatic arthritis; Ps, psoriasis; MS, multiple sclerosis; IL, interleukin; CAPS, cryopyrin-associated periodic syndrome; BLyS, B-lymphocyte stimulator;
APRIL, a proliferation-inducing ligand; TACI, transmembrane activator and CAML interactor; SLE, systemic lupus erythematosus; RA, receptor A; Fab, fragment antigen binding;
PEG, polyethylene glycol; RANKL, receptor activator of nuclear factor kappa-B ligand; IBD, inflammatory bowel disease; PML, progressive multifocal leukoencephalopathy; GPA,
granulomatosis with polyangiitis; MPA, microscopic polyangiitis; NHL, non-Hodgkin lymphoma; CLL, chronic lymphocytic leukaemia; GCA, giant cell arteritis; Th, T helper cell.
102
section 2  Background to medicine
mAbs but comprise the extracellular domain of a cell surface receptor
fused to an immunoglobulin (Ig) Fc to form a soluble molecule that
is specific for the receptor’s ligand (Fig. 2.7.2). For further discussion
see account by H Waldmann and G Winter in Chapter 3.8.
Classification
The structure of a therapeutic mAb can be discerned by its no-
menclature and is comprised of: prefix, substem A, substem B, and
suffix. The prefix is unique to the drug and the suffix, -​mab, indicates
that it is a mAb. Substem B provides the species upon which the Ig
sequence is based: u for human, o for mouse, xi for chimeric, and zu
for humanized. Substem A indicates the class of drug (e.g. -​l(i)-​ for
immunomodulatory, k(i) for interleukin and t(u) for tumour). Some
examples are provided in Table 2.7.2.
Targets for biological therapies in
immune-​mediated inflammatory diseases
In immune-​mediated inflammatory diseases, biological therapies
act in three predominant ways: to neutralize proinflammatory cyto-
kines or growth factors; to modulate intercellular interactions; or to
deplete pathogenic cells (see Fig. 2.7.3 and Table 2.7.1).
Cytokine and growth factor targets
Tumour necrosis factor (TNF)
TNF is a pivotal cytokine mediator in a variety of immune-​mediated
inflammatory diseases. It exists in soluble and transmembrane
forms, and binds to type 1 and type 2 receptors to mediate a var-
iety of proinflammatory and cell survival effects. TNF blockade
was the first of the modern-​day biological therapy paradigms and
has been licensed for adults and children with a variety of rheum-
atological conditions (rheumatoid arthritis, ankylosing spondyl-
itis, psoriatic arthritis, juvenile idiopathic arthritis), as well as for
psoriasis, Crohn’s disease, ulcerative colitis, uveitis and hidradenitis
suppurativa (Table 2.7.1). There are five anti-​TNF drugs currently
licensed. Two are fully human mAbs, one is a chimeric mAb, one
(certolizumab) is a PEGylated Fab’ fragment and the last is a sol-
uble receptor. The latter (etanercept) also binds lymphotoxin α3 but
there is nothing to suggest this is relevant to its efficacy or toxicity.
Certolizumab is the only anti-​TNF manufactured in E. coli, which
should increase yields over mammalian cell culture. The PEG moiety
increases the half-​life by reducing renal clearance and proteolysis.
TNF blockade was licensed for rheumatoid arthritis at the start of
the 21st century and was revolutionary. Patients who had been re-
fractory to all treatment modalities suddenly experienced previously
unimagined symptomatic improvements in joint pain and stiffness,
and fatigue. Furthermore, TNF blockade prevented the relentless
joint damage seen with more traditional therapies. Consequently,
Heavy chain
Light chain
CDRs
VARIABLE REGION
CONSTANT REGION
(a)
(b)
MURINE
CHIMERIC
-XIMAB
HUMANISED
-ZUMAB
‘FULLY
HUMAN’
-MUMAB
Fig. 2.7.1  The different types of therapeutic antibody. (a) Basic antibody
structure. (b) Basic structure of a murine, chimeric, humanized, and
human monoclonal antibody. Red indicates murine sequence and black
indicates human sequence. CDR, complementarity-​determining region.
Reproduced from: Isaacs JD. Antibody engineering to develop new antirheumatic
therapies. Arthritis Res Ther 2009, 11, 225.
disulphide bonds
VH
VL
CH1
CL
(b)
VH
VL
CH1
CL
VH
VL
CH1
CL
CH2
CH2
CH3
CH3
Hinge
(a)
Fc
(c)
CH2
CH2
CH3
CH3
receptor fragment
receptor fragment
Fig. 2.7.2  The domain structures of an antibody molecule and derivatives. (a) An antibody molecule.
(b) A fragment antigen-​binding (Fab) fragment. (c) A receptor-​immunoglobulin fusion protein. CH, heavy
chain constant domain; CL, light chain constant domain; Fc, fragment crystallizable; VH, heavy chain variable
domain; VL, light chain variable domain. NB: A Fab’ fragment is a Fab fragment plus the hinge region.
Reproduced from: Isaacs JD. Antibody engineering to develop new antirheumatic therapies. Arthritis Res Ther 2009, 11, 225.
2.7  Biological therapies for immune, inflammatory, and allergic diseases
103
patients experienced less disability than previous generations with
retained employment prospects and reduced need for joint surgery.
Registry data demonstrated a slightly enhanced risk of serious in-
fection (about 20% higher than with synthetic antirheumatic drugs
such as methotrexate) but no enhanced risk of malignancy. An un-
expected complication was the emergence of tuberculosis (TB),
often disseminated and with unusual histology. Because of their
targeted nature biological therapies have provided unexpected in-
sights into pathophysiology. In the case of TB, we learnt that TNF
was essential for the maintenance of granulomas, which in turn con-
trolled latency in individuals with prior exposure. The dissolution of
granulomas with TNF blockade thereby released TB bacilli into the
bloodstream with extrapulmonary reactivation. Patients prescribed
TNF blockade are now routinely screened for prior TB exposure and
treated with appropriate prophylaxis where indicated. The soluble
receptor etanercept is less likely to reactivate TB. The precise reason
is unclear but may reflect a reduced capacity to disrupt granulomata
due to an incapacity to perform so-​called ‘inside-​out’ signalling
upon binding cell surface TNF. TNF blockade is also associated with
an enhanced susceptibility to opportunistic intracellular pathogens
such as listeria, salmonella, and some fungi.
The effects of TNF blockade were similarly impressive in patients
with severe Crohn’s disease, particularly those with fistulae and peri-
anal disease in which healing was rapid and complete. When used in
the first two years of disease, around 50% of patients achieve remis-
sion and even mucosal healing. Perhaps with parallels to TB reactiva-
tion, etanercept is not effective in this granulomatous disease of the
gastrointestinal (GI) tract. TNF blockade is also effective in ulcera-
tive colitis. Psoriasis provided a further successful disease target, at
a time when systemic treatments were relatively limited. Infliximab
provided 75% improvement in the Psoriasis Area and Severity Index
(PASI) in more than 80% of patients, and 90% improvements in 50–​
60%. These drugs were also effective in psoriatic arthritis and an-
kylosing spondylitis, for which therapeutic options were even more
limited than for rheumatoid arthritis. In both diseases they provide
significant improvement in disease activity, including spinal inflam-
mation and function.
In contrast, TNF blockade was not successful in other indica-
tions characterized by TNF excess, perhaps illustrating homeostatic
functions of TNF. These included multiple sclerosis and congestive
heart failure, in which trials were stopped prematurely for disease
worsening. Uncommon and unexpected adverse effects of TNF
blockade include neutropenia and vasculitis and, paradoxically,
psoriasis can be triggered in some patients. Drug-​induced lupus can
also occur, including renal and cerebral manifestations.
IL-​6
IL-6 is a key local and systemic inflammatory mediator. Its receptor
exists in membrane-​bound and soluble forms. After binding IL-​6,
the soluble receptor pairs with its signalling component, gp130, on
any cells that express it. Tocilizumab binds both forms of the IL-​6
receptor, effectively preventing IL-​6 signalling. The protean effects
of IL-​6 include: activation of B and T cells, and macrophages; the
differentiation of macrophages into osteoclasts; synovial fibroblast
proliferation and synovial neovascularization; production of plate-
lets from megakaryocytes; production of acute phase reactants
and hepcidin from the liver; and systemic effects such as fever and
Table 2.7.2  Mab nomenclature
Prefix
Substem A
Substem B
Suffix
Infliximab
Inf
li
Xi
Mab
Tocilizumab
Toci
li
Zu
Mab
Canakinumab
Cana
ki
nu
Mab
Rituximab
Ri
tu
xi
Mab
(a)
 
(b)
(c)
Fig. 2.7.3  Potential mechanisms of action of immunomodulatory mAbs and receptor-​immunoglobulin
fusion proteins. (a) Neutralization of soluble mediators. (b) Cell lysis. (c) Inhibition of intercellular or
cell: endothelial interactions.
104
section 2  Background to medicine
fatigue. Il-​6 levels are elevated in rheumatoid arthritis, particularly
within the joint, and administration of tocilizumab provides re-
ductions in disease activity similar to TNF blockade. In line with
hepatic effects of IL-​6 there is a robust normalization of the acute
phase response and, in patients with anaemia of chronic disease, a
rise in haemoglobin as hepcidin levels fall. Elevated platelet counts
also fall. Dyslipidaemia may occur in patients receiving tocilizumab
with elevations in HDL, LDL, and triglycerides, usually without a
rise in atherogenic index. Nonetheless lipid rises should be treated
according to relevant guidelines. Diverticular perforations have oc-
curred in patients receiving tocilizumab, which should therefore be
used with caution where there is a history of intestinal ulceration or
diverticulitis. The aetiology is unclear but could reflect an important
homeostatic role of IL-6. Reversible neutropenia and transaminase
elevation are also common. In other respects, the efficacy and ad-
verse event profile of IL-6 blockade resemble those of TNF blockade.
In particular there is a slightly enhanced risk of serious infections.
IL-​1
IL-​1α and IL-​1β are proinflammatory and bind the IL-​1 receptor
(IL-​1R). The naturally occurring IL-​1 receptor antagonist (IL-1Ra)
competes with IL-​1α and β for binding to IL-​1R. IL-​1 inhibition can be
achieved with anakinra (recombinant IL-1ra), or with canakinumab
(anti-​IL-​1 mAb). Anakinra effectively blocks IL-​1α and IL-​1β but
its short half-​life requires daily dosing. It is licensed for rheuma-
toid arthritis and for the autoinflammatory cryopyrin-​associated
periodic syndromes (CAPS), in which the activated inflammasome
results in increased secretion of IL-​1β. These include neonatal-​
onset multisystem inflammatory disease (NOMID), chronic in-
fantile neurological, cutaneous, articular syndrome (CINCA),
Muckle–​Wells syndrome, and familial cold autoinflammatory
syndrome (FCAS). In rheumatoid arthritis, despite elevated IL-​1
levels in plasma and synovial fluid, the efficacy of anakinra is less
than that of the other biological drugs. Canakinumab is an anti-​IL-​
1β human mAb, licensed for the treatment of CAPS, systemic ju-
venile idiopathic arthritis (JIA), and gouty arthritis. In gout, urate
crystals are ingested by neutrophils and monocytes, activating the
inflammasome with the release of IL-​1β.
IL-​17
IL-​17A is the signature cytokine of Th17 T cells which, in health,
protect against bacterial and fungal infections, particularly candid-
iasis. Other members of this cytokine family include IL-17F, IL-17C,
and IL-17A/​F. The various family members bind receptors that share
the subunit IL-​17RA. Th17 T cells are also associated with a number
of autoimmune diseases including rheumatoid arthritis, psoriasis,
inflammatory bowel disease, and ankylosing spondylitis. The func-
tions of IL-​17 overlap with those of TNF and IL-​6, while also stimu-
lating the production of additional proinflammatory cytokines and
chemokines. Akin to IL-​6, IL-​17 promotes osteoclastogenesis and
angiogenesis. Secukinumab is a fully human anti-​IL-​17A antibody.
