Академический Документы
Профессиональный Документы
Культура Документы
Lecture Notes
Epidemiology
and Public Health
Medicine
Richard Farmer
MB, PhD, FFPH, FFPM
Professor of Epidemiology
Postgraduate Medical School
University of Surrey
Stirling House
Surrey Research Park
Guildford
Surrey, UK
Ross Lawrenson
MRCGP, FAFPHM, MD
Dean of Medicine & Professor of Primary Health Care
Postgraduate Medical School
University of Surrey
Stirling House
Surrey Research Park
Guildford
Surrey, UK
Fifth Edition
PIDPR 5/21/04 11:21 AM Page iv
The right of the Authors to be identified as the Authors of this Work has been asserted in accordance with the
Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any
form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK
Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.
First published in 1977 under the title Lecture Notes on Epidemiology and Community Medicine
Second edition 1983
Third edition 1991
Fourth edition 1996
Reprinited 1997, 1998
Fifith edition 2004
Farmer, R. D. T.
Lecture notes on epidemiology and public health medicine / Richard D.T. Farmer, Ross Lawrenson. — 5th ed.
p. ; cm.
Includes bibliographical references and index.
ISBN 1-4051-0674-3
1. Epidemiology. 2. Public health.
[DNLM: 1. Epidemiologic Methods. 2. Health Services. 3. Preventive Medicine. WA 950 F234L 2004] I. Title:
Epidemiology and public health medicine. II. Lawrenson, Ross. III. Title.
RA651.F375 2004
614.4 — dc22
2004000864
ISBN 1-4051-0674-3
A catalogue record for this title is available from the British Library
Set in 8/12 Stone Serif by SNP Best-set Typesetter Ltd., Hong Kong
Printed and bound in India by Replika Press Pvt. Ltd.
Contents
v
PIDPR 5/25/04 2:22 PM Page vi
Preface
The UK Government is committed to improving is relevant effective and efficient are evident
the nation’s health and reducing health inequali- within the NHS in the UK as in many other coun-
ties. Whilst the provision of health care is in a state tries. This is exemplified in the NHS plan The New
of constant change it is important to remember NHS; modern, dependable (1997).
that the key objective is to maintain and improve This new edition of Lecture Notes: Epidemiology
the health of the population. This was recognized and Public Health Medicine, as before, covers
by Derek Wanless in his report Securing Good Health the basic tools required for the practice of epidemi-
for the Whole Population published on 25th Febru- ology and preventive health. The chapters in the
ary 2004. This document focused on prevention first section of the book outline the principles of
and the wider determinants of health. To prevent epidemiology and lead the reader to some classic
disease and improve health it is essential to under- examples from the medical literature. A new chap-
stand why diseases arise; and conversely why, in ter has been included on the practice of evidence-
many cases, they do not. To do this it is necessary based medicine. The second section of the book
to study the distribution and natural history of dis- covers the areas of prevention and control of dis-
eases in populations and to identify the agents re- ease — in particular the chapter on health promo-
sponsible; effective strategies can then be planned. tion has been updated to reflect the advances that
In the same way that the provision of health care have occurred over the last eight years. The chapter
should be evidence based, the introduction of new on occupational health has been dropped from
preventive strategies should be rigorously evaluat- this edition.
ed and researched. The application of evidence- The final section has been updated to reflect the
based medicine is applicable to both clinical and changes in the provision of health care. Change is
public health practice. now a constant in the health services and the shift
In the past the importance of public health between central control and devolution of respon-
medicine and the related basic medical sciences, in sibility will continue to ebb and flow. At the time of
particular medical statistics and sociology applied writing we are seeing more devolution of responsi-
to medicine, was not emphasized in the under- bility and the primary care trusts have a tremen-
graduate medical education. This relative neglect dous opportunity to deliver health services that are
changed in the 1990s with the GMC's recommen- truly responsive to patient needs. We should also
dation on undergraduate medicine Tomorrow’s recognise the successes brought about through the
Doctors. This publication recommended that the introduction of health targets — the incidence of
theme of public health medicine should figure heart disease is falling; the mortality from breast
prominently in the undergraduate curriculum, en- and cervical cancer has fallen as screening for these
compassing health promotion and illness preven- diseases has increased; and many infectious dis-
tion, assessment and targeting of population needs eases, for practical purposes, have been eliminated.
and awareness of environmental and social factors We still have many challenges — obesity and dia-
in disease. This explicit and forceful advocacy for betes are increasing rapidly, alcohol abuse has been
the discipline from a body as influential as the recognized as a growing social problem and the
GMC undoubtedly gave added momentum to spread of sexually transmitted disease and HIV still
the development of medical education. Similar poses challenges.
changes emphasising the importance of disease We hope readers will find that this new edition
prevention and the need to ensure that health care continues to provide a basic structure to under-
vi
PIDPR 5/21/04 11:21 AM Page vii
Preface
standing epidemiology and public health and that diseases and immunization. We must also recog-
many of our readers will be encouraged to delve nise the contribution of Emeritus Professor David
deeper into the subject. Miller who was the co-author of the first four
editions of this book. We would also like to thank
Mrs Pat Robertson, our PA at the University, for her
Acknowledgements
help and support.
We are greatly indebted to Dr Peter English of the
Health Protection Agency for his help and support Richard Farmer
in the updating of the chapters on infectious Ross Lawrenson
vii
PIDPR 5/21/04 11:21 AM Page viii
List of Abbreviations
viii
PID1 5/21/04 11:22 AM Page 1
Part 1
Epidemiology
PID1 5/21/04 11:22 AM Page 2
PID1 5/21/04 11:22 AM Page 3
Chapter 1
General principles
The word epidemiology is derived from Greek and demiology is needed to appraise critically other
literally means ‘studies upon people’. Modern people’s contributions.
methods of epidemiological enquiry were first de-
veloped in the course of investigating outbreaks
The investigation of causes and
of infectious diseases in the 19th century. In
natural history of disease
contemporary medical practice the scope and
applications of epidemiology have been greatly One of the most important roles of epidemiology is
extended. Similar methods are now used in the to provide a broader understanding of the causes
investigation of the causes and natural history and natural history of diseases than can be gained
of all types of disease. They are also used in from the study of individuals. Clearly, the experi-
the development and assessment of preventive ence of an individual doctor is limited because the
programmes and treatments, the assessment number of patients with a particular condition
of the safety of medicines and in the planning with whom he or she comes into contact is rela-
and evaluation of health services. In contrast to tively small. The less frequent a disease, the more
clinical medicine, epidemiology involves the fragmentary is an individual doctor’s experience
study of groups of people (populations) rather and understanding of it. If the experience of many
than the direct study of individuals. This does doctors is recorded in a standard form and pro-
not diminish its relevance to clinical medicine. perly analysed then new and more reliable knowledge
On the contrary, it enhances the practice of medi- may often be acquired. This will assist in diagnosis,
cine by increasing the understanding of how give a better understanding of prognosis and point
diseases arise and how they might be managed to optimum management policies. Such systema-
both in the individual and in societies as a tic collection and analysis of data about medi-
whole. cal conditions in populations is the essence of
Most doctors find themselves involved with epi- epidemiology.