The drug penetrates the skin and reduces local inflammation asso-
ciated with psoriasis. In a head-​to-​head study its efficacy exceeded
that of etanercept, with 54% of patients achieving PASI 90 and 24%
PASI 100 after 12 weeks. It is also an effective treatment for psoriatic
arthritis and ankylosing spondylitis. While IL-17A is abundantly
present in the gut mucosa in inflammatory bowel disease (IBD),
secukinumab was ineffective in Crohn’s disease and associated with
infectious complications. Furthermore, IBD exacerbations were re-
ported in trials of secukinumab for other indications. Blockade of
IL-​17A is associated with infections, including mucosal and, occa-
sionally, systemic candidiasis. Brodalumab targets IL-17-​RA, thereby
antagonizing the different IL-​17 family members; it is licensed for
use in moderate to severe plaque psoriasis.
B-​lymphocyte stimulator
B-​lymphocyte stimulator (BLyS), also known as B-​cell activating
factor, is a B-​cell survival factor and a member of the TNF super-
family. Levels are elevated in certain autoimmune diseases such as
systemic lupus erythematosus (SLE), where there is an association
with disease activity. Inhibition of BLyS leads to impaired B-​cell
survival and reduced differentiation of B cells into plasma cells.
Belimumab is a fully human anti-​BLys mAb. It is the only biological
drug to be approved in the treatment of SLE, with a licence for pa-
tients with active, autoantibody positive disease. Administration of
belimumab is associated with reductions in IgG levels, including
autoantibodies; longer-​term administration is associated with a fall
in circulating B cells. Bacterial infections are more common in pa-
tients receiving belimumab. Atacicept (TACI-​Ig) is a fusion protein
comprising the ligand binding portion of the BLyS receptor TACI
(transmembrane activator and CAML interactor) and a human IgG1
Fc. As well as binding BLyS, TACI also binds a proliferation-​inducing
ligand (APRIL), which is an additional B-​cell growth factor. Hence
atacicept has a broader anti-​B-​cell profile than belimumab, and is
currently under investigation in SLE.
IL-23
Th17 differentiation is partly controlled by IL-​23, produced by den-
dritic cells and by Th17 cells themselves. IL-​23 comprises two sub-
units, p40 shared with IL-​12 (important in the differentiation of
Th1 cells), and p19. Ustekinumab is a fully human anti-​p40 mAb
that blocks the actions of both IL-​23 and IL-​12. It is effective for
treating severe plaque psoriasis and psoriatic arthritis although
showed inferiority to secukinumab in a head-​to-​head trial in psor-
iasis. It is also licensed for use in Crohn’s disease, in contrast to IL-17
blockade. Guselkumab, which blocks IL-23 p19 (and therefore spe-
cific for IL-23 vs. IL-12) is in development for similar indications,
and is licensed for use in moderate to severe plaque psoriasis.
Modulation of intercellular interactions
A second strategy for biological therapies involves the inhibition of
cell interactions, which can prevent activation or migration.
Costimulation blockade
T cells require two critical signals for their activation. The first is
the recognition, by the T-​cell receptor, of their cognate antigen (a
complex of major histocompatibility complex (MHC) molecule and
peptide) on an antigen presenting cell (APC). The second is the rec-
ognition of CD80 and CD86 on the APC by CD28, which provides
an essential costimulatory signal. Following activation the T-​cell
upregulates CTLA4, which displaces CD28 from CD80 and CD86
due to its higher affinity. CTLA4 imparts a negative signal to the
T-​cell, thereby acting as a ‘brake’ to activation, returning the T-​cell
to its resting state. Abatacept, or CTLA4-​Ig, is a fusion protein of the
extracellular domain of CTLA4 and a modified Fc region of human
IgG1 (Fig. 2.7.4). This molecule competes with CD28 for binding to
2.7  Biological therapies for immune, inflammatory, and allergic diseases
105
CD80 and CD86, thereby blocking costimulation and inhibiting the
activation of naïve T cells. The Fc modification is designed to pre-
vent abatacept from killing APCs to which it binds. It is an effective
agent in patients with rheumatoid arthritis, polyarticular juvenile
idiopathic arthritis, and psoriatic arthritis. It can be administered by
monthly infusion or weekly subcutaneous injection.
Adhesion molecule blockade
Natalizumab is a humanized anti-​α4-​integrin mAb which blocks
the interaction of α4β1 integrin with its receptor vascular cell ad-
hesion molecule-​1 (VCAM-​1) and α4β7 integrin with mucosal
addressin cell adhesion molecule-​1 (MadCAM-​1). This prevents
the transendothelial migration of integrin-​expressing leucocytes
into inflamed tissues. In multiple sclerosis VCAM-​1 is upregulated
on CNS endothelium and possibly glial cells and natalizumab is an
effective therapy, administered by four-​weekly infusion. However,
its use is associated with an increased risk of progressive multifocal
leukoencephalopathy (PML), the opportunistic infection associated
with John Cunningham (JC) virus. Patients receiving natalizumab
for multiple sclerosis must be fully informed of this risk and carry
a patient alert card. While also effective in Crohn’s disease, by
inhibiting leucocyte adhesion to MadCAM-​1 on gut endothelial
cells, natalizumab is rarely used for this indication as a result of the
PML risk.
In contrast, vedolizumab is a humanized anti-​α4β7 mAb. α4β7
integrin is expressed preferentially on a specific subset of gut-​
homing memory T cells that cause inflammation in inflammatory
bowel disease. Vedolizumab therefore blocks leucocyte trafficking
to the gut, with a licence for both Crohn’s disease and ulcera-
tive colitis. Etrolizumab is an anti-​β7 humanized mAb that binds
both α4β7 and αEβ7 integrins. The latter is selectively expressed
on gut mucosal intraepithelial T cells, where it binds E-​cadherin.
Hence etrolizumab can prevent T cells trafficking to the gut but
may also block their retention. It is currently under investigation
in inflammatory bowel disease, having demonstrated efficacy in a
phase 2 trial in ulcerative colitis.
Depletion of pathogenic cells
Alemtuzumab, which targets CD52 on lymphocytes, monocytes,
macrophages, and NK cells, was the first humanized therapeutic
mAb. It is unusual in having its origins in a rat rather than a mouse
mAb. It is potently cytotoxic and, while first developed to treat leu-
kaemia and lymphoma, it was subsequently studied in a range of
autoimmune diseases. It is licensed for relapsing remitting mul-
tiple sclerosis, administered as two short, annual treatment cycles.
Its use is paradoxically associated with the development of sec-
ondary autoimmunity. The most common manifestation is thy-
roid autoimmunity, but more serious complications have included
thrombocytopenic purpura, antiglomerular basement membrane
disease, and other haematological cytopenias. Administration of
alemtuzumab is also associated with significant infusion reactions,
which can be reduced with appropriate premedication that includes
a glucocorticoid. Oral prophylaxis for herpes virus infections is also
recommended for at least one month after each treatment cycle.
Rituximab is a chimeric human IgG1 anti-​CD20 mAb, originally
developed for the treatment of B-​cell malignancies but subsequently
established as an effective treatment for rheumatoid arthritis and
certain types of vasculitis. It is potently cytotoxic, probably a conse-
quence of efficient antibody-​dependent cellular cytotoxicity, comple-
ment-​mediated cytotoxicity, and apoptosis induction. In rheumatoid
arthritis it is administered as two intravenous infusions spaced two
weeks apart, again with prophylaxis against infusion reactions. CD20
is expressed on all B cells but not on haematopoietic stem cells,
pro-​B cells, or plasma cells. Consequently, after initial depletion B
cells reconstitute and there is minimal loss of humoral immunity
due to survival of plasma cells. Reconstitution occurs typically
after 6 months, but with differential rates in distinct B-​cell subsets.
Retreatment for rheumatoid arthritis is usually administered when
MHC
TCR
CD28
CD80/86
Antigen
presenting cell
T cell
Signal 1
Signal 2
(a)
CD80/86
Antigen
presenting cell
T cell
Signal 1
(b)
CTLA4
MHC
TCR
CD28
MHC
CD80/86
Antigen
presenting cell
TCR
T cell
Signal 1
(c)
CD28
CTLA4-Ig
Fig. 2.7.4  The structure and mode of action of abatacept. (a) T cells require two signals for full activation.
Signal 2 (coactivation) derives from the interaction of CD28 with B7.1 and B7.2. (b) CTLA4 is upregulated
following T-​cell activation and competes with CD28 due to its higher affinity, de-​activating the T-​cell.
(c) Abatacept (CTLA4-​Ig) is a fusion protein of the extracellular domain of CTLA4 and an IgG Fc.
106
section 2  Background to medicine
symptoms start to recur but at least 6 months after the initial course.
There is no recommended upper limit to the number of rituximab
courses that can be administered but repeated courses are associ-
ated with secondary hypogammaglobulinaemia, most commonly
IgM but occasionally IgG. The latter may be associated with an en-
hanced infection risk and the risk:benefit ratio of continued treat-
ment should be considered before each course of therapy. PML cases
have also been reported in patients who have received rituximab but
most commonly in patients with leukaemia or lymphoma who have
received additional immune suppression. Nonetheless all patients
should be counselled about this risk and receive a patient alert card.
Rituximab is also licensed for use in granulomatosis with polyangiitis,
microscopic polyangiitis, and pemphigus vulgaris.
Immunogenicity
The development of chimeric, humanized, and fully human mAbs
was expected to reduce immunogenicity associated with murine
components of antibody molecules. While this has largely been the
case, immunogenicity cannot be eliminated altogether. MAb CDRs
are intrinsically immunogenic, being created by gene recombination
events and somatic mutations. There are also Ig allotypes which vary
between individuals. Antidrug antibodies (ADAs) can be difficult
to measure, hence their importance and extent is yet to be fully es-
tablished. Nonetheless they can undoubtedly interfere with efficacy
by neutralizing an mAb’s binding site or accelerating its clearance.
Clinically this is most likely to manifest as secondary non​response,
where an individual loses the benefit of a previously effective therapy.
Hypersensitivity reactions also occur, such as infusion reactions as-
sociated with anti-​infliximab ADAs, but anaphylactic reactions are
fortunately rare. Soluble receptors are less immunogenic as they
contain minimal foreign sequence.
Immunogenicity is influenced by many factors such as the primary
mAb sequence, its formulation (particularly the presence of aggre-
gates), the route of administration and precise treatment regime,
disease under treatment, and coadministered therapy. For example,
anti-​TNF mAbs are more immunogenic in rheumatoid arthritis than
in ankylosing spondylitis or psoriatic arthritis; and the frequency of
ADAs in rheumatoid arthritis patients is reduced by coadministration
of methotrexate. Even with such cotherapy, however, almost a third
of patients receiving the fully human mAb adalimumab generated
ADAs, most commonly within the first 6 months of therapy, which
significantly reduced efficacy. Measurement of circulating drug is
often easier than measuring ADAs, and a number of algorithms have
been proposed whereby measurement of drug levels with or without
ADAs should render mAb therapy more personalized and cost-​
effective. Nonetheless, controlled trials with economic analyses are
ideally needed before so-called ‘therapeutic drug monitoring’ can be
recommended in routine practice.