demiology through the use they make of the The value of pooling doctors’ experience in elu-
results of studies or sometimes as participants in cidating the causes of disease is well illustrated by
investigations. It is important that all professionals the story of the epidemic of fetal limb malforma-
involved in health care should have an under- tions (phocomelia) that was caused by women tak-
standing of the subject so that they can use epi- ing the drug thalidomide during the first trimester
demiological methods in the study of health and of pregnancy. Phocomelia, a major deformity in
disease. More importantly, a knowledge of epi- the development of the limbs, was a recognized
3
PID1 5/21/04 11:22 AM Page 4
congenital abnormality long before the invention The thalidomide incident underlines the need to
of thalidomide. A drawing by Goya called ‘Mother collect, collate and analyse data about the occur-
with deformed child’ bears witness to the fact that rence of disease in populations as a matter of rou-
it occurred in 18th century Spain (Fig. 1.1). Under tine. This will increase the probability that causes
normal circumstances it is a very rare abnormality. will be identified early and, whenever possible,
Any doctor may encounter such rare conditions at eliminated. However, even with the most efficient
some time during his or her professional life. Little and complete system of recording medical obser-
can be done to correct the malformation and, be- vations, it is unlikely that the causes of all disease
cause the condition is well known, it is unlikely to will be identified. It is interesting to speculate
warrant the preparation of a case report for publi- about what would have happened had thalido-
cation. If, over a short period of time, each of a mide been universally lethal to the fetus before the
dozen or so doctors or midwives throughout the 12th week of pregnancy. The excess spontaneous
country delivered a child with such an abnormal- abortions might have passed unnoticed, some
ity, each would be personally interested but the even to the pregnant woman, and the possibility
significance of these individual cases would pass that thalidomide had any deleterious effect on the
unnoticed unless the doctors or midwives commu- human fetus would not have come to light. The
nicated with each other or there was a central re- discovery of such causal relationships requires
porting system. This is what happened early in the other approaches, but still depends on the study of
course of the thalidomide episode. One of the les- populations and cannot be established by exami-
sons learned was highlighted by the Chief Medical nation of individual cases. The same is true for
Officer in his 1966 annual report. He said that it most proposed causes (agents) and other factors
‘. . . focused attention on the lack of information which may determine or predispose to the occur-
concerning the different types of congenital mal- rence of disease.
formations. Had a national scheme for notification
been available at this time, it is probable that the
Disease in perspective
increase in limb deformities would have been
noticed earlier and perhaps some of the tragedies Another application of epidemiological tech-
could have been avoided’. niques is to give perspective to the range of diseases
facing doctors and the diversity of their natural
history. The individual clinician only sees a sele-
cted and comparatively small proportion of sick
people, and so may gain an erroneous impression
of the relative frequency of different conditions in
the community as a whole. He or she may also fail
to appreciate the range of different ways in which
diseases present and progress. This is important
since, consciously or not, the clinician tends to rely
on his or her personal experience to assess the like-
lihood of particular diagnoses and their prognosis
when deciding management policy. Rather they
should rely on unbiased evidence obtained from
population studies.
4
PID1 5/21/04 11:22 AM Page 5
dence or prevalence may also distort the view of services can be provided. Such trials are becoming
the health care needs of the community. In the increasingly numerous, but they usually need to
National Health Service and in most health care be on a large scale to produce reliable results.
systems throughout the world, attempts are made Although this is expensive and time consuming
to organize services according to priorities set by it is necessary in the long-term interests of health
objective criteria rather than allowing them to be care.
dictated solely by subjective judgements and tradi-
tional provision. An important report published in
Clinical medicine and epidemiology
the early 1980s called Social Inequality and Health
(The Black Report) drew attention to some of the It will be clear from the above that there are impor-
major differences that persist in the patterns of ill- tant contrasts between the approaches to disease
ness and disability in England and in the use of by clinicians and by epidemiologists. Recognition
health services between different socioeconomic of these differences helps understanding of the
groups. For example, men in social class V were subject. The clinician asks the question ‘What dis-
reported to suffer from long-standing illnesses ease has my patient got?’ whereas the epidemiolo-
almost twice as often as those in social class I but gist asks ‘Why has this person rather than another
they consulted their general practitioner only developed the disease? How could it be prevented?
about 25% more often. This observation suggests a Why does the disease occur in winter rather than
serious failure to match needs with appropriate summer? Why in this country but not in another?’
services. It calls for detailed investigation of the rel- In order to answer such questions it is necessary to
evant population groups to elucidate the reasons compare groups of people, looking for factors that
for it and the implications for future health care distinguish people with disease from those with-
provision. out. Underlying the investigation of disease in
this way is the belief that the misfortune of an
individual in contracting a disease is not due to
Evaluation of medical interventions
chance or fate but to a specific, definable and
Epidemiology is of value in testing the usefulness preventable combination of circumstances or
(and safety) of medical interventions. Although personal characteristics.
many existing remedies have never been subjected For a clinician, the utility of a diagnosis is a
to trial, everyone nowadays recognizes the neces- pointer to management decisions. Therefore the
sity to conduct clinical trials of a new drug or diagnostic precision required is related to the speci-
vaccine before it is introduced into medical prac- ficity of treatments that are available. For an epi-
tice. This is the only way to demonstrate that a demiologist, diagnosis has different significance. It
particular drug or vaccine is likely to improve the is a way of classifying individuals in order to make
patient’s prospects of recovery or to prevent disease comparisons between groups. Lack of diagnostic
from occurring or progressing. Once a product precision will result in poor definitions of cate-
has been launched on the market it is necessary to gories. This makes it difficult to identify the subtle
continue to monitor its effects (both beneficial and yet important differences between groups which
adverse) in order to ensure that patients are are critical to the understanding of the causes and
being prescribed effective and safe medication. In prevention of disease.
recent years the application of epidemiological The clinician is interested in the natural history
methods to the assessment of medicines has of disease for prognostic purposes in an individual
become firmly established and is referred to as patient. He or she is usually content to express
pharmacoepidemiology. prognosis in terms such as ‘good’, ‘bad’, ‘about 6
The same principles are being applied to other months’, etc. It is unhelpful to the clinician and
treatments, such as surgery or physical therapy, the patient to attempt to introduce mathematical
and even to the alternative ways in which health precision into prognostic statements, such as ‘He
5
PID1 5/21/04 11:22 AM Page 6
has a 10.9% chance of surviving symptom-free for similar conditions, 26.5% survive symptom-free
5 years’, though it may sometimes be appropriate for 5 years. What accounts for this difference
to give a range of expected survival times, for ex- which could assist in planning treatment or pre-
ample between 3 and 7 years. By contrast, in popu- ventive strategies?
lation studies precision is helpful because it may While there are these clear differences between
allow the investigator to identify variables that clinical and epidemiological approaches to med-
have significant effects on outcome. For example, ical problems and while their immediate purposes
it may be informative to investigate why in one are different, it is also clear that the results of epi-
group of patients 10.9% survive symptom-free for demiological investigations can contribute greatly
5 years while in another group with approximately to the scientific basis of clinical practice.
6
PID2 5/21/04 11:24 AM Page 7
Chapter 2
‘Cause’ and ‘risk’, and types of
epidemiological study
7
PID2 5/21/04 11:24 AM Page 8
Statistically significant associations between ex- to the direct cause. Thus, the investigation of cause
posure and the occurrence of disease may occur by is usually a complex exercise that involves the
chance, i.e. they are neither causal factors nor con- identification of both the characteristics of suscep-
founding factors. tible individuals (and sometimes characteristics of
• A determinant is an attribute or circumstance individuals who appear to be unusually resistant)
that affects the liability of an individual to be and the types of exposure to external agents that
exposed to or, when exposed, to develop disease are necessary for the disease to occur.
(e.g. hereditary predisposition, environmental Ideally, causal hypotheses should be explored by
conditions). carefully controlled experiments in which the
• A confounding variable is a factor that is signifi- effects of each of the postulated causes can be ex-
cantly associated both with the occurrence of a dis- amined independently of other factors. In animal
ease in a population and with one of its causes or studies, for example, it is usually possible to ex-
determinants, but is not itself a cause. For example, clude the effects of inheritance by breeding a
heavy cigarette smoking and a high alcohol family of animals for study. The possible effects of
consumption tend to occur together. Smoking the general environment and diet that are not of in-
is causally associated with carcinoma of the terest for a particular investigation can be eliminated
bronchus and because heavy drinking is associated by rearing the whole family under standard condi-
with cigarette smoking, alcohol consumption will tions. Then the effects of a suspected causal agent
tend to correlate with carcinoma of the bronchus, can be assessed by exposing a sample of the ani-
even though it is not a cause. mals to it whilst protecting others from it. In such
The concept of risk includes both the ‘risk’ that a experiments the only major difference between
person exposed to a potentially harmful agent will the two groups is their exposure to the agent under
develop a particular disease and the ‘risk’ that a study. Such a study design allows the observed
particular intervention will beneficially or adver- effects, if any, to be attributed unequivocally to the
sely influence the outcome. The indices com- agent under investigation. It is impractical and un-
monly used to measure risk are set out below. ethical to undertake studies of such experimental
Risk factors are different but are involved in both purity amongst human subjects. The identification
concepts. They are factors that are associated with of the causes of diseases and factors that alter
a particular disease or outcome. They can be asso- the course of a disease in humans necessitates
ciated either by chance or because they influence adopting methods whereby hypotheses can be
the course of events. All causal agents and determi- tested without prejudice to the individuals being
nants are ‘risk factors’ but not all ‘risk factors’ are studied.