Biosimilars
The availability of biological therapies has been restricted or ra-
tionalized in most parts of the world, due in large part to their high
cost. However, as the patents protecting the original biological
drugs expire, there is the potential for biosimilar molecules to be
developed at lower cost. Unlike synthetic small molecule drugs, it
is not possible to create an exact replica of a therapeutic mAb. This
is because many features, particularly post-​translational modifica-
tions, depend less on the primary amino acid sequence and more
on cell culture conditions and downstream manufacturing, pro-
cessing, and production. Furthermore, some of these features are
difficult to characterize precisely, and a given batch of therapeutic
mAb is also characterized by ‘microheterogeneity’—​a mixture of
closely related but distinct molecules. Consequently, both the Food
and Drug Administration (FDA) and the European Medicines
Agency (EMA) have developed regulations for the development of
biosimilar biological products. These rely heavily on in vitro com-
parisons of the physicochemical characteristics and biological ac-
tivities of the originator and biosimilar biologicals, and less on data
from clinical trials in humans. While reducing the cost of biosimilar
development these guidelines mean that biosimilars can be licensed
having been administered to relatively few patients. Because it is im-
possible for a biosimilar to be identical to the originator biological,
this limited clinical experience has led to concerns over safety and
efficacy. There are additional questions around: indication extrapo-
lation (biosimilars can be licensed for the same indications as the
originator biological without requiring clinical trial data in each
indication); differences in immunogenicity compared to the ori-
ginator biological; substitution (the potential for a pharmacist to
substitute a biosimilar when an originator biological has been pre-
scribed); nomenclature and identification of the biosimilar product,
particularly when adverse effects occur. While smaller and less
complex than mAbs, biosimilar growth factors have been marketed
for several years without significant problems. Furthermore, there
is batch-​to-​batch variability of originator biologicals, particularly
when manufacturing processes change. Nonetheless the launch of
biosimilar mAbs has been seen as a major milestone and a number
of anti-TNF and rituximab biosimilars are now licensed and in use,
so far without unexpected complications.
Future developments
Stratified medicine
There are currently four classes of biological therapy for rheumatoid
arthritis: anti-​TNF, anti-​IL-​6R, B-​cell depletion, and costimulation
blockade. Surprisingly, the efficacy profile of each class is very similar,
with only 10–​20% of established rheumatoid arthritis patients
achieving optimal responses. Adverse event profiles are also largely
similar. Currently, however, there are no biomarkers available that
help the rheumatologist decide which biological class to try first in
a particular patient. A similar situation pertains to TNF blockade,
anti-p40 blockade, and integrin blockade in IBD. Because biological
therapies are costly this has negative health economic implications
as well as delaying the achievement of disease control, which may
require cycling through more than one agent. Consequently, there
has been a major investment in the identification of ‘stratifiers’ to
guide therapy examining, for example, genetics, transcriptional and
proteomic profiles in blood and, particularly in IBD, tissue charac-
teristics. In cancer the tissue often provides a clue, such as the use of
trastuzumab in HER2 positive breast cancer. In the respiratory field
omalizumab (humanized anti-​IgE) is indicated in allergic asthma pa-
tients who have a positive skin test or in vitro reactivity to a perennial
2.7  Biological therapies for immune, inflammatory, and allergic diseases
107
aeroallergen, and dosed according to serum IgE levels. Mepolizumab
(humanized anti-​IL-​5) is indicated only for patients with refractory
eosinophilic asthma. Once stratifiers are identified, new tests (com-
panion diagnostics) will be developed to assist in the most appro-
priate allocation of therapy.
Antibody fragments and novel species
Novel species include bispecific antibodies, which can now be pro-
duced using various technologies, as well as smaller antibody frag-
ments. Bispecific antibodies permit the simultaneous targeting of
two specificities by a single molecule. In inflammatory diseases, for
example, there is significant interest in the dual targeting of TNF and
IL-​17. Smaller antibody fragments, while suffering from a shorter
half-​life, should achieve better tissue penetration. This will be par-
ticularly important not only in oncology, but potentially also in in-
flammatory diseases.
FURTHER READING
Baker KF, Isaacs JD (2018). Novel therapies for immune-mediated in-
flammatory diseases: What can we learn from their use in rheumatoid
arthritis, spondyloarthritis, systemic lupus erythematosus, psoriasis,
Crohn’s disease and ulcerative colitis? Ann Rheum Dis, 77, 175–87.
Beniwal-​Patel A, Saha S (2014). The role of integrin antagonists in the
treatment of inflammatory bowel disease. Expert Opin Biol Ther, 14,
1815–​23.
Billiet T, et al. (2014). Targeting TNF-​α for the treatment of inflamma-
tory bowel disease. Expert Opin Biol Ther, 14, 75–​101.
Dörner T, Kay J (2015). Biosimilars in rheumatology: current perspec-
tives and lessons learnt. Nat Rev Rheumatol, 11, 713–​24.
Elvin JG, Couston RG, van der Walleb CF (2013). Therapeutic anti-
bodies: Market considerations, disease targets and bioprocessing.
Int J Pharmaceutics, 440, 83–​98.
Garbers C, et al. (2018). Interleukin-6: designing specific thera-
peutics for a complex cytokine. Nat Rev Drug Discov, 17(6),
395–412.
Hawkes JE, et al. (2018). Discovery of the IL-23/IL-17 signaling pathway
and the treatment of psoriasis. J Immunol, 201(6), 1605–13.
Isaacs JD (2009). Antibody engineering to develop new antirheumatic
therapies. Arthritis Res Ther, 11, 225.
Keating GM (2013). Abatacept: a review of its use in the management
of rheumatoid arthritis. Drugs, 73, 1095–​19.
Kohler G, Milstein C (1975). Continuous cultures of fused cells se-
creting antibody of predefined specificity. Nature, 256, 495–​7.
Krieckaert C, Rispens T, Wolbink G (2012). Immunogenicity of bio-
logical therapeutics: from assay to patient. Curr Opin Rheumatol, 24,
306–​11.
Maxwell LJ, et al. (2015). TNF-​alpha inhibitors for ankylosing spon-
dylitis. Cochrane Database Syst Rev, 4, CD005468.
Reich K, et al. (2008). Comparative effects of biological therapies on
the severity of skin symptoms and health-​related quality of life in
patients with plaque-​type psoriasis: a meta-​analysis. Curr Med Res
Opin, 24, 1237–​54.
Schmid AS and Neri D (2019). Advances in antibody engineering for
rheumatic diseases. Nat Rev Rheumatol, 15(4), 197–207.
Thalayasingam N, Isaacs JD (2011). Anti-​TNF therapy. Best Pract Res
Clin Rheumatol, 25, 549–​67.
Tiller KE, Tessier PM (2015). Advances in antibody design. Annu Rev
Biomed Eng, 17, 191–​216.
Winter G, Milstein C (1991). Man-made antibodies. Nature, 349(6307),
293–9.

2.8 Traditional medicine exemplified by traditiona
2.8 Traditional medicine exemplified by traditional Chinese medicine 108
ESSENTIALS
Traditional (often synonymous with complementary) medicine
can be defined as the knowledge, skill, and practices, based on
the theories, beliefs, and experiences indigenous to different cul-
tures, which are used in the maintenance of health and the pre-
vention, diagnosis, improvement, or treatment of illness. These
practices are found in almost every country, and demand for them
is increasing.
For individual patients, when of proven quality, safety, and ef-
ficacy, traditional medicine contributes to the goal of ensuring
that all people have access to care that they feel they need. For
medical science, investigators in traditional fields have discovered
new medicines with which to combat important diseases, includ­
ing the sympathomimetic ephedrine, the antimalarial artemisinin,
and arsenic trioxide used in the treatment of promyelocytic
leukaemia.
The World Health Organization Traditional Medicine Strategy
2014–​2023 has two key goals: to support Member States in harness­
ing the potential contribution to health, wellness, and people-​centred
healthcare, and to promote the safe and effective use of traditional
medicine through the proper and organized regulation of products,
practices, and practitioners. Much benefit will emerge if the science
of the West and the rich experience and traditions of medicine in
China can work together in an integrated way.
Introduction
Traditional and complementary medicine are important and often
underestimated aspects of healthcare. Not only are the practices
found in almost every country; demand is increasing. The im-
portance of the approach resides in two qualities: first, when it is
of proven quality, safety, and efficacy, traditional and complemen-
tary medicine contributes to the goal of ensuring that all people
have access to care that they feel they need. Secondly, several in-
vestigators in these traditional fields have over the years discovered
new medicines with which to combat important diseases. Here we
define traditional medicine and discuss several therapeutic contri-
butions exemplified by traditional Chinese medicine—​including
the strategic issues of disease prevention, health promotion, and
affordability.
What is traditional medicine?
The World Health Organization (WHO) defines traditional medi-
cine as follows: ‘Traditional medicine has a long history. It is the sum
total of the knowledge, skill, and practices based on the theories,
beliefs, and experiences indigenous to different cultures, whether
explicable or not, used in the maintenance of health as well as in the
prevention, diagnosis, improvement, or treatment of physical and
mental illness.’
The terms ‘complementary medicine’ or ‘alternative medicine’ are
used interchangeably with traditional medicine in some countries.
They refer to a broad set of healthcare practices that are not part of
that country’s own tradition or conventional medicine and are not
fully integrated into the dominant healthcare system.
Traditional Chinese medicine, Ayurveda, Unani, naturopathy,
osteopathy, homeopathy, and chiropractice are frequently included
in traditional medicine. In China, the earliest initiatives originated
from Shennong, thousands of years ago: he examined the proper-
ties and tasted many hundreds of herbs. Ayurveda, or the ‘science
of life’, was developed by the seers of ancient India. Practitioners
and dispensers now use Ayurveda for treatment. Unani medicine
is the traditional practice of South Asian countries and its use
is growing in other parts of Asia, such as Bangladesh and India.
Naturopathy emphasizes prevention and the promotion of optimal
health through the process of self-​healing. Homeopathy originated
from the belief that a substance which causes the symptoms of a
disease in healthy people would cure similar symptoms in those
who are unwell.
Commonly used treatments and therapeutic techniques for each
of these traditional approaches are listed in Box 2.8.1. Certain popu-
lations in Africa, Asia, and Latin America continue to use traditional
medicine to meet their primary healthcare needs. Meanwhile,
in Australia, Europe, and North America, ‘complementary and
2.8
Traditional medicine exemplified
by traditional Chinese medicine
Fulong Liao, Tingliang Jiang, and Youyou Tu
2.8  Traditional medicine exemplified by traditional Chinese medicine
109
alternative medicine’ is increasingly used in parallel to dominant
practices of modern medicine, particularly for managing chronic
diseases. In these countries, a desire for more personalized healthcare
is likely to explain the increasing take up of these approaches.
Since traditional Chinese medicine is long established and pro-
vides some notably successful therapies, we provide examples that
illustrate how traditional approaches remain so much a feature
of medicine in China; in some instances, the treatment has been
adopted in the parallel world of modern medicine that is dominant
elsewhere.
Products in the global pharmacopoeia but which
originated from traditional Chinese medicine
Ephedrine—​the first drug transferred to the Western
pharmacopoeia from Chinese medicine
Since the Han dynasty (206 BCE-200 CE), the herb Mahuang (Ephedra
sinica) in traditional Chinese medicine has been known to have
antiasthmatic properties. A decoction of ephedra is recorded in the
classical ‘Treatise on Febrile and Miscellaneous Diseases’ by Zhang
Zhongjing (150–​215 CE). Ephedrine was the first drug that origin-
ated from traditional Chinese medicine to appear in the Western
pharmacopoeia.
The first isolation of ephedrine from Ephedra
The renowned Japanese organic chemist and pharmacologist Nagai
Nagayoshi who had studied in Europe, first succeeded in extracting
the active principle, ephedrine, from Ephedra in 1885. He isolated
the alkaloid (ephedrine hydrochloride) from the plant and deter-
mined its formula and chemical structure. At that time, the pharma-
cological effect of ephedrine was best known in the eye in which it
causes pupillary dilatation. The agent was largely forgotten until the
early 1920s when it was rediscovered by K.K. Chen and C.F. Schmidt.
The discovery of ephedrine as a sympathomimetic
Dr Chen’s research at Peking Union Medical College sought to
understand the pharmacology of chemical components obtained
Box 2.8.1  Commonly used therapies and therapeutic techniques in branches of traditional
medicine globally
Traditional medicine includes diverse health practices, approaches, knowledge and beliefs incorporating plant,
animal and/or mineral based medicines, spiritual therapies, manual techniques and exercises, applied singularly or
in combination to maintain well-being, as well as to treat, diagnose or prevent illness.