causal agents or determinants. The methods that are used in epidemiological
The purpose of epidemiological studies is to studies represent practical compromises of the
identify causes and determinants and to define and above ‘ideal’ design. It is essential therefore that
measure risks by the application of the scientific the results of any investigation are interpreted in
methods set out in the next four chapters. full knowledge of the limitations imposed by the
compromises. In particular, it is important to take
account of the effects of confounding variables
Causes and determinants
and, when these cannot be controlled in the study
Few diseases have a single ‘cause’. Most are the design, to allow for them in the analysis.
result of exposure of susceptible individuals to one
or more causal agents. Even in the case of some of
Distinguishing causes and determinants
the most straightforward illnesses, for example
from chance association
infections, exposure to the causal agent does not
inevitably result in disease. Many other factors may The observation that a disease is statistically associ-
influence the development of disease in addition ated with a suspected agent is clearly not proof that
8
PID2 5/21/04 11:24 AM Page 9
the suspected agent causes the disease. For exam- (HIV). Most people with AIDS could have become
ple, there is a higher prevalence of alcoholism infected with HIV on many occasions. By the time
amongst publicans and bar staff than in most other the disease is apparent it is impossible to prove that
occupational groups. This does not necessarily a particular exposure or type of activity led to the
mean that being a publican causes alcoholism. infection. In some circumstances it is not possible
There are several other possible explanations of to date the start of the disease; for example, carci-
this phenomenon, including the fact that people noma of the endometrium usually occurs many
who tend towards excessive alcohol consumption years before symptoms are manifest and the dis-
may seek jobs in bars. ease is diagnosed. In such cases, although it is usu-
The types of evidence that can be used to distin- ally possible to date exposures to suspected causal
guish a causal from a fortuitous association are dis- agents they cannot be related in time to the
cussed below. Many of the criteria appear to be disease.
simple and straightforward but it can be seen that
each of them can present practical difficulties.
Distribution of the disease
The spatial or geographical distribution of the dis-
ease should be similar to that of the suspected
Distinguishing cause from association
causal agent. For example, endemic goitre occurs
• Strength of association in areas where the iodine content of drinking water
• Time sequence is low. Sometimes a geographical association
• Distribution of the disease
between the distribution of the disease and its
• Gradient
suspected causal agent may be difficult to
• Consistency
• Specificity demonstrate. This is a particular problem if there is
• Biological plausibility a significant time interval between exposure and
• Experimental models manifestation of disease and there have been
• Preventive trials movements in the population during that interval.
For example, legionnaires’ disease commonly
occurs in people who become infected as a result of
casual or transient exposure to the source and who
Strength of association
may be widely scattered before they develop symp-
The stronger the association the more likely it is toms of the disease. In these circumstances it is
to be causal. This is usually measured in terms necessary to map the location of cases to the place
of relative risk, i.e. the incidence of disease in where they were at the time it is hypothesized that
people exposed to the suspected agent compared they were exposed to the causal agent.
with the incidence in those not so exposed (see
below).
Gradient
The incidence of disease should correlate with the
Time sequence
amount and duration of exposure to the suspected
If an agent causes a disease then exposure must cause (population dose–response). For example,
always precede its onset. Thus eating contaminated mesothelioma was noted to be more common than
food can cause diarrhoea and vomiting 24 h later. A expected in people working with asbestos and in
practical problem is that it is often difficult to date those living near to factories that emitted asbestos
exposure to a suspected causal agent; for example, dust into the atmosphere. The incidence was
the acquired immune deficiency syndrome (AIDS) greatest in workers exposed for the longest periods
is usually not manifest until many years after in- and those living in closest proximity to the
fection with the human immunodeficiency virus factories.
9
PID2 5/21/04 11:24 AM Page 10
Preventive trials
Specificity
Control or removal of the suspected agent results
Specificity was amongst the criteria that could be in decreased incidence of disease. For example,
used to distinguish chance associations from cause when it was appreciated that the use of thalido-
suggested by Hill in 1965. He proposed that a sin- mide for treatment of morning sickness in preg-
gle true cause should lead to a single effect, not nancy was associated with a high incidence of
multiple effects. This criterion is particularly useful phocomelia, the drug was withdrawn and the epi-
for infectious agents. It is not necessarily valid for demic rapidly ceased.
non-infectious disease since it is widely accepted
that a single agent can be causally associated with
Risk
a number of outcomes; for example smoking ciga-
rettes can cause lung cancer, heart disease and There are three common indices of risk: absolute,
chronic obstructive airway disease, amongst other relative and attributable.
diseases.
10
PID2 5/21/04 11:24 AM Page 11
11
PID2 5/21/04 11:24 AM Page 12
Whole
No disease
population
or random
Disease
sample
Case–control Present
Cases of disease
Absent
Present
Matched controls
Absent
Figure 2.1 Comparison of cohort
and case–control study designs.
12
PID2 5/21/04 11:24 AM Page 13
and quicker to complete than cohort studies, rarely ies of treatment, prevention and control measures
give clear-cut proof of cause. and the way in which health care is provided. They
can also be used to assess the comparative effec-
tiveness and efficiency of different interventions.
Intervention studies
The most familiar study design of this type is the
These are essentially experiments designed to clinical trial. Ethical considerations are particularly
measure the efficacy and safety of particular types important when considering the design and execu-
of health care intervention. This can include stud- tion of any kind of intervention study.
13
PID3 5/21/04 11:25 AM Page 14
Chapter 3
Descriptive studies
14
PID3 5/21/04 11:25 AM Page 15
1000
800
Time Chemotherapy and
600 BCG vaccination
15
PID3 5/21/04 11:25 AM Page 16
data are plotted on a logarithmic scale (Fig. 3.2) it due to a complex series of changes. Until the
becomes clear that the introduction of specific 1950s, these were mainly an increase in the resist-
measures for the control and treatment of tubercu- ance of the population to infection and environ-
losis was associated with an acceleration in the es- mental changes that reduced the chances of
tablished decline in mortality. It is now thought acquiring infection. After the early 1950s, the rate
that the decline in mortality from tuberculosis was of decline in mortality was accelerated by the
newly available methods of management.
It is frequently necessary to examine secular
2000 trends both as changes in rates (arithmetic scale)
and as rates of change (logarithmic scale) if the
1000 nature of a trend is to be fully appreciated.
The secular trend in mortality from carcinoma of
the bronchus shows the opposite picture to that for
tuberculosis (Fig. 3.3). Until quite recently it had
SMR (base years 1950–52)
Periodic changes
10
1871 1891 1911 1931 1951 1971
These are more or less regular or cyclic changes in
Years
incidence. The most common examples are seen in
infectious diseases. For example, until a vaccine
Figure 3.2 Tuberculosis mortality in England and Wales,
1871–1971 (logarithmic scale). (Reproduced with permis-
was introduced, measles showed a regular biennial
sion from Prevention and Health: Everybody’s Business, cycle in incidence in England and Wales (Fig. 3.4).
HMSO, 1976.) The cycles were probably the result of changes in
120
Cigarette consumption ¥ 109
1000
100
Deaths per million
800 Cigarett
e consum
ption 80
600
60
400 (d eaths)
Female 40
Figure 3.3 Carcinoma of lung,
200 20 bronchus and trachea. Deaths per mil-
lion population in England and Wales,
0 0 1955–92, and cigarette consumption
1955 1960 1965 1970 1975 1980 1985 1990
Year per year. (Reproduced with permission
of the Office of National Statistics).