Commonly used therapies and therapeutic techniques
Chinese
Medicine
Ayurveda Unani Naturopathy Osteopathy Homeopathy Chiropractic
Herbal medicines
Acupuncture/acupressure
Manual therapies
Spiritual therapies
Exercises
= commonly incorporates this therapy/therapeutic technique
= sometimes incorporates this therapy/therapeutic technique
= incorporates therapeutic touch
Reprinted from Traditional Medicine Growing Needs and Potential—​WHO Policy Perspectives on Medicines, No. 002, May 2002.
Copyright © World Health Organization 2002.
110
section 2  Background to medicine
from Chinese herbs and was supported by the Rockefeller founda-
tion; his work on ephedrine became very well known. Since the iso-
lation of ephedrine by Nagai was unknown to him, Chen set out
to develop a method of extraction to obtain the active compound
from Ephedra, which he isolated in only a few weeks. Later, he col-
laborated with Schmidt to determine the pharmacological actions of
ephedrine. The agent was found to increase myocardial contractility,
induce sustained elevation of carotid arterial blood pressure, con-
tract the blood vessels in skin, mucosa, and internal organs, as well as
stimulate central awareness. These actions resemble those of adren-
aline (epinephrine), but with long-​lasting effects and relatively low
toxicity. Chen and Schmidt published their findings in the Journal of
Pharmacology and Experimental Therapeutics in 1924. Soon after-
wards, the Eli Lilly Company marketed ephedrine (in 1926) for the
indications of nasal congestion and bronchial spasm.
With the discovery of ephedrine, traditional Chinese medicine
had thus borne its first commercial fruit in Western pharmacology.
As a sympathomimetic agent, ephedrine is able to cause release of
adrenaline and noradrenaline (epinephrine and norepinephrine)
and induce relaxation of bronchial smooth muscle. Among other ef-
fects, stimulation of ß-​adrenergic receptors by ephedrine appears to
be responsible for inducing relaxation of bronchial smooth muscle.
Ephedrine became a highly popular and effective treatment for
asthma, particularly because, unlike the standard therapy of adren-
aline at that time, conveniently it could be administered orally. As a
treatment for asthma, ephedrine reached its zenith of popularity in
the late 1950s.
Artemisinin—​a gift from traditional Chinese medicine
Compared with the time taken to introduce many of contemporary
discoveries through drug development into widespread clinical use,
the discovery of artemisinin largely followed a relatively simple and
direct logical route. This route was fortuitously directed by findings
that had been long-​reported in the ancient Chinese literature and in
this sense, artemisinin, with its unique sesquiterpene lactone gener-
ated by phytochemical evolution, is truly a gift from the old practi-
tioners of Chinese medicine.
In the event, the modern achievement was accomplished by the
nationwide collaboration of ‘project 523’ (a national antimalarial
project initiated by Chairman Mao Zedong during the war with
the United States in Vietnam). This national cooperative project
included institutions in Beijing, Shanghai, as well as Shandong,
Yunnan, and Guangdong provinces. Each institution contributed
to the successful progress of the project in different ways: these in-
cluded the identification of the agricultural area where local strains
of Artemisia which are rich sources of artemisinin, grew most pro-
lifically; the simplified procedure for extraction of artemisinin
suitable for large-​scale production; determination of the unique
stereo-​structure of artemisinin; development of active artemisinin
derivatives; and, not least, the facility for experimental and clinical
investigation of the actions of artemisinin and its derivatives.
Breakthrough in antimalarial research in the 1970s
Malaria caused by Plasmodium falciparum has been a life-​threatening
human disease for millennia. After a failed international attempt to
eradicate malaria in the 1950s, the disease rebounded, largely due
to the emergence of malarial parasites which were resistant to the
existing antimalarial drugs, such as chloroquine, which had been
introduced in 1947. Resistance to quinine, often the drug of last
resort, had been noted in Brazil in 1907 and had been reported
from the Thai–​Cambodian border in the mid-​1960s. In 1969,
the Institute of Chinese Materia Medica, Academy of Traditional
Chinese Medicine was engaged in the national ‘523’ antimalarial
project, and the senior author of this chapter (YY Tu) was appointed
head of an antimalarial research group. The group collected and
selected more than 2000 Chinese herb preparations and identi-
fied 640 recipes from ancient literatures that might have some
antimalarial activities. More than 380 extracts were evaluated in a
living model of malaria in mice infected with Plasmodium berghei,
but contradictory to what was recorded in the ancient literature,
no promising or reproducible result was obtained. In an effort
to explain this, Tu intensively re-​examined the literature:  in A
Handbook of Prescriptions for Emergencies authored by Ge Hong
(Jin dynasty, 284–​346 CE), the use of Qinghao (the Chinese name
of sweet wormwood, Artemisia annua) for alleviating malaria
symptoms is described: ‘A handful of Qinghao is immersed in 2
litres of water; wring out the juice and drink it all’ (Fig. 2.8.1). Tu
was struck by the possibility that the heating involved in a con-
ventional extraction step might have destroyed the active compo-
nents and that extraction at a low temperature might be necessary.
First extracting by ether reflux, YY Tu and colleagues subse-
quently resolved the extracts into the acidic and neutral fractions
and finally obtained a non​toxic neutral extract (No. 191). On 4
October 1971, extract 191 was found to be completely effective
in suppressing parasitaemia in the P. berghei-​infected mice and
later, in monkeys with malaria due to Plasmodium cynomolgi. To
accelerate the clinical evaluation of extract 191, Tu and colleagues
voluntarily tested the safety of the extract in themselves. Shortly
after this, in October 1972, a trial with 30 patients mainly from
southern China infected with P. vivax and P. falciparum showed a
rapid disappearance of fever and intraerythrocytic parasites. One
month later, these results were formally notified to the national
‘523 project’ office, after which further collaborative research on
Artemisia annua was promoted. The clear demonstration of anti-
malarial efficacy in extract 191, stimulated by the ancient and ac-
curate account many centuries earlier by Ge Hong, symbolizes the
breakthrough which led to the later discovery and pharmaceutical
development of artemisinin.
The discovery of artemisinin
YY Tu’s research group set out to purify the components with anti-
malarial activity that were present in extract 191 of Artemisia annua.
In 1972, a colourless crystalline substance was identified with a mo-
lecular weight of 282 Da, a molecular formula of C15H22O5 absent ni-
trogen, and a melting point of 156–​157°C. The name of ‘Qinghaosu’
(artemisinin, ‘su’ means basic element in Chinese) was given to this
newly identified compound. Afterwards, the team found out that
only the species of A. annua in the Artemisia genus and its leaves
in the alabastrum stage contain abundant artemisinin. The chem-
ical properties and behaviour of the compound were identified as a
sesquiterpene lactone; finally in 1975, the stereo-​structure of arte-
misinin was determined unambiguously in collaboration with col-
leagues at the Institute of Biophysics, Chinese Academy of Sciences.
The structure was first published in 1977 and the new molecule and
the paper were immediately cited by the Chemical Abstracts (C.A.
1977, 87, 98788g).
2.8  Traditional medicine exemplified by traditional Chinese medicine
111
Worldwide attention to artemisinin
The successful treatment of several thousand patients with artemi-
sinin and its derivatives for malaria in China attracted worldwide
attention in the 1980s. In 2002, the WHO announced a switch in the
recommendations for malaria treatment to artemisinin combin-
ation therapy. In 2006, to prevent malaria parasites from developing
resistance to the drug, WHO requested pharmaceutical com-
panies to end the marketing and sale of ‘single-​drug’ artemisinin,
Artemisinin combination treatment is currently widely used, and
has saved many lives globally, and especially in Africa. The remedy
impressively reduces the intensity of malaria in Africa due to its
activity against the gametocyte stages of P. falciparum in the blood
which allow transmission from host to host via the mosquito vector.
Murray et al. have examined the global mortality due to malaria
over the period 1980–​2010 (see Fig. 2.8.2): since the global peak in
2004, there has been a substantial decrease in malaria deaths that is
H
H
H
D
C
B
A
6
7
8
9
10
3
4
2
1
1
5
3
2
0
0
5
0
0
0
4
12
13
11
H
CH3
H3C
15
CH3
14
Fig. 2.8.1  From ancient literature to artemisinin molecule. Life: A Handbook of Prescriptions for Emergencies by Ge Hong (Ming dynasty version,
1574 ce); Middle: ‘A handful of Qinghao immersed with 2 litres of water, wring out the juice, and drink it all’ is printed in volume 3 in the fifth
column from the right. Right: the molecular structure of artemisinin.
Image courtesy of Prof Liao FL.
2000000
1800000
1600000
1400000
1200000
1000000
800000
600000
400000
200000
0
1980
1985
Deaths (n)
1990
1995
Year
2000
2005
2010
≥70 years
50–69 years
15–49 years
5–14 years
<5 years
Fig. 2.8.2  Malaria deaths by age from 1980 to 2010.
Reprinted from The Lancet, 379(9814), Murray CJ, et al., Global malaria mortality between 1980 and 2010: a systematic
analysis, 413–​31. Copyright © 2012, with permission from Elsevier.
112
section 2  Background to medicine
attributable to the rapid scale-​up of control activities, including the
intensified use of artemisinin combination therapy.
Arsenic trioxide and the treatment of acute
promyelocytic leukaemia
Brief medicinal history of arsenic
Arsenic is one of the oldest drugs in Western medicine and trad-
itional Chinese medicine, since it was cited by Hippocrates (460–​
370 BCE) as a treatment of skin ulcer and by the Yellow Emperor’s
Internal Classic (263 BCE) for treatment of periodic fever (likely to
be attributable to malaria). Arsenic exists in different forms: As4S4,
known as realgar; and As2S3, known as orpiment or yellow arsenic;
the common term for As2O3 (arsenic trioxide) is white arsenic. The
highly poisonous nature of arsenic has long been known and ac-
cording to the traditional principle of ‘combating an evil with a toxic
agent’ limited amounts of arsenic have been included in traditional
medical recipes; it was also recorded by the 16th century Li Shi-​Zhen
in the Compendium of Meteria Medica. In the late 18th and early
19th centuries, Fowler’s solution, 1% potassium arsenite (KAsO2)
solution, was introduced in clinics in Europe to treat periodic fever,
chronic myelogenous leukaemia, Hodgkin’s disease, and many other
conditions. Some practitioners in Europe used arsenicals as a ‘tonic’
to improve energy in those who complained of fatigue or reduced
libido. However, due to its toxicity, including the development of
squamous cancers of the skin, and the advent of modern chemo-
therapy, the use of Fowler’s solution and other arsenicals was largely
discarded in the early decades of the 20th century.
The application of arsenic to acute promyelocytic leukaemia
Acute promyelocytic leukaemia is one of the most lethal forms of
leukaemia. In the 1970s, the combination of anthracyclines and
cytosine arabinoside was adopted as the mainstream treatment for
acute promyelocytic leukaemia, but the effects were often not cura-
tive and adverse events with bleeding related to disseminated intra-
vascular coagulation—​often initiated by chemotherapy. Enlightened
by the ancient eastern wisdom, Chinese doctors never stopped their
attempts to blaze a new trail. In the 1980s, inspired by the Chinese
philosophy that ‘the evil can be educated to be the good’ and the
concept of cancer cell differentiation from the Western literature, Dr
Zhenyi Wang’s team of Shanghai Institute of Haematology first ap-
plied all-​trans retinoic acid to the treatment of acute promyelocytic
leukaemia and achieved a complete remission rate of 90%. More im-
portantly, this provides a first example that a human cancer can be
treated effectively using inducers of cell differentiation rather than
cytotoxic agents. The application of arsenic to the treatment of leu-
kaemia was another breakthrough initiated by Chinese clinical in-
vestigators and physicians.
In 1971, Dr Tingdong Zhang and colleagues affiliated to the First
Hospital at Harbin Medical University, chanced upon the potential
anticancer effect of arsenic trioxide from a countryside herbalist.
Accordingly, Zhang and colleagues prepared a solution for par-
enteral administration termed ‘Ailin’ (broadly translated from the
Chinese as a cancer-​curing mixture): this contained arsenic trioxide
and a trace amount of mercury chloride and was administered to
patients with chronic myeloid leukaemia. The first report on the ef-
fects of Ailin solution was published in 1973 in a local journal en-
titled Medicine and Pharmacy of Heilongjiang. In 1976, the authors
reported five patients with acute leukaemia who had had a complete
remission after administration of Ailin. Later, in 1979, Zhang pub-
lished a summary of 55 cases of acute leukaemia treated by Ailin
solution or related remedies in which arsenic trioxide had been
combined with other agents. All 55 cases were improved to some ex-
tent, with a remission rate of 70%; this included 12 patients in whom
complete remission occurred. Of note, few or only slight toxic side
effects occurred with the small doses of arsenic used. Zhang and col-
leagues clearly pointed out that the key component in the Ailin so-
lution was arsenic. From 1984 to 1992, they further observed that
acute promyelocytic leukaemia was particularly susceptible to ar-
senic treatment.
In the mid of 1990s, a research group at the Shanghai Institute of
Haematology, led by Dr Zhu Chen, established a cooperation with
Dr Tingdong Zhang. They then launched the first systematic study
of the cellular and molecular mechanisms of arsenic trioxide and
confirmed its antileukaemia effects. Grounded in their laboratory
findings, Dr Chen’s team conducted a controlled clinical trial of pure
arsenic trioxide in relapsed and newly diagnosed patients with acute
promyelocytic leukaemia in which high (80%) rates of complete re-
mission were observed in both settings. Based on the convincing re-
sults of clinical and basic studies by Dr Chen’s group, in August 1999,
arsenic trioxide was approved as a new drug by the China Food and
Drug Administration.
Mechanism of action of arsenic in acute
promyelocytic leukaemia
Dr Zhu Chen, Dr Saijuan Chen, and colleagues of Shanghai Institute
of Haematology investigated the mode of action of arsenic trioxide.
A  combination of apoptosis induction and partial differentiation
appears to be the cellular mechanisms associated with induction
of remission. In molecular terms, the presence of the transloca-
tion, t(15;17), causing fusion of the retinoic acid receptor α gene
on chromosome 17 with the promyelocytic leukaemia (PML) gene
coding for a nuclear protein on chromosome 15, leads to formation
of the PML-​RARα fusion protein containing sequences from the
PML moiety and retinoic acid receptor α. It is notable most patients
with acute promyelocytic leukaemia who harbour PML-​RARα re-
spond well to the therapeutic effect of arsenic trioxide, while there
is evidence that those very rare patients without the t(15;17) trans-
location and therefore without expression of the PML-​RARα fusion
protein, such as those with the t(11;17) translocation and resultant
PLZF-​RARα, are resistant or relatively resistant to arsenic. It has been
shown that arsenic trioxide binds directly to cysteine residues located
within the specific RING-​B box-​coiled coil domain of PML-​RARα.
The identification of the PML-​RARα fusion protein as a direct target
of arsenic trioxide and at in situ concentrations that are considerably
lower than those required to induce apoptosis in other cancer cells,
indicates a remarkable selectivity of the molecular target. It is worth
noting that the wild-​type PML protein is a tumour suppressor in-
volved in growth arrest and regulation of apoptosis and senescence.
In healthy cells, PML is located in a specific nuclear subdomain (the
PML nuclear body). However, in promyelocytic leukaemia, complex
formation of PML-​RARα with wild-​type PML breaks up the nuclear
microspeckle pattern and the functions of the protein. Arsenic tri-
oxide at therapeutic dosage corresponding to in situ concentrations
as low as 10-​7 M induces degradation of both the wild-​type PML
2.8  Traditional medicine exemplified by traditional Chinese medicine
113
and PML-​RARα fusion proteins with posttranslational modifica-
tion by sumoylation, which appears to restore accumulation of the
wild-​type protein in the formerly leukaemic cells. The sumoylation
of the fusion protein is then followed by its degradation through the
proteasome-​dependent pathway (Fig. 2.8.3). At the level of cells and
the whole organism, arsenic trioxide causes partial differentiation of
acute promyelocytic leukaemia cells as well as their apoptosis; not-
ably it has little effect on other cancer cells under similar conditions.
A  striking additional finding is that only the disease-​associated
PML-​RARα, but not the wild-​type RARα receptor is degraded upon
exposure to the drug.
Given the striking effects of arsenic trioxide in a leukaemia that
is otherwise very challenging to cure, clinicians worldwide have
rapidly adopted the original findings. In the first instance, the agent
found favour with clinicians who explored use of parenteral ar-
senic in patients who had proved resistant to or had relapsed after
receiving recommended first-​line drugs including the targeted mo-
lecular therapy of all-​trans retinoic acid which, by also targeting
the PML-​RARα oncoprotein, can lead to terminal differentiation
of malignant promyelocytes to mature neutrophils. However, it
is apparent that this treatment alone does not eradicate the ma-
lignant clone. Cytotoxic chemotherapy may lead to remission but
through the mechanistic studies of the group of Shanghai Institute of
Haematology it is now recognized that achievement of a cure in the
great majority of acute promyelocytic leukaemia patients requires
the addition of arsenic trioxide.
The treatment of acute promyelocytic leukaemia is complex, and
involves management and prevention of several potentially life-​
threatening complications such as the promyelocytic leukaemia cell
differentiation syndrome, occurring in a small subset of patients
within the first three weeks of treatment with all-​trans retinoic acid
or arsenic trioxide. However, the past three decades have witnessed
a great progress in the treatment of the disease largely as a result of
original and modern studies related to the traditional use of medi-
cinal arsenic in China.
Based on the observations that both arsenic trioxide and all-​trans
retinoic acid bind and trigger proteasomal degradation of PML/​
RARα fusion protein but through distinct mechanisms, a syner-
gistic targeting strategy has been developed by Dr Zhu Chen and
Dr Saijuan Chen’s group for newly diagnosed patients with acute
promyelocytic leukaemia. This involves a combination of all-​trans
retinoic acid and arsenic trioxide, and incorporates chemotherapy
in the postremission treatment. This regimen not only reduces the
early deaths caused by bleeding or differentiation syndrome, but
more importantly yields up to 90% five-​year disease-​free survival in
acute promyelocytic leukaemia. Recent reports from the same group
have confirmed the safety of the all-​trans retinoic acid/​arsenic tri-
oxide combination therapy, since no obvious long-​term toxicities
NB4 cells
0 hr
0.5 hr
6 hr
24 hr
PML
(a)
(b)
(d)
(c)
PML-RARα
RARα
0
0
10 20 30 60
0 10 20 30 60
0
10 20 30 60
As2 O3(hr)
As2 O3(min)
121
77
181
121
77
181
121
77
181
121
77
kDa
kDa
kDa
48
77
48
77
Anti-FLAG
Anti-FLAG
Anti-FLAG
Mock
Arsenic (ng/mg)
0
5
10
15
20
25
PML
P
S
PML-RARα >
RARα >
1
6
12
24
<
<
<
<
<
Fig. 2.8.3  The effect of arsenic trioxide (ATO) on the posttranslational modification and degradation of PML-​RARα and PML is rapid and
specific. (a and b) ATO-​induced degradation of PML-​RARα in the APL cell line NB4 cells, as assessed by Western blotting with RARα antibody and
immunofluorescence staining with PML antibody (green). The cells were treated with 1 mM ATO. Scale bar, 10 mm. (c) A time-​course study showing
the effects of ATO on PML, PML-​RARα, or RARα. (<) points to parental proteins, (▼) to modified proteins, and (▽​) to degraded fragments. Transfected
HEK 293T cells were treated with 2 mM ATO and then lysed in radioimmunoprecipitation assay (RIPA) buffer and fractionated into supernatants
(S) and pellets (P) by centrifugation. (d) The arsenic content of pellets from HEK 293T cells that had been mock-​transfected or transfected with a
vector encoding PML.
Reproduced from Chen Z, Chen SJ, et al. Arsenic trioxide controls the fate of the PML-​RARalpha oncoprotein by directly binding PML. Science, 2010, 328:240–​3.
114
section 2  Background to medicine
were observed among patients in whom treatment is stopped. Based
on the risk factors for prognosis in acute promyelocytic leukaemia,
optimization of the treatment emphasizes the role of arsenic in in-
duction, consolidation and maintenance therapy as a substitute
to chemotherapy in low-​ and intermediate-​risk patients, and in
potential reduction of chemotherapy in the patients with a high-​
risk of complications without adverse impact on the outcome. See
Chapter 22.3.3 for further discussion.
Evolution of pharmaceutical use and commercial
development of medicinal arsenic
The all-​trans retinoic acid/​arsenic trioxide/​chemotherapy triad
therapy achieves a cure rate of more than 90% and costs less than
CNY 100 000 in Mainland, China. Since the new round of health
reforms launched in 2009, both all-​trans retinoic acid and ar-
senic trioxide have been listed in the China’s National Essential
Drug List; moreover, the costs of treatment are covered univer-
sally by the health insurance. An oral mixture called Realgar-​
Natural Indigo Formula containing realgar, natural indigo, Salvia
miltiorrhiza, and Radix pseudostellariae, which was developed by
Dr Shilin Huang of the 210 Military Hospital in the 1980s, can
achieve a 5-​year survival rate of 85% by targeting PML-​RARα and
orchestrating related signal networks. This formula is also effective
and affordable.
In the recent years, researchers of Hong Kong University, led
by Dr Wing-​Yan Au and Dr Yok-​Lam Kwong, have promoted use
of a convenient oral formulation to replace intravenous arsenic in
Hong Kong and to avoid exorbitant charges: 150 patients with acute
promyelocytic leukaemia have been treated with this alternative,
convenient, and affordable first-​line therapy.
Introduction to compound formulae and disease
prevention in traditional Chinese medicine
The role of compound formulae
The finding of ephedrine as a sympathomimetic, the discovery of
the antimalarial, artemisinin and the exploration of arsenic trioxide
for treating acute promyelocytic leukaemia have strictly followed
the modern pharmaceutical approach. All the three discoveries
were based on a single herbal formula and the application of con-
temporary molecular science in translational research to develop
modern and effective medicines for important conditions.
However, the investigation and use of a single herbal principle
for the treatment of a specific disease is the rare case in the use of
traditional Chinese medicine today. The reality is that popular treat-
ments are characterized by the holistic syndrome differentiation for
a patient; this leads to the prescription of a herbal formula composed
of a group of well-​tailored herbs that are related by the practitioner
to the syndrome for which treatment is requested. The relationship
between individual syndromes and the herbal formulae used, while
largely symptom-​based and antithetic to the teaching of Western
medicine, are established as part of the rich experience in the ad-
vancement of traditional Chinese medicine that has evolved over
thousands of years. Currently, marketed products of traditional
Chinese medicines are mostly patented compound formulae; these
account for more than 20% of the whole pharmaceutical market in
China. In the 2015 edition of the Chinese pharmacopoeia, 1493 for-
mulae and single herb preparations are recorded (of which formulae
constitute the majority); 618 medicinal materials and processed
herbal slices are the exception.
Herbal formulae for treatment
of cardiovascular diseases
Cardiovascular diseases are common life-​threatening conditions
and naturally come to the frequent attention of practitioners and
doctors. The term of ‘blood’ in Chinese medicine has a meaning that
corresponds closely with that in Western medicine (i.e. the flowing
or circulating blood). The syndromes associated with stasis of blood
also occur frequently as part of syndrome differentiation, and are
linked closely with vascular diseases. The principle of activating
or improving the circulation of blood is a therapeutic concern in
many aspects of medicine and is an important aspect of traditional
Chinese medicine.