16
PID3 5/21/04 11:25 AM Page 17
800 000
600 000
Notifications
400 000
the levels of child population (herd) immunity (see mon in summer months when the ambient tem-
p. 105). Other infectious diseases such as whoop- peratures favour the multiplication of bacteria in
ing cough, rubella and infectious hepatitis show food. The regular seasonality of gastrointestinal in-
less regular, but nevertheless distinct, cycles with fections is shown in Fig. 3.5 in which the number
longer intervals between peaks. of notifications of food poisoning for each quarter
in 1974–89 are plotted. A particularly interesting
Seasonality feature of food poisoning incidence is that the
This is a special example of periodic change. The marked seasonality is combined with a noticeable
environmental conditions that favour the pres- secular trend. The number of cases notified from
ence of an agent, and the likelihood of its success- late 1988 and early 1989 was much higher than in
ful transmission, change with the seasons of the previous years. This could be due to contamination
year. Respiratory infections, which spread directly in the food chain, a decline in standards of food
from person to person by the air-borne route, are storage, distribution or preparation, or the result of
more common in winter months when people live an increase in notification rates following public-
in much closer contact with each other than in the ity given to the problem of food poisoning.
summer. By contrast, gastrointestinal infections, Some non-infectious conditions, for example
which spread by the faecal–oral route, often allergic rhinitis, deaths from drowning and road
through contamination of food, are more com- accidents, also display distinct seasonality. For
16 000
14 000
12 000
Cases notified
10 000
8 000
6 000
4 000
2 000
0
Figure 3.5 Quarterly notifications of ‘74 ‘75 ‘76 ‘77 ‘78 ‘79 ‘80 ‘81 ‘82 ‘83 ‘84 ‘85 ‘86 ‘87 ‘88 ‘89
food poisoning in England and Wales, Years (1974–89)
1974–89.
17
PID3 5/21/04 11:25 AM Page 18
Table 3.1 Seasonality of birth of schizophrenic and neurotic people compared with that of the general population
(expected) showing an increased frequency of births of schizophrenic people in the first part of the year but no
seasonality amongst neurotic people. (Adapted from Hare E, Price J, Slater E. Br J Psychiatry 1974; 124: 81–86.)
Quarter of birth
Schizophrenic people
Observed 1383.0 1412.0 1178.0 1166.0
Expected 1292.1 1342.8 1293.1 1211.1
Observed as a percentage of expected 107 105 91 96
Neurotic people
Observed 3085.0 3172.0 2949.0 2882.0
Expected 3024.1 3150.6 3042.0 2844.2
Observed as a percentage of expected 101 101 97 101
most of these, the explanation for the seasonality It should be noted that the seasonality in disease
is not difficult to infer. There are seasonal varia- patterns related to climatic conditions is reversed
tions in the incidence of certain other conditions, in the southern Hemisphere.
however, for which there is as yet no rational ex-
planation. For example, schizophrenic people are
Epidemics
more likely than the general population to be born
in the early months of the year (February and These are temporary increases in the incidence of
March) (Table 3.1). Many hypotheses have been disease in populations. The most obvious epidemics
offered to explain this observation, including the are of infectious diseases such as influenza (Fig. 3.6)
proposition that the disease is caused by an in- but non-infectious epidemics do occur. For exam-
trauterine infection, that the mothers of schizo- ple, there was an increase in asthma deaths in the
phrenic people are more likely to miscarry at 1960s associated with the increased use of pressur-
certain times of the year (thereby resulting in a ized aerosol bronchodilators (Fig. 3.7).
deficit of births in months other than January to The word ‘epidemic’ is also sometimes used to
March) and that the mothers are more likely to describe an increase in incidence above the level
conceive in April to June than are other women. expected from past experience in the same popula-
However, none has yet been proved. tion (or from experience in another population
1400
1200
Number of deaths
1000
800
600
400
200
0
1 14 27 40 1 14 27 40 1 14 27 40 Figure 3.6 Weekly deaths from in-
Week fluenza in England and Wales,
1975–77.
18
PID3 5/21/04 11:25 AM Page 19
800
500
400 Prescriptions
300
100
0
40
30
Deaths
20
19
PID3 5/21/04 11:25 AM Page 20
Table 3.2 Geographical variation in the incidence of disease. Comparison of death rates in England and Wales with those
in Japan (1979) for various causes shows considerable discrepancies. Both are highly industrialized countries with well-
developed health services, but they have very different cultures and racial origins. (Data from World Health Statistics
Annual, WHO, Geneva, 1981.)
ply becomes contaminated, the illnesses that result dwellers is often blurred. Table 3.3 shows some
from the contamination will be clustered in people differences in mortality between urban and rural
living within the distribution area of the water. areas in England and Wales.
‘Spot-maps’ on which cases are marked may show
local concentrations that suggest possible sources.
Variations within single institutions
In interpreting such maps, it is important to relate
the spatial distribution of cases to the density of In institutions such as schools, military barracks,
population. The classical study of the 1854 cholera holiday camps and hospitals, variations in attack
outbreak in the Golden Square area of London by rates by class, platoon, chalet or ward may focus at-
John Snow used such a technique and led him to tention on possible sources or routes of spread. For
identify the particular water pump that was the example, in an outbreak of surgical wound infec-
source of the infection. In this instance, cases were tion, identifying the bed positions of patients,
clustered in the streets close to the Board Street ward duties of staff and theatres used may suggest
pump, while comparatively few cases occurred in the identity of a carrier or other source of infection.
the vicinity of other pumps in the area. Similarly, in places of work the danger of develop-
A special kind of locality difference is that which ing disease may be shown to be inversely related to
exists between urban and rural environments. In distance from source of a chemical hazard.
general, people who live in urban areas are subjec- A high incidence of a disease amongst people
ted to different hazards from those experienced by who share the same environment does not prove
people who live in rural areas. These differences that a factor within the environment was the cause
alter their risk of certain diseases, sometimes to the of the disease. It may be that the people have cho-
advantage of the country person and sometimes to sen, or have been chosen, to share the same envi-
the benefit of the town dweller. In urban areas, ronment because they have an increased
there may be better housing and sanitation but susceptibility to that disease or because of pre-
more overcrowding and air pollution; more leisure existing disease or disability.
but less exercise, fresh food and sunlight; more
industrial hazards but fewer risks of infection from
Personal characteristics
animal contacts and vectors. In industrial societies,
however, where commuting is a common practice, The chances of an individual developing a disease
the distinction between town and country may be affected by personal characteristics. The
20
PID3 5/21/04 11:25 AM Page 21
Table 3.3 Differences in mortality amongst males between urban and rural districts in England and Wales 1969–73
(SMRs).
Malignant neoplasms
Bronchus, trachea and lung 118 109 98 90 79
Bladder 112 109 99 96 82
Chronic rheumatic heart disease 114 110 88 94 85
Ischaemic heart disease 99 106 107 101 95
Influenza 84 98 90 116 111
Bronchitis 117 109 98 96 76
Motor vehicle accidents 87 95 98 99 124
Accidental poisoning 126 110 100 89 67
Homicide 151 99 95 71 56
analysis of data on the incidence of disease in rela- tumours are the dominant determinants of the
tion to the personal characteristics of victims pro- patterns of illness.
vides useful indicators of possible causes. The The fact that the incidence of most diseases
personal characteristics can be classified as shown varies with age can complicate the comparison of
below. morbidity and mortality between populations
with dissimilar age structures. For example, the age
structure of a population of military personnel is
Variation of disease due to personal likely to be substantially different from that of a
characteristics group of practising physicians. Therefore, it is to be
Intrinsic factors (affect susceptibility if exposed to causal expected that the two groups will differ in their in-
agents) cidence of many diseases. In order to make a valid
• Age comparison between these populations it is essen-
• Gender tial to adjust the data to take account of differences
• Ethnic group in their age structure. This procedure is called stan-
Personal habits or lifestyle (affect exposure)
dardization (see Chapter 9).
• Family
Age differences in the incidence of disease may
• Occupation and socioeconomic group
also be accounted for by a so-called ‘cohort effect’.
This occurs when individuals born in a particular
year, or living at a particular point in time, are ex-
Intrinsic factors
posed to the same noxious agent. They then carry
Age an enhanced risk of the disease caused by that nox-
Most diseases vary in both frequency and severity ious agent for a long period, sometimes for the rest
with age. In general, children are more susceptible of their lives. For example, the children who were
to infectious diseases, young adults are more acci- exposed to radiation in Hiroshima and Nagasaki in
dent prone and older adults tend to suffer the re- 1945 when the atomic bombs were detonated have
sults of long exposure to occupational and other had higher than expected incidence of leukaemia
environmental hazards. In infancy, immaturity throughout their lives.
and genetic defects affect susceptibility to disease.