Since the 1970s, Keji Chen and colleagues have conducted nu-
merous clinical trials by applying herbal medicine in efforts to stimu-
late the circulation of blood as part of the treatment of coronary heart
disease, including anginal attacks and restenosis after percutaneous
coronary intervention. In collaboration with physicians in working
in 16 hospitals in Beijing, Chen and this clinical research team in-
vestigated the use of a herbal compound formula simply termed
‘coronary heart II’ to treat chronic stable anginal pectoris. The for-
mula consists of five herbs with properties considered to stimulate
or improve the circulation of blood, including Salvia miltiorrhiza
and Ligusticum wallichii. The randomized double–​blind controlled
clinical trial showed that the compound formula could reduce the
frequency of anginal attacks; at the same time laboratory studies
indicated that the effect might in part be attributed to antiplatelet
activity with increased fibrinolysis. The trial supported clinical use
of the formula preparation in the treatment of chronic stable an-
gina pectoris compared with the placebo. Tetramethylpyrazine,
an antiplatelet drug used in China, was detected in extracts from
Ligusticum wallichii contained in the aforementioned formula
preparation.
In the past 30 years, active treatment for acute myocardial in-
farction and coronary syndromes has saved many lives. However,
coronary artery restenosis after percutaneous coronary interven-
tion continues to present a dilemma. To investigate frequently
encountered complications such as plaque rupture and the con-
tribution of inflammatory factors to obstructive complications of
atheromatous coronary heart disease, Chen and colleagues took
a holistic view of the ischaemic complications of blood stasis in
the coronary circulation and selected XueFu ZhuYu Tang, a well-​
known Chinese compound formula developed by Wang Qingren
(1768–​1831 CE) for various syndromes attributed to blood stasis.
Pursuing evidence from pharmacological studies, they used
Ligusticum phenols and Paeonia glycosides extracted from the for-
mula in a preparation for a multicentre clinical trial carried out in
335 patients with potential coronary arterial restenosis after per-
cutaneous coronary intervention. Based on a routine treatment
of aspirin and ticlopidine, XS0601 capsule (extracted Ligusticum
phenols and Paeonia glycosides) showed significant benefit for
preventing restenosis compared with placebo. The restenosis rate
2.8  Traditional medicine exemplified by traditional Chinese medicine
115
was significantly reduced in the XS0601 group as compared with
that of the placebo (26.0% compared with 47.2%, p <0.05; see
Fig. 2.8.4). At up to 1 year of follow-​up, the Kaplan–​Meier survival
curve for freedom from clinical end-​point events showed a sig-
nificant difference between XS0601 and the placebo; in addition,
follow-​up angiographic findings support the salutary findings in
relation to survival outcome.
To request approval for compound formulae from American
Food and Drug Administration (FDA) is a new effort of narrowing
the gap between traditional Chinese medicine and mainstream
Western medical practice. The patent medicine, Compound
Danshen Dripping Pill in the Chinese pharmacopoeia, is a typ-
ical example. The pill, containing extracts from Salvia miltiorrhiza
and Panax notoginseng, is mainly used to treat angina and cor-
onary heart diseases. So far it is the first Chinese patent traditional
medicine to have completed the FDA’s clinical phase II trials and
it is encouraging that this medication remains under evaluation
in rigorous clinical phase III trials. Also, the drug in tablet form
has been approved by drug watchdogs in Canada, Russia, and
some Asian/​African countries. The manufacturer’s data shows
that more than millions of people worldwide take this medication
annually.
Disease prevention, health promotion, and traditional
Chinese medicine
Disease prevention and health promotion are strategic issues in
medicine. ‘Preventing disease is far superior to treating a dis-
order’ is also one of the sacred principles underlying traditional
Chinese medicine. A drawing of physical and breathing exer-
cises on silk was unearthed from a Han Tomb at Mawangdui,
central China in the 1970s. The finding provided early archaeo-
logical evidence of exercise employed in healthcare in China
some 2000 years ago (Fig. 2.8.5). For disease intervention and
healthcare, biomechano-​pharmacology, a new borderline dis-
cipline, pursues the combinatorial effects of traditional medicine
with biomechanical factors that are designed to improve blood
flow. Many studies have revealed that endothelial release of ni-
tric oxide, endothelin-​1, prostacyclin (PGI2), von Willebrand
factor, tissue plasminogen activator, intercellular cell adhesion
molecule-​1, vascular cell adhesion molecule-​1, endothelial
cell growth factor, and inflammatory factors are all regulated
by blood shear stress. Therefore, vessel tension, thrombosis,
thrombolysis, cell adhesion, angiogenesis, atherosclerosis can
be argued on first principles to be influenced by blood shear
stress and hence circulatory flow. The argument in Chinese trad-
itional medicine is made that cardiovascular and cerebrovas-
cular diseases may benefit from interventions that influence the
biological responses of endothelium regulated by biomechanical
factors. In most cases, sedentary life style reduces blood flow
and cardiac output and with this the level of blood shear stress.
As a modern interpretation of blood stasis syndrome, decreased
blood shear stress is a common pattern of the syndrome. The an-
cestors of modern Chinese utilized exercise to prevent this syn-
drome and promote health. The great surgeon, Hua Tuo, before
205 CE, not only created five animal-​mimic boxing as a phys-
ical fitness exercise, but also realized ‘Diseases are prevented as
blood flow is promoted’. This appears today to be a really wise
and prescient foresight of the biological effects of blood flow.
Pharmacological experiments have recently shown that the joint
application of exercise combined with administration of ex-
tracts from Chinese medicine may prevent atherosclerosis. For
example, the experimental combination of a Yindan Xinnaotong
capsule and swimming in experimental rats may prevent athero-
sclerosis through a synergistic effect between the capsule and
50
40
26.1
Per patient
restenosis
Per lesion
restenosis
New lesion
Post-stenting
Pre-stenting
Follow-up
Stent-dilating
Restenosis rate (%)
47.2
23.6
39.2
12.3 13.9
Control
XS0601
30
20
10
0
Fig. 2.8.4  Restenosis comparison between XS0601 and the controls. Left: Restenosis rates per patient and per lesion and percent of new lesion.
XS0601 significantly lowered both per patient and per lesion restenosis as compared with the placebo group (p <0.05). As for new lesions, there
was no significant difference between the two groups. Right: Angiographic evidences for a patient in the XS0601 group in the stages of prestenting,
stent-​dilating, poststenting, and follow-​up.
Image courtesy of Prof Chen KJ.
116
section 2  Background to medicine
swimming in improving blood circulation, rheological param-
eters in the blood, concentration of proatherogenic lipopro-
teins, and the vascular endothelium. Vascular remodelling may
contribute to the prevention of atherosclerosis by upregulating
smooth muscle protein responses—​certainly, this whole field
continues to be an active field of contemporary research in
Chinese medicine.
Making traditional medicine available
and affordable
Large swathes of the populations in the world’s poorest coun-
tries are those most in need of inexpensive, effective treatments.
The WHO estimates that one-​third of the global population still
lacks regular access to essential drugs, and that in the poorest
parts of Africa and Asia, this figure rises. In these regions, some
form of traditional medicine is often a more widely available and
more affordable source of healthcare. Clearly to provide a sound
basis for efforts to promote traditional medicine and to prevent
illness and wasted resources, safe and effective therapies must be
identified, The focus should be on safe and effective treatments
for diseases which represent the greatest burden for poor popu-
lations, in this context for malaria and HIV/​AIDS are of great
importance but the value of a safe supply of potable water and
simple hygiene also must go hand in hand with treatment for all
infectious diseases.
Many Member States of WHO have made great efforts to advance
traditional medicine, some of which can be attributed to the imple-
mentation of the WHO Traditional Medicine Strategy 2002–​2005.
In recent years, WHO has collaborated with many countries and
areas to develop these programmes and to bring traditional medi-
cine into the mainstream healthcare system. WHO has developed
technical guidelines and standards and organized trainings/​work-
shops in support of Member States. Fig. 2.8.6 shows the progress of
Member States regarding established national policies on traditional
medicine and national regulations on herbal medicines.
The newly established WHO Traditional Medicine Strategy
2014–​2023 has two key goals:  to support Member States in
harnessing the potential contribution to health, wellness, and
people-​centred healthcare and to promote the safe and effective
use of traditional medicine through the proper and organized
regulation of products, practices, and practitioners. We very much
hope that the implementation of this strategy will accelerate the
availability and affordability of traditional medicine in the near
future. As stated by Mao Zedong, Chinese medicine and pharma-
cology are a great treasure-​house and we should explore them and
raise them to a higher level.
Fig. 2.8.5  Drawing of physical and breathing exercise on silk unearthed from a Han Tomb
at Mawangdui, central China. It is an early archaeological evidence of exercise employed in
healthcare in China some two thousand years ago.
2.8  Traditional medicine exemplified by traditional Chinese medicine
117
So much of benefit for humankind will emerge if the science of the
West and the rich experience and traditions of medicine in China
can work together in an integrated way.
Acknowledgements
The authors wish to record their gratitude to Professors Chen
(Saijuan) and Chen (Zhu) for kindly reviewing this chapter.
FURTHER READING
Au WY, et al. (2011). Oral arsenic trioxide-​based maintenance regi-
mens for first complete remission of acute promyelocytic leukemia:
a 10-​year follow-​up study. Blood, 118(25), 6535–​43.
Chen GQ, et al. (1997). Use of arsenic trioxide (As2O3) in the treat-
ment of acute promyelocytic leukemia (APL): I. As2O3 exerts dose-​
dependent dual effects on APL cells. Blood, 89(9), 3345–​53.
Chen KJ, et al. (2006). XS0601 reduces the incidence of restenosis:
a prospective study of 335 patients undergoing percutaneous cor-
onary intervention in China. Chin Med J, 119(1), 6–​13.
Chen KK, Schmidt CF (1924). The action of ephedrine, the active
principle of the Chinese drug Ma Huang. J Pharmacol Exp Ther, 24,
339–​57.
Hu J, et al. (2009). Long-​term efficacy and safety of all-​trans ret-
inoic acid/​arsenic trioxide-​based therapy in newly diagnosed
acute promyelocytic leukemia. Proc Natl Acad Sci U S A, 106(9),
3342–​7.
Huang ME, et al. (1988). Use of all-​trans retinoic acid in the treatment
of acute promyelocytic leukemia. Blood, 72(2), 567–​72.
Klayman DL (1985). Qinghaosu (artemisinin): an antimalarial drug
from China. Science, 228, 1049–​55.
Li GQ, et al. (1984). Randomised comparative study of mefloquine,
qinghaosu, and pyrimethamine-​sulfadoxine in patients with falcip-
arum malaria. Lancet, 2(8416), 1360–​1.
Liao FL, et al. (2006). Biomechanopharmacology: a new borderline
discipline. Trends Pharmacol Sci, 27, 287–​9.
Lo-​Coco F, et al. (2013). Retinoic acid and arsenic trioxide for acute
promyelocytic leukemia. N Engl J Med, 369(2), 111–​21.
Luo J, et al. (2015). Compound danshen (salvia miltiorrhiza) dripping
pill for coronary heart disease: an overview of systematic reviews.
Am J Chin Med, 43(1), 25–​43.
Murray CJ, et al. (2012). Global malaria mortality between 1980 and
2010: a systematic analysis, Lancet, 379(9814), 413–​31.
Shen ZX, et al. (1997). Use of arsenic trioxide (As2O3) in the treat-
ment of acute promyelocytic leukemia (APL): II. Clinical efficacy
and pharmacokinetics in relapsed patients. Blood, 89(9), 3354–​60.
Sun HD, et al. (1992). Ai-​Lin I treated 32 cases of acute promyelocytic leu-
kemia. Chin J Integrat of Chinese and Western Medicine, 12, 170–​1.
Tu YY (2011). The discovery of artemisinin (qinghaosu) and gifts from
Chinese medicine. Nat Med, 17(10), 1217–​20.
Tu YY (2015). Discovery of Artemisinin—​a gift from traditional
Chinese medicine to the world. Karolinska Institutet, Nobel Lecture,
The Nobel Foundation, Stockholm.
Wang JG, et al. (2018). What has traditional Chinese medicine de-
livered for modern medicine?, Expert Rev Mol Med, 20, e4.