In later life, physiological changes, degenerative Gender
processes and an increased liability to malignant There is evidence that males are intrinsically more
21
PID3 5/21/04 11:25 AM Page 22
Table 3.4 Death rates at different ages for males and these ethnic characteristics from a number of other
females in England and Wales, 2003 (deaths per 1000). factors which affect the incidence of disease, for ex-
Age Males Females ample dietary habits, religious practices, occupa-
tion and socioeconomic status. The effect of
Stillbirths 5.7 4.9
ethnicity on the incidence of disease is best studied
Under 1 year 5.9 4.5
in communities where people of different groups
1–4 years 0.25 0.20
are living side by side and in similar circumstances.
5–9 years 0.12 0.10
10–14 years 0.16 0.11
For example, studies in the UK have shown a high-
15–19 years 0.49 0.24 er prevalence of type 2 diabetes in Asians compared
20–24 years 0.78 0.27 with the white population. This is probably due to
25–34 years 0.94 0.44 genetic differences. On the other hand, in New
35–44 years 1.58 0.94 Zealand the differences in the cot death rate be-
45–54 years 3.85 2.54 tween Maoris and Europeans is related principally
55–64 years 9.7 6.0 to the lower socioeconomic status of most Maoris
65–74 years 27.2 17.0 and lifestyle factors such as maternal smoking.
75–84 years 73.6 50.5
85 and over 188.1 159.8
Personal habits or lifestyle
22
PID3 5/21/04 11:25 AM Page 23
Table 3.5 SMRs for ages 15–64 years (England and Wales) showing trends by social class for specific causes of death.
occupations whilst others may chose ‘sheltered’ with similar smoking habits in the general popula-
occupations because they already suffer mentally tion. Interactions such as this are often very com-
or physically disabling diseases. Some, because of plex and the analysis of observed distributions can
chronic disease, may be unable to keep demanding do no more than indicate possible determinants
jobs in the higher socioeconomic groups; they which merit more detailed and carefully controlled
tend to move down the social scale (social class enquiry. Time, place and personal interactions can
migration). be separated if circumstances arise in which one of
Social class is derived from occupation and status the variables can be kept constant while the others
within an occupational group (i.e. manager, fore- change. For example, comparison of disease fre-
man, unskilled). The concept of social class en- quency in migrant populations with the frequency
compasses income group, education and social in their place of origin is often informative, partic-
status, as well as occupation. Most diseases show a ularly where migrants move from an area with a
positive social class gradient, with a higher inci- high incidence of disease to one with a low inci-
dence in manual workers than in professional dence, or vice versa. When they migrate, they take
groups (Table 3.5). with them their original hereditary susceptibilities
but they change their risk of exposure to harmful
agents. For example, the incidence of cancer of the
Interactions of time, place and
stomach is higher in Japanese living in Japan than
personal characteristics
those living in the USA, while for cancer of the
Frequently, two or more factors correlate with the large bowel the reverse is true. In time, when
incidence of a disease and also with each other. It migrants are assimilated into the host culture, they
may be that only one factor is a causal agent or de- may be exposed to new risks in that culture. Thus,
terminant and that the correlation with a second studies of migrant groups can also be used to meas-
factor is fortuitous. Sometimes, however, two sepa- ure the latent period between exposure and onset
rate causes of disease interact with each other in of disease. For example, the incidence of multiple
such a way that the effect of the two acting togeth- sclerosis is higher in Europeans who migrated to
er in the same individuals is greater than that of ei- South Africa before the age of 15 than in those
ther acting alone. For example, while people who born in South Africa.
work with asbestos and who do not smoke have a It must be stressed that caution is needed in studies
higher incidence of bronchial carcinoma than of migrants because they are self-selected from the
other non-smokers, those who smoke have a much original population and their risks of disease may
higher incidence than would be expected in people have been different from those who did not migrate.
23
PID4 5/21/04 11:27 AM Page 24
Chapter 4
Surveys, survey methods and bias
24
PID4 5/21/04 11:27 AM Page 25
and some diagnostic categories may not be carried where these points are close together, it does not
forward from one revision to the next. matter very much which is chosen, but in the case
In addition the diagnosis may involve a meas- of chronic diseases the intervals may be months or
urement that is not made routinely and/or even years. In such circumstances the reference
recorded for the whole of the population. point must be stated and be consistent.
The above difficulties with routinely available
Example It is extremely difficult to study the epi- data can be partly overcome by well-designed
demiology of hypertension in the community routine information systems. Nevertheless, these
without doing special surveys because the defini- cannot meet all requirements and many of the
tion of hypertension differs from one GP to ano- problems can only be overcome by surveys in
ther. By contrast, birth weight can be studied in which the data and means of collection are speci-
some detail because all newly born babies are fied in advance and in which the study population
weighed and their weight is usually recorded. is clearly defined.
25
PID4 5/21/04 11:27 AM Page 26
12
10
Per cent with diabetes
0
40–44 45–49 50–54 55+
Age
Figure 4.1 The prevalence of dia-
New European Known European betes both known and previously un-
diagnosed in Maori and European
New Maori Known Maori workers. (From Scragg R et al. N Z
Med J 1991; 104: 395–7.)
Sampling
Simple random sample
It is usually unnecessary to study the whole of a
population in order to obtain useful and valid in- In this sample each individual in the parent
formation about that population. The investiga- population has an equal chance (probability) of
tion of a sample has many practical advantages. In being selected. One way of obtaining a random
particular it reduces the number of individuals sample is to give each individual a number and
who have to be interviewed, examined or investi- then to use a computer-generated table of random
gated. It is also often easier to obtain high response numbers to decide which individuals should be
rates and high-quality information on smaller included.
numbers. This is always preferable to poor-quality
data on larger numbers. If a sample is used, it is es-
Systematic sampling
sential to ensure that the individuals included in
the sample are genuinely representative of the This form of sampling is more convenient and is
population being investigated — the ‘parent’ popu- adequate for most purposes. People are selected at
lation. There are many methods available for se- regular intervals from a list of the total population.
26
PID4 5/21/04 11:27 AM Page 27
It has the advantage of being easy for field workers mly selected from all possible groups of the same
to use. type, for example a random sample of all house-
holds in England as in the General Household Sur-
Example If a 1 in 10 sample of school children is re- vey undertaken routinely by ONS. All members of
quired then every 10th child on the school roll the selected groups are included in the study. The
could be included. In some circumstances this underlying assumption is that the individuals be-
method can lead to bias, for example when the longing to any particular group do so for reasons
school roll (or similar list) is compiled by class (or unconnected with the disease being studied and
other grouping), which may affect randomness. the presence of any factor under investigation. The
main advantage of this method of sampling is that
the field work is concentrated and therefore sim-
Stratified sample
pler and cheaper. The principal disadvantage is
In this sample the probability of an individual that diseases and associated factors themselves
being included varies according to a known and may have determined the group to which individ-
predetermined characteristic. The aim of this uals belong, which the investigator may not
method is to ensure that small subgroups which suspect.
are of particular interest to the investigator are
adequately represented.
Multistage sampling
Example Suppose one of the attributes being inves- This combines the above sampling techniques. For
tigated in a cross-sectional study of school children example, a series of ‘clusters’, say schools, might be
is the consequences of being an immigrant to the identified and a random sample of them selected.
country. If immigrants comprise only 5% of the Then within each school, a random sample of
population, then a simple random sample would pupils stratified by class would be recruited to the
produce a group in which 5% are immigrants. Un- study.
less the sample is very large, the number of immi-
grants in the group may be insufficient for a
Bias in sampling
conclusive analysis. To avoid this problem, the
sample has to be weighted in favour of the selec- There are five important potential sources of bias in
tion of immigrant children. This is done by draw- selecting any sample.
ing separate random samples from amongst 1 Any deviation from the rules of selection can de-
immigrant and indigenous children, e.g. 50% of stroy the randomness of the sample. One of the
immigrants and 10% of the indigenous group. most common temptations is to recruit volunteers
Thus, all immigrant and all indigenous children to the study. This is in effect self-selection of par-
have equal chances of selection although the ticipants and such individuals tend to be unrepre-
chance of an immigrant being selected is greater sentative of the parent population.
than the chance of a locally born child being 2 Bias is introduced if people who are hard to iden-
selected. When the data are analysed, the fact that tify in the parent population under study are omit-
the sample was recruited in this way must be taken ted from the study. Thus, in investigating the
into account. health of school children the omission of children
who are persistent absentees may seriously bias
results if the reason for their absence is chronic
Cluster sample
illness.