WHO (2013). WHO Traditional Medicine Strategy 2014–​23. https://​
www.who.int/​medicines/​publications/​traditional/​trm_​strategy14_​
23/​en/​
Zhang TD, Rong FX (1979). Treatment of acute granulocytic leu-
kemia by Ailin No.1 and dialectic theory. Medicine and Pharmacy of
Heilongjiang, (4), 7–​11.
Zhang XW, et  al. (2010). Arsenic trioxide controls the fate of the
PML-​RARalpha oncoprotein by directly binding PML. Science,
328(5975), 240–​3.
120
100
80
60
40
20
0
1999a
25
65
39
82
45
92
48
69
110
119
2003b
2005c
2007d
2012e,f
Number of MS with TM policy
Number of MS regulating herbal medicines
Fig. 2.8.6  The national policies on traditional medicine and national regulations on herbal medicines in WHO
member states.
Reprinted from WHO Traditional Medicine Strategy 2014–​2023, page 21. Copyright © World Health Organization 2013.

2.9 Engaging patients in therapeutic development 1
2.9 Engaging patients in therapeutic development 118
ESSENTIALS
Engagement of patients during the development of new treatments
has evolved rapidly in the last decade. No longer should patients be
viewed as passive subjects of research: they should be active partici-
pants in a more open and ambitious ecosystem of academic investi-
gators, companies, and government entities. Enhanced engagement
of patients throughout the development process provides important
benefits for the introduction of useful and effective drugs by chan-
ging the understanding of the disease, the measurement of the dis-
ease during treatment, and the interpretation of benefit-​risk. This
system is critical for the discovery, study, development, regulation,
delivery, and use of innovative therapies.
Introduction
Increasing involvement of patients and their advocacy organizations
in therapeutic development provides an opportunity to innovate
experimental medicine and render challenging or invasive clinical
trials possible. The recent excitement in the field of ‘rare diseases’
has a long history—​but since the accelerated introduction of novel
treatments as a result of the introduction of the Orphan Drug Act
(first in the United States, in 1983, later followed in Europe in 2001),
participation of patients in the whole process of therapeutic devel-
opment has burgeoned at many levels. There can be little doubt that
this trend of practice substantially improves the development of
drugs and promotes a healthy social contract between all parties that
can accelerate achievement of one common goal: introduction of ef-
fective and safe treatments. For academic investigators, clinicians,
companies, and government entities, the goal is common; but the
processes undertaken by, and investment of each party is strikingly
different.
As to the involvement of patient organizations, this may include
funding creative breakthrough research at a university that might
otherwise not be able to provide adequate support for the neces-
sary preclinical or clinical studies. Patient groups have contributed
materially to the development of new ways to measure or score the
manifestations of a given disease, as well as its response to potential
treatments—​not least by helping to recruit patients for exploratory
clinical research. More recently, patient groups have also partici-
pated in the collection of data on the benefits and risks for treatments
relative to their disease, as perceived by the individual experien-
cing that condition (the ‘patient perspective’). Finally, many patient
groups assist regulatory authorities, such as the European Medicinal
Agency (EMA), directly and extensively. This involvement relates to
policy changes that are needed to allow early access to therapy or
other policies that have significance for patients. Input from patient
organizations has broadened the perspective of the regulatory agen-
cies as they evaluate new treatments for approval: this involvement
is relatively new but provides insights beyond those offered by pro-
fessionals in academia and industry—​moreover it helps to maintain
the focus of the approval process on those most critically affected by
a given condition.
Enhanced engagement of patients throughout the development
process provides important benefits for the development of useful
and effective drugs by changing the understanding of the disease;
the measurement of the disease during treatment, and the interpret-
ation of benefit-​risk. The understanding of disease is deepened as a
result of detailed disease surveys or studies into the natural course
of the condition in the absence of specific treatment, the course of
the condition treated or in the context of standard care; here again
such work is often supported or organized by patient groups. By
these means, those who develop drugs, as well as investigative phys-
icians gain a sharper and more quantitative understanding of the
true burden of illness in a given disease. It is surprising how often
these more intensive and systematic studies cast aside historical in-
accuracies and mistaken assumptions created decades before as a
result of received clinical wisdom. In the era before treatments were
introduced, much knowledge was based on available authorities
who had, in reality either seen only one subset of a disease and were
unaware of its true diversity: lacking the ability to advance clinical
care, little experience of the condition gained through long-​term
clinical monitoring and follow-​up.
To study how a treatment affects a disease, studying the disease
at a deep and quantitative level is critical for determining how ex-
perimental therapies can have an impact on the clinical course.
Such detailed knowledge allows investigators to devise more ef-
fective instruments of measurement, and at times radically different
2.9
Engaging patients in
therapeutic development
Emil Kakkis and Max Bronstein
2.9  Engaging patients in therapeutic development
119
evaluations and determinants of the progression of a disease, by
which treatments can be assessed authentically. Many patients
now participate in ‘no-​drug’ studies:  here, testing methodologies
are studied carefully to develop the best measures by which cap-
ture the particularities of a given disease. Lately, one of the more
compelling effects of patient engagement has been revealed in the
assessment of benefit and risk of treatments. For some devastating
disorders, the patient and their family’s view of relative risk and
benefit can be starkly different from that of physicians or regulators.
The quantitative collective of the risks for a disease relative to the
additional effects of treatment, allows investigators and regulators to
scrutinize the merits of a treatment while accounting for the risks in
a devastating disease at the same time.
Orphan medicinal products
In recent years, the development of therapies for rare or ultra-​rare
diseases, often referred to as ‘orphan drugs’ or medicines is receiving
increasing attention. This is a welcome sea change for patients and
their respective advocacy organizations, although it is estimated of
the 7000 rare diseases, only 5% have marketed therapies available
to patients. There are a wide variety of challenges and opportunities
that are unique to rare disease therapy development, but economics
may be the most pronounced stumbling block. That is, financing the
testing and successful development for a small patient population
may be difficult to statistically demonstrate (as a consequence of
small sample size) and cost prohibitive for investors and others who
seek to earn a return on investment. Nonetheless, breakthroughs in
genetic testing, gene therapy, enzyme replacement therapy have al-
ready demonstrated substantial clinical benefit and potential for pa-
tients who previously had little hope of access to effective medicines.
While the science around rare diseases has progressed rapidly,
in some cases, it has outpaced the ability of regulatory agencies to
effectively monitor these developments and account for them in ap-
proval decisions. In general, both the Food and Drug Administra­
tion (FDA) and EMA are largely on par for approval of orphan
therapies (see Fig. 2.9.1), but the EMA is seen as often providing
additional flexibility on the types of data packages submitted to sup-
port approval and marketing of an innovative therapy. For example,
several orphan therapies have been approved earlier for marketing
in Europe, but have been declined or delayed in the United States
usually because of issues surrounding the phase 3 studies. In com-
plex settings, the EMA has relied on their scientific and medical in-
sights into the total package of data to allow a patient population
with high unmet medical need access to the drug whereas in the
United States, a missed primary endpoint in a phase 3 study has gen-
erally not been approved.
The contemporary clinical development process
While in the minds of many scientists and most employees in com-
panies, the clinical trial is the principal focus of their endeavour, the
truth remains that clinical trials are indeed another form of experi-
ment, with effectiveness and safety still incompletely understood
in broader human populations. All the clinical phases of trial pro-
grammes are part of a sequence that is in effect an uncertain journey
to the ultimate set of goals involving regulatory approval of a safe
and effective therapy; and one that is both useful and reimbursed.
Before this stage is even contemplated, even the road to initiating
an early clinical trial can be long, arduous, and very costly. Identifying
a suitable drug target can take years of research and development,
with a notably high failure rate. Most early research ideas never
make it to clinical trials, and those that do, still have a small possi-
bility of reaching an approval. The regulatory process is designed to
test safety and efficacy into various stages, and over time these tests
do help provide some basic level confidence that a drug is safe and
50
45
40
35
30
25
20
15
10
5
0
2010
2011
2012
2013
Year
Approval of new molecular entities (NMEs)
2010–2015
Number of NME approvals
2014
2015
Orphan EMA approvals
Orphan FDA approvals
All EMA approvals
All FDA approvals
Fig. 2.9.1  Approval of new molecular entities.
120
section 2  Background to medicine
works. No doubt also, however rigorous the regulatory system, it can
also fail by approving a drug for a population for which the risks
were not well defined, or by failing to approve a drug that might ac-
tually benefit some patients, but not those originally studied.
Nonetheless, once a suitable small molecule or biological drug
candidate has been identified, typically employing in vitro and/​or
in vivo models of disease, then extensive pharmacological studies,
including toxicology are undertaken. This is to verify the reasonable
likelihood of safety of the compound before the clinical trial pro-
gramme can be planned in any detail.
Once completed, and submitted for scrutiny to regulatory au-
thorities, then the clinical trial process can move forward in that
country for which regulatory approval for clinical trials has oc-
curred. In recent years, patient groups have become well-​organized
and highly engaged with the early research process: some, directly
fund preclinical research projects through grant programmes or
other charitable mechanisms. In the intense world of rare diseases,
families of patients have been known to go so far as to form their
own biotechnology company in order to recruit the funding re-
quired to treat their family member (see, for example, The Cure by
Geeta Anand).
Clinical trials are typically divided into several phases—​each
encompassing a growing cohort of patients. At each stage, there are
rich opportunities for patients and their organizations to engage.
Phase 1 of clinical development
This phase marks the first opportunity for a novel therapy to be
administered to patients. The primary objective at this stage is to
determine the safety profile of the compound. During phase 1 clin-
ical research, to determine how dosing impacts the safety profile,
varying doses of the compound or biological agent are administered
to small cohorts of patients. Throughout this process, patients are
closely monitored for adverse events and researchers will observe
how the compound is metabolized in the body. In addition, some
clinical trials may also experiment with different routes of admin-
istration for the compound (i.e. oral, infusion, sublingual, and so
on). It is important to note that if the phase 1 study reveals high or
unacceptable toxicity in patients, the trial may be halted and future
development may be completely scrapped.
Phase 2 of clinical development
Once a safety profile has been established using clinical data, the
trial can proceed to the next stage in order to explore the clinical
effectiveness of the therapy —​‘efficacy’. During this period, clinical
researchers are obliged to select various endpoints or health metrics
that will unambiguously determine the potential benefit of the pu-
tative treatment on patients for whom the licensed product is envis-
aged. Clearly the selection of what regulatory authorities will agree
to as informative clinical trial endpoints vary widely and depend
entirely on the disease in question. Some more traditional measures
are: overall survival time, severity of symptoms, and quality of life.
Alternatively, if known, clinical researchers may attempt to use a
biomarker or other alternative measure, which is often referred to
as a surrogate endpoint. In some quarters, surrogate endpoints are
becoming increasingly favoured by those who develop innovative
medicines; this is because they can provide more precise measures
of disease progression or remission and often can be rapidly de-
termined in accessible biological fluids using facile modern assays.
Sequential measurement of biomarkers, in turn, can expedite and
facilitate the clinical trial process, thereby reducing the burden of
assessment of traditional endpoints. However, the predictive value
of biomarkers for clinical outcomes is frequently questioned, and a
high level of proof is often demanded for their acceptance in clinical
trial endpoints as opposed to time-​honoured monitoring in clin-
ical practice. Improved scientific methods are thus needed to select
biomarkers for disease assessment from an often-​overwhelming
array of potential biomolecules whose abundance is known to be
altered as a subsequence or consequence of the pathogenesis of a
given condition.
Phase 3 of clinical development
The final and most extensive stage of the evaluation of putative new
therapies in clinical trials seeks to elucidate whether the experi-
mental therapy is safe and effective as intended for use. Phase 3 trials
are typically completed in a randomized controlled design in which
patients are randomly assigned to receive the experimental therapy
or a placebo or in some situations, an established, active therapy.
These studies are typically double-​blind, that is both patients and
doctors or sponsor are unaware as to which treatment is being ad-
ministered. Once the trial is completed, the sponsors will assemble
a large package of data to be submitted to the regulatory authority
for review. Regulatory review in Europe can take usually about one
year: in this time, the authorities are charged to make a determin-
ation about whether the benefits of a new therapy outweigh the risks,
for if so, the novel therapy can be approved for marketing.
Phase 4/​Postmarket surveillance
Once a new therapy has been approved for marketing and distribu-
tion, sponsors and public health authorities will continue to monitor
patients for adverse events. Postmarketing studies may continue for
years and new data may emerge that requires regulators to recon-
sider the benefit-​risk determination (Fig. 2.9.2).
Expanded access/​named patient
treatment/​Compassionate use
Some patients facing acute, life-​threatening diseases, may seek out
experimental treatment outside of the normal clinical trial process
for unapproved therapies. This increasingly common scenario is
referred to as ‘compassionate use’ or ‘expanded access’ or named
patient treatment. In many cases, the impetus to obtain access to
experimental treatments is patient-​driven and typically deployed as
a final or ‘last-​ditch’ therapeutic measure. The increasing complexity
of the regulatory process and with it, the prolonged development
period for many drugs, has had consequential effects on patients
with deteriorating health. Delays in the development process can
cause an emerging a crisis in patients suffering from the effects of
severe and progressive diseases, and has led to calls for earlier access.
This state of affairs applies particularly to drugs for patients with
life-​threatening conditions where the risk of adverse and unwanted
effects becomes less of an ethical concern. Advocacy groups sup-
porting patients in such a position are exploring various options to
enable patients to obtain access to an experimental drug (ostensibly
with appropriate clinical surveillance and a practical framework for
its administration). The difficulty with this worthy principle is that
2.9  Engaging patients in therapeutic development
121
the clinical development process alongside regulatory approval is in-
tended to protect patients from unnecessary risks.
In the United States, the number of requests for expanded ac-
cess to experimental therapies, a process regulated by the Food
and Drug Administration, increased from 1014 in 2010 to 1873 in
2014. This burgeoning, and international trend poses several chal-
lenges for patients, the biopharmaceutical industry, and regulatory
bodies. The complexities of these interdependent relationships are
particularly exposed in instances when patients, their families and
advocacy groups deploy large-​scale social media campaigns to
pressurize companies into providing early access to experimental
therapies that are still in development and under regulatory scru-
tiny (see Box 2.9.1).
Patient groups partnering with industry
to fund drug development
Cystic fibrosis is an inherited genetic disorder caused by a muta-
tion in the cystic fibrosis transmembrane regulator gene, CFTR; the
disease causes progressive damage to the respiratory and digestive
systems (see Chapter 18.10). One of the leading patient groups is
the Cystic Fibrosis Foundation (CFF), which began funding spe-
cific drug development efforts in the late 1990s with investments in a
biotechnology company called Aurora Biosciences. In 2001, Aurora
was acquired by Vertex Pharmaceuticals and received continued in-
vestments from CFF that would prove vital in sustaining the risky
cystic fibrosis research programme. This programme would go on
to develop a novel pharmacological chaperone therapy, Kalydeco,
which was approved for marketing in the United States in 2012, and
was the first approved drug designed to treat the cause of cystic fi-
brosis as opposed to symptoms. The Foundation will receive a one-​
time pay-​out of $3.3 billion (US) in exchange for the rights to the
royalties from the drug. This scenario represents a novel paradigm
for funding drug development for a rare disease. CFF is utilizing
those revenues to continue investing in further cystic fibrosis re-
search and treatment.
Patient engagement in drug development:
Disease burden, disease measurement,
and benefit-​risk
Unravelling the true burden of disease requires close input and
collaboration with patients. This is especially true for diseases
PHASE I
TENS
HUNDREDS
NUMBER OF VOLUNTEERS
THOUSANDS
NDA/BLA SUBMITTED
APPROVAL
PHASE II
PHASE III
PHASE IV
Regulatory
approval
POST-APPROVAL
RESEARCH &
MONITORING
CLINICAL TRIALS
REVIEW
BASIC
RESEARCH
DRUG
DISCOVERY
PRE-
CLINICAL
POTENTIAL NEW MEDICINES
IND SUBMITTED
Fig. 2.9.2  The contemporary clinical development process.
Box 2.9.1  Case study: Josh Hardy
One recent and widely publicized case involved Josh Hardy, who
had experienced severe health complications stemming from a rare
kidney cancer. After a bone marrow transplant, Josh was infected
by an adenovirus that, if untreated, would probably result in death.
However, the patient’s responsible physician at St. Jude Children’s
Research Hospital recognized that an experimental therapy,
brincindofovir had the potential to treat the infection. Chimerix
Inc., the company that was developing the treatment, refused to
provide access, citing concerns about cost, workload, and sustain-
ability of such requests: ‘The company says helping him will slow
efforts to get it on the market ...’, as reported by CNN. Chimerix
argued that as a small 54-​person company in Durham, North
Carolina, diverting their resources to this one case could inhibit
their broader effort to get the drug approved as quickly as possible.
What followed was a social media firestorm, with thousands of
emails, Facebook, and Twitter messages being directed to the com-
pany Chief Executive Officer, Dr Ken Moch, and other employees
imploring them to ‘Save Josh.’ The company also reported receiving
more extreme messages, including death threats. Ultimately, the
company relented and with the engagement of the Food and Drug
Administration created a new clinical trial to enrol Josh as the first
patient rather than provide the treatment through the expanded
access pathway. Josh responded well to the treatment and did well
for 2 years but later declined and died due to other causes in 2016.
122
section 2  Background to medicine
that may not be well-​established in the medical literature, as many
assumptions or clinical dogma may turn out to be unfounded or
tenuous after further investigation. Disease burden is particularly
poorly understood in rare diseases in part due to the limited op-
portunity for surveys conducted by investigators, and the failure
to invest in tools for measuring disease. Patient engagement in
rare diseases is now almost the rule as patient groups fund and
develop surveys of disease or natural history studies. The Cystic
Fibrosis (CFF) Foundation is one example of an organization that
established a network of centres of excellence, developed common
methods for measuring and tracking cystic fibrosis (CF), and pub-
lished information on the disease that helped all investigators or
sponsors considering development treatments. This work has suc-
cessfully supported efforts to develop new treatments for CF. Many
other rare disease groups are replicating this effort to some degree,
and developing data on their disease in the precompetitive space so
that diverse therapeutic avenues, predicated on deep phenotyping
and extensive clinical knowledge, can be explored. The increasing
use of computerized international disease registries has had a
useful effect in this field.
Quantifying disease severity
Disease measurement is one of the more complex areas of de-
velopment; and patient groups alone are not usually able to pro-
vide all the information concerning clinical measurement or
patient-​reported outcomes. Here an important synergy can de-
velop, based on consultation and collaboration with experts who
are able to develop quantitative instruments. These activities
are expensive and often there is a need for industrial sponsor-
ship to facilitate completion of such enabling work. Challenges
in obtaining access to these data as public resources has led to
more efforts to assure that all such natural history data sets do
not end up as proprietary information of corporate sponsors, but
are available for anyone working on a new treatment or develop-
ment of tools.
Determination of the true benefit-​risk assessment for any treat-
ment requires comprehensive understanding of the risks of un-
mitigated disease and a deep appreciation of the trade-​off between
one disease risk and another risk—​that which emanates or might
emanate from the treatment. In the past, industrial sponsors and
regulatory authorities spoke for the patients: it is now clear that
there this tradition is an unwarranted presumption that lacks any
depth of understanding about the experience of a chronic severe
disease or indeed what patients, often supported by their families
and the larger disease-​family are willing to do to contribute to the
management of their disease and future therapies.
A recent example of patient engagement has been the work of a
Duchenne muscular dystrophy patients’ group in the United States,
called Parent Project Muscular Dystrophy (PPMD). Duchenne
muscular dystrophy (DMD) is a genetic disease that leads to severe
loss of muscle bulk and function. Males with the disease typically
live into their late twenties. In 2014, PPMD, convened a variety of
stakeholders from throughout the Duchenne community and de-
veloped a comprehensive survey tool to rigorously capture the key
aspects of benefit:risk. One critical finding was that most parents
would support a treatment that stabilizes their child’s strength for
one more year at the cost of one less year of life. That is a trade-​off
that would not normally be expected and, in most instances, runs
not only counter to the usual traditions of professional ethics and
decision-​making, but also provides an unexpected foil to the con-
ventional view of what patients really want.
After the outcomes of this survey were disseminated, a broad co-
alition was assembled, including parents of DMD patients, DMD
patients themselves, academic researchers, and physicians, as well
as representatives from the biopharmaceutical industry to discuss
the implications for the development of treatments. The outcome
of this collaboration was a guidance document that explores key
areas for drug development including:  benefit-​risk assessments,
diagnostic tools, natural history studies, clinical trial designs, and
outcome measures including the use of biomarkers. Ultimately, this
document was submitted to the Food and Drug Administration for
consideration and was incorporated in part as an official guidance;
it provides a practical roadmap for the biopharmaceutical industry
as it seeks to pioneer useful new treatments.
Conclusions
Successful treatment of patients is at the heart and goal of medi-
cine but the increasing complexity of treatments and under-
standing of diseases has led to a beneficial societal metamorphosis
which involves patients at all levels of therapeutic development.
Including patients throughout the process of drug development
enhances understanding of pathogenesis, measurement, and the
personal impact of disease; genuine participation focuses attention
on the authentic therapeutic targets of the disease. In summary,
increasing involvement of patients and cognate organizations
and advocacy groups broadly in drug development, has driven
important changes to the regulatory process and almost always
expedites achievement of the common goal to deliver safe and
salutary new therapies.
FURTHER READING
Clinical trials
U.S. Food and Drug Administration. The FDA’s Drug Review Process:
Ensuring Drugs Are Safe and Effective. http://​www.fda.gov/​drugs/​
resourcesforyou/​consumers/​ucm143534.htm
Patient engagement
FasterCures (2011). Back to basics: HIV/​AIDS advocacy as a model
for catalyzing change. FasterCures and HCM Strategists (14 June
2011).
Kim E, Lo AW (2019). Venture Philanthropy: A Case Study of the
Cystic Fibrosis Foundation (April 23, 2019). https://ssrn.com/
abstract=3376673
National Health Council (NHC) and Genetic Alliance (2015). Advancing
Meaningful Patient Engagement in Research, Development and Drug
Review. (22 September 2015). https://​www.nationalhealthcouncil.org/​
sites/​default/​files/​PatientEngagement-​WhitePaper.pdf
Parent Project Muscular Dystrophy, Franson TR, Peay H (eds).
Benefit-​Risk Assessments in Rare Disorders, the Case for Therapeutic
Development in Duchenne Muscular Dystrophy as the Prototype for
New Approaches. http://​www.parentprojectmd.org/​site/​DocServer/​
2.9  Engaging patients in therapeutic development
123
br_​paper_​v11_​_​2_​.pdf;jsessionid=2C381495CB3753608053FD8D
D624B686.app247d?docID=14503
Biomarkers as primary endpoints
Institute of Medicine (2010). Evaluation of Biomarkers and Surrogate
Endpoints in Chronic Disease. Micheel CM, Ball JR (eds) http://​www.
nap.edu/​catalog/​12869/​evaluation-​of-​biomarkers-​and-​surrogate-​
endpoints-​in-​chronic-​disease
Kakkis E, et  al. (2015). Recommendations for the development
of rare disease drugs using the accelerated approval pathway and
for qualifying biomarkers as primary endpoints. Orphanet J Rare
Dis, 10, 16.
Compassionate use/​expanded access
European Medicines Agency. Compassionate Use. http://​www.ema.
europa.eu/​ema/​index.jsp?curl=pages/​regulation/​general/​general_​
content_​000293.jsp&mid=WC0b01ac058007e691
U.S. Food and Drug Administration. Expanded Access:  Information
for Patients. https://​www.fda.gov/​NewsEvents/​PublicHealthFocus/​
ExpandedAccessCompassionateUse/​default.htm