This involves the use of groups as the sampling 3 The replacement of previously selected individ-
unit rather than individuals (e.g. households, uals by others can easily introduce bias. If, for ex-
school classes or residents within blocks on a grid ample, it proves difficult to trace a person who has
map). The groups to be studied should be rando- been selected or if that person refuses to cooperate,
27
PID4 5/21/04 11:27 AM Page 28
it is not acceptable to replace him or her with an ments (questionnaires, laboratory or other measur-
easily traceable or cooperative individual. Replace- ing equipment) used. Without clarity of definition
ment of a selected individual is acceptable only if it in the design of the study and consistency in its ex-
transpires that they were included in the sample in ecution, errors will occur (see below).
error, for example a selected subject was subse-
quently found not to satisfy study criteria.
4 If large numbers of individuals in the sample re- Error and bias in numerator data
fuse to cooperate in a study, the results may be • Subject variation
meaningless. Therefore, it is essential to make in- • Observer variation
tensive efforts to enlist the cooperation of and • Limitations of the technical methods used
trace all the individuals who have been sampled.
5 If the list of people used as a sampling frame is
out of date, bias will be introduced owing to omis-
Subject variation
sion of recent additions and the inclusion of peo-
ple who have departed. Differences in observations made on the same sub-
ject on different occasions may be due to many fac-
tors, including those outlined below.
Error in rates
• Physiological changes in the parameter ob-
The analysis of epidemiological survey data usual- served, for example blood pressure, blood glucose.
ly entails the calculation of rates, for example inci- • Factors affecting the response to a question, for
dence and prevalence rates, in exposed and example recollection of past events, motivation to
non-exposed population groups. Rates may be af- respond and mood at time of interview, reaction to
fected by errors and bias in either the numerator or environment and rapport with interviewer.
the denominator or both. Such errors can invali- • Induced changes because the subject is aware
date comparisons between rates, and result in mis- that he or she is being studied. (This is sometimes
leading conclusions. referred to as the Hawthorne effect.)
28
PID4 5/21/04 11:27 AM Page 29
• Errors in executing the test or variations in the bias might arise. Some of the more straightforward
phrasing of a question, for example failure to be principles are given below.
consistent in the use of a procedure, carelessness in • The criteria used in diagnostic classification
setting up instruments or reading a scale, failure to must be clearly defined and rigidly adhered to
follow instructions when administering a ques- (even at the risk of missing a few cases). The fea-
tionnaire, omission of some questions or tests, tures that must be present (or absent) for a diagno-
errors in recording of results. sis to be made must be specified.
• Lack of experience or skill, and idiosyncrasies of • Classification of severity or grade of disease
observers, especially when classification depends should be in quantitative terms where possible and
on a subjective assessment, for example misinter- it should cover the full range of possible types of
pretation by the interviewer of an answer to a ques- case.
tion, lack of skill in the manipulation of • All subjects should be observed under similar
instruments, poor motivation, lack of interest in biological conditions on each occasion. Avoid
the project. uncomfortable circumstances. Design simple
• Bias in the execution of the test, for example pre- questions and use check questions for consistency
conception of what is ‘normal’ or ‘to be expected’, of response.
digit preference (i.e. tendency to ‘round off’ read- • The number of observers used should be kept to
ings to whole numbers, 5s and 10s), inflection of a minimum. They should be trained properly to
voice in asking questions. enhance their skills and test their variation on
dummy subjects (or specimens). Take duplicate
readings and record the mean value. Arrange for
Limitations of the technical methods used
the classification to be reassessed by different ob-
Technical methods may give incorrect or mislead- servers, for example independent assessment of
ing results for the following reasons. histopathology specimens by more than one
• The test does not measure what it is intended to pathologist.
measure; for example, the presence of albuminuria • Where possible, subjects and observers should be
in pregnancy, for which there are many causes, is a unaware of (blinded to) the specific hypothesis
poor index on its own of the presence of toxaemia. under investigation in order that they are not in-
Therefore, a study of toxaemia in which cases are fluenced by personal perceptions of the signifi-
identified solely by albuminuria will give mislead- cance of the variables being recorded.
ing results. • The tests selected should be relevant to the pur-
• The method used is intrinsically unreliable or in- pose. Those that give the most consistent results
accurate, and thus yields results that are not re- and are least disturbing to the subject are preferable.
peatable or correspond poorly with those obtained • Equipment should be simple, reliable and easy to
by alternative methods, or do not correlate well use.
with the severity of the condition being measured, • Test methods should be standardized by, for ex-
for example peak flow rate in asthma. ample, the use of standard reagents, sets of graded
• Faults in the test system, for example defective X-rays or slides, standard wording of questions and
instruments, erroneous calibration, poor reagents, instructions on probing and interpretation of
etc. answers, and calibration of instruments against a
standard reference. Quality control should be
maintained to avert ‘drift’ from standards.
Avoidance of numerator error and bias
There are no hard and fast rules that can be applied
Error and bias in denominators
to ensure that errors do not arise in surveys and
that bias is avoided. Each project will require care- Errors occur when the population being investigat-
ful thought and consideration of where errors and ed is not fully defined. Such errors can be mini-
29
PID4 5/21/04 11:27 AM Page 30
30
PID4 5/21/04 11:27 AM Page 31
31
PID5 5/21/04 11:28 AM Page 32
Chapter 5
Cohort studies
32
PID5 5/21/04 11:28 AM Page 33
under investigation. The data on the control group exposed for over 7 weeks clearly have a greatly re-
must be the same and collected in the same way as duced risk of developing the dyscrasia since they
those on the exposed group. have passed through the critical exposure period.
These two types of cohort study can be equally If, in a study designed to assess the risk of agranu-
valuable in epidemiological studies. The choice de- locytosis in patients exposed to chlorpromazine,
pends on the question being studied and the avail- the total number of treatment weeks is used as a de-
ability of suitable study populations. nominator, it will give a distorted indication of the
level of risk. In this case, the definition of exposure
must specify the time period during which the
Time at risk
individual consumed the drug.
In an ideal situation, all members of either type of
cohort are recruited to a study at about the same
time and followed up for the same period of time. Advantages and disadvantages of
Sometimes, it is not possible to recruit sufficient cohort studies
numbers to yield significant results in a short peri-
od, particularly if exposure to the agent or the dis- Advantages
ease being investigated is relatively rare. Moreover,
in most studies, some patients are lost to follow-up. • The main advantage of the cohort study design is
Either situation will result in variations in the that it is possible to distinguish antecedent causes
length of time during which individual members from concurrent associated factors in the aetiology
of cohorts are observed. This gives rise to problems of disease.
in the analysis of the data. • In both types of cohort study, the incidence of
One way to handle variations in the periods dur- disease in exposed and non-exposed groups can be
ing which individuals have been observed is to use determined, allowing the calculation of absolute,
the total time at risk in each group as the denomi- relative and attributable risks (see p. 10).
nator. It is calculated by summing the units of time • It is possible to study several outcomes from ex-
during which each person in the group was ob- posure to the same hazard.
served. It is expressed as the number of units of • Bias in controls is less of a problem than in
‘time at risk’, for example 1 person-year = one indi- case–control studies because the necessary com-
vidual at risk (or observed) for 1 year (or two people parison groups (exposed and non-exposed) are
for half a year each). built into the study design from the start. Even so,
Caution must be exercised in the use of ‘time at it is important to bear in mind that the two groups
risk’ as a denominator. It is only valid if the risk of may not be equally susceptible to the disease under
developing the disease in an individual is not in- study.
fluenced by the period of exposure or the time at • Because the study is prospective, it is possible to
which the exposure occurred. If there is reason to standardize methods, thereby reducing error due
believe that the risk of a disease is affected by the to observer, subject and technical variation (see
length of time an individual is exposed to an agent, p. 28).
the summation of the exposed time within a group
will be misleading. For example, in 1969 Pisciotta
Disadvantages
demonstrated that chlorpromazine can cause
agranulocytosis in some people and that it usually • It is not possible to be certain that supposed aeti-
occurs after 5–7 weeks of continuous exposure (Pis- ological factors are in fact causal. This requires ex-
ciotta AV. JAMA 1969; 208: 1862). It follows that periments of a kind referred to in Chapter 3, which
patients who are exposed for less than 5 weeks do are rarely possible in human populations.
not have the same risk as those exposed for longer, • Even with common diseases, large populations
even though they might be susceptible. Patients are usually required to obtain significant differ-
33
PID5 5/21/04 11:28 AM Page 34
ences in incidence in exposed and non-exposed on changes in smoking habits and other data were
groups. Also, if the incubation period of the disease sent to the male doctors in 1957, 1960, 1972, 1978
is prolonged, the results of the study may be great- and 1990. The fact that all the individuals being
ly delayed. These factors tend to make cohort stud- studied were doctors on the Medical Register aided
ies very expensive in resources. follow-up considerably. Deaths of doctors are noti-
• One of the major difficulties encountered in co- fied to the Medical Register, for reasons quite un-
hort studies is in the follow-up of all subjects. connected with the study, which enabled the
Migration and withdrawal of cooperation may bias investigators to follow up a cohort many years after
the results. It is necessary therefore to build into it was recruited with comparative ease. The first
the study design a system for obtaining basic infor- stage of the analysis was to divide the doctors into
mation on the personal characteristics and out- those exposed to the suspected harmful agent
come of those who cannot be followed up in detail (smokers) and those not exposed (non-smokers).
for the full duration of the study. This allows com- The mortality of the two groups was then
parisons to be made between subjects who are fully compared.
studied and those who are not. In this way, serious The conclusions of the investigators have had
selective bias may be detected and can be allowed far-reaching consequences. An association was
for in the analysis and interpretation of the results. found between smoking and seven different can-
• Finally, even though standard methods and cers, most notably lung cancer, as well as with
diagnostic criteria are adopted, these may change chronic obstructive lung disease, vascular disease,
owing to ‘drift’ over a prolonged follow-up period peptic ulcer and several other fatal diseases. The ex-
and results in later stages may not be comparable cess mortality was almost twice as high in the sec-
with those obtained earlier in the study. ond half of the study as in the first half (Fig. 5.1). It
now seems that about half of all regular smokers
will eventually be killed by their habit. There was a
Examples of cohort studies pronounced correlation between the death rate
from lung cancer and the number of cigarettes
smoked (Fig. 5.2). The data also revealed that the
Mortality in relation to smoking: 40 years’
risk of death from lung cancer fell substantially in
observations on male British doctors (Doll
those who gave up smoking, a benefit which in-
R, Peto R, Wheatley K, Gray R, Sutherland I.
creased with time.
Br Med J 1994; 309: 901–11)
This study yielded two observations that could
The classic study of the effects of smoking amongst not have been made from descriptive studies
British doctors is a good example of a study based alone. Firstly, the sequence of events was clearly
upon a cohort that was used because it was admin- identified — smoking was followed by lung cancer,
istratively easy to identify and follow up. It in- and secondly, a dose–response effect was demon-
volved the use of an internal control group. In strated. Both of these findings weigh heavily in
1951, the research team sent a simple question- favour of the causal hypothesis. However, it should
naire to all of the 59 600 doctors whose names were be remembered that the investigation was stimu-
on the Medical Register of the UK at the time. The lated by the results of descriptive studies which
questionnaire enquired about their past and cur- showed a correlation between mortality from lung
rent smoking habits. Over 34 000 (69%) of the cancer and sales of cigarettes in England and
male doctors and more than 6000 (60%) of the Wales.
female doctors who were contacted completed The problem with a cohort recruited in this way
the questionnaire. The responding doctors were is that if it is used to study the effects of an agent or
divided according to their past and current smoking factor which is very rare, or if the disease is a rare
habits and their subsequent mortality was record- consequence of exposure, then the size of the co-
ed. Further questionnaires to obtain information hort has to be very large in order to yield sufficient
34
PID5 5/21/04 11:28 AM Page 35
100 1951–71
100
80
60
Per cent alive
40
Key
1–14 Cigarettes/day
20 15–24 Cigarettes/day
Figure 5.2 Overall survival after age ≥ 25 Cigarettes/day
35 years among cigarette smokers and Never smoked regularly
non-smokers: life-table estimates,
0
based on specific death rates for the 40 55 70 85 100
entire 40-year period. (From Doll et al. Age
1994.)
35
PID5 5/21/04 11:28 AM Page 36
numbers of cases to detect a significant difference Table 5.1 Average incidence of confirmed leukaemia in
between the risks in the exposed group and the residents of Hiroshima and Nagasaki (1947–58) by city of
non-exposed group. exposure and distance from epicentre.
36
PID5 5/21/04 11:28 AM Page 37
cooperate. These men were asked to complete a events (defined as fatal IHD or myocardial infarc-
questionnaire which included questions on occu- tion). The crude attack rates and the attack rates ad-
pation, smoking habits and indicators of heart dis- justed for the risk factors set out above are shown
ease. They were also examined by a research nurse. in Table 5.3.
Ninety-nine per cent of the men were followed up The results indicate that, after taking account
for an average of 6 years. of the differences in smoking habits, systolic
The data were analysed using a multiple logistic blood pressure, serum cholesterol and age
regression model in order to adjust simultaneously between the social class groups, there remains
the incidence rates of major IHD events for smok- an unexplained higher incidence of major
ing, systolic blood pressure, serum cholesterol, age IHD events amongst men in manual occupa-
and social class. tions compared to those in non-manual
Of these men, 336 experienced major IHD occupations.
37
PID6 5/21/04 11:29 AM Page 38
Chapter 6
Case–control studies
38
PID6 5/21/04 11:29 AM Page 39
blocking agents or 5-alpha reductase inhibitors. • People registered with the same general practice.
Those who fail to respond may be referred to • Students at the same school or educational
hospital for catheterization or surgery. For institute.
conditions such as this the general practice is • Workers in the same factory or office.
often the most appropriate source of cases for an • People on the same electoral or other population
investigation. register.
For most studies it is desirable to recruit incident • Patients being treated in the same hospital.
cases (cases newly presenting with the disease) • People randomly selected from the community
rather than prevalent cases (cases with established in which the case lives. This can be achieved by, for
disease). The problem with prevalent cases is that example, random digit dialling. Using this
they represent the survivors from an earlier cohort method, controls that fail to meet basic recruit-
of incident cases. Thus, in a study of the factors ment criteria are discarded after a few key ques-
that lead to end-stage renal failure, it would be tions and the remainder are included in the
inappropriate to recruit the cases for study from investigation.
amongst patients undergoing regular dialysis at • Hospital patients, either all attenders or, more
the time of the investigation. Patients who have usually, those with conditions believed to be unre-
had regular dialysis for, say, 1 year will be the sur- lated to the factors under study. The main limita-
vivors of a larger group who started dialysis a year tion of using hospital patients as controls in any
before. The survivors are unlikely to be representa- study is that, even though they may not have the
tive of all of those who started treatment and there- disease being investigated, they are unlikely to be a
fore the findings from any study of them might random sample of the general population from
have limited value. which the cases are drawn. For example, even
Even when incident cases only are recruited to a though a hospital patient does not have the disease
study the stage of the disease at which they present under investigation he or she may have another
will vary. This should be taken account of during disease caused by the same agent or whose pres-
the analysis of the data. For example, women may ence could have affected exposure to the causal
first present with breast cancer at any stage, includ- agent of the disease under investigation. Moreover,
ing metastatic disease. people who live in poor social environments are
more likely to be admitted to hospital than those
who live in better circumstances and their use as
Selection of controls
controls may introduce a social class bias.
Control subjects are essential in order to establish Controls may be selected as a group or selected
the frequency with which the suspected causal for each of the cases as they arise. Thus, it is legiti-
agents or determinants occur in people who do not mate to recruit, say, 50 cases from a particular com-
have the disease under investigation. Control sub- munity and the same number or more controls
jects must not have the disease being investigated from the same community. It is also legitimate to
at the time the case arose (index day). Controls recruit one or more controls per case as they arise
must be a representative sample of the population without reference to any personal characteristic. In
from which the cases were recruited and thus are at a matched case–control study the controls are re-
similar risk of having been exposed to the suspec- cruited for each case on the basis that they share
ted agent. Once selected, controls should be neither one or more characteristics.
discarded nor replaced for any reason other than In many ways the matched case–control study is
that they fail to meet the selection criteria; for more efficient; however, it will not provide any in-
example, if they were mistakenly drawn from formation on the variable or variables used for
another population. matching. Thus, in an investigation of the effect of
Depending on the condition being investigated oral contraceptives on the risk of venous thrombo-
controls could be selected from amongst: sis it might be appropriate to match the controls to
39
PID6 5/21/04 11:29 AM Page 40
the cases on age. When doing this the study will causal agent and those not exposed. By contrast,
provide no information on the effect of age on risk. case–control studies only provide data from
Likewise a study could be designed to identify the which the rate of exposure to suspected harmful
risk factors for acute myocardial infarction where agents in diseased and non-diseased individuals
the controls are matched to the cases by age, gen- can be calculated. This means that neither the
der and body weight. Such a study would provide absolute risk nor the attributable risk nor the
no information on the effect or age, gender or body precise relative risk resulting from exposure can
weight on the risk of acute myocardial infarction. be calculated. The difficulty is shown schematically
This is acceptable if the effects of the matched vari- in Tables 6.1 and 6.2.
ables are well known and therefore of little imme- In a cohort study, the subjects studied are all
diate interest. those exposed (A + C) and all those not exposed (B
It is important to note that error (or bias) in the + D) to the suspected causal agent (Table 6.1). The
selection of controls will have exactly the same subjects subsequently reveal themselves as dis-
effect on the outcome of the study as bias in the eased or non-diseased within these categories.
selection of cases. In case–control studies (and in There is therefore no difficulty in calculating the
cohort studies that involve selection of external disease rate in the total population (A + B/A + B + C +
controls) as much attention must be given to the D) or in the exposed persons (A/(A + C)) and those
identification of, and collection of data from, the not exposed (B/(B + D)). The subsequent calcula-
control subjects as is given to the cases. tion of relative risk (RR) and attributable risk (AR)
presents no problem:
40
PID6 5/21/04 11:29 AM Page 41
derive the total numbers of cases in the population Thus the increase in risk of heart disease associated
who were exposed and not exposed because a with alcohol may be the result of smoking.
case–control study is not based upon a known pro- There are two ways to take account of confound-
portion of the population in either category. Con- ing variables.
sequently, neither the incidence rate in the 1 Analysis of subsets of the data (partitioning) defined
population as a whole nor the incidence rate by the confounding variable. This can be illustrated
amongst the people exposed to the suspected by considering a study of the effect of age at first
harmful agent can be derived. It follows from this birth on women’s risk of carcinoma of the breast.
that risk cannot be calculated directly. However, an Women who have their first child whilst young
approximation of the relative risk can be derived tend to have more children than women whose
from case–control data. This approximation, al- first child is born late in their reproductive life. It
though often referred to as the relative risk, is more follows that if there is a statistical association be-
correctly termed the odds ratio (OR). It is calcul- tween age at first birth and the risk of breast cancer
ated as follows. Using the notation in Table 6.1 the it is likely that there will also be an association be-
true relative risk is: tween family size and risk of breast cancer. The two
effects can be separated by restricting the analysis
A ¥ ( B + D)
to women who have had only one child (thereby
A ¥ ( A + C)
separating out the effects of parity) and calculating
In the case of most diseases, the proportion of the the risk according to age at first pregnancy. Alter-
population who are affected, whether or not they natively the analysis can be restricted to women
are exposed to the suspected causal agent, is small. who had their first child at a given age and the risks
Thus, A is small in relation to C; likewise B is small calculated according to parity. The disadvantage of
in relation to D. It follows that D will approximate this technique is that not all the available data can
to B + D and C will approximate to A + C. The be used in the critical analyses.
approximation to the relative risk, the OR, then 2 Use of a multivariate analysis technique to adjust the
becomes: relative risk for the effect of confounding variables. The
most usual statistical model is logistic regression.
A¥D a¥d
= An advantage of this method is that it allows
B¥C b¥c
simultaneous adjustment for the effects of more
This approximation to relative risk is used in all than one confounding variable. Similar methods
case–control studies but it is only valid if the inci- are used to adjust for the effects of confounding
dence of the disease is low. In most circumstances variables in cohort studies.
it is not possible to calculate attributable risk from
a case–control study.
Effects of high incidence
of exposure
Confounding variables
Essentially, the success of a case–control study is
A confounding variable is a characteristic or expo- dependent upon there being a significant differ-
sure that is associated with both the exposure ence between the proportion of cases exposed to
being investigated and the outcome (disease). For the suspect agent and the number of controls so
example, in an investigation of the association be- exposed. If the incidence of exposure is very high,
tween alcohol consumption and the risk of heart it may be impossible to demonstrate such a differ-
disease it is likely that amongst the cases there ence. Consider an extreme example of a case–con-
would be a greater proportion of smokers amongst trol study designed to identify the possible causal
the cases than amongst the controls. This is be- agents of carcinoma of the bronchus which is con-
cause smoking is correlated with alcohol consump- ducted in a population where the prevalence of cig-
tion and smoking is associated with heart disease. arette smoking over the age of 15 years is 100%. In
41
PID6 5/21/04 11:29 AM Page 42
such a situation there can be no difference between people who choose to smoke are also constitu-
the proportion of cases and controls exposed to tionally predisposed to lung cancer or are exposed
cigarettes. It follows that smoking could not be re- to another noxious agent more often than are
vealed as a risk factor. It does not follow that no risk non-smokers. This problem is less conspicuous
factors will be revealed by such a study, but those when dealing with highly specific agents such
identified may be associated rather than causal and as microorganisms or in situations where the
the principal cause will be missed. time between exposure and onset of symptoms is
short.
Advantages and disadvantages of • There are sometimes difficulties in selecting and
case–control studies recruiting appropriate controls. This is important
because the value of the results obtained from a
case–control study is as dependent upon the prop-
Advantages er selection of ‘controls’ as it is on the identifica-
Despite the approximations that have to be made tion of affected individuals.
in the analysis of case–control studies, they do • Because case–control studies are not based on de-
have some important advantages over cohort fined populations, the incidence of the disease
studies. within that population cannot be calculated from
• By concentrating effort on the identification of the study.
affected individuals and recruiting controls from
the unaffected population, the number of subjects
required to obtain significant results is kept to a Examples of case–control studies
minimum.
• Results can be obtained relatively quickly be- Sexual activity, contraceptive method,
cause the investigation does not have to wait for genital infections and cervical cancer
the disease to develop, as it does in cohort studies. (Slattery M, Overall JC, Abbott TM et al. Am J
This means that it is a relatively inexpensive type Epidemiol 1989; 130: 248–58)
of study.
It has been suggested that cervical cancer is a sexu-
ally transmitted disease. Between 1984 and 1987 a
Disadvantages
case–control study was carried out in Utah, USA,
• Case–control studies generally rely upon retro- where a high proportion of the population are
spective data, which have their own inherent prob- active members of the Church of Jesus Christ of
lems. The ability of individuals to recall past events Latter Day Saints (Mormons). The study was de-
tends to be unreliable due to a tendency for mem- signed to explore the relationship between cervical
ory to be selective. Records of past events may be cancer and sexual activity, the use of barrier meth-
incomplete in respect of variables that are the sub- ods of contraception and certain types of genital
ject of investigation, and the ways in which the rel- infection. The subjects were women aged 20–59
evant observations and measurements were made years, newly diagnosed with cervical cancer. Con-
are not usually standardized. This gives rise to trols were identified by use of a random digit di-
uncertainty regarding their validity. alling telephone sampling technique. They were
• Because the data are collected after the event matched to cases by 5-year age intervals and coun-
(retrospectively) it is difficult to be sure whether a ty of residence. Interviews were completed for 266
demonstrable correlation is causal or not. Thus, women with histologically confirmed carcinoma
the finding that a history of cigarette smoking is in situ or invasive squamous cell cervical cancer
common amongst individuals with lung cancer and for 408 matched controls.
does not prove that the former preceded and After adjustment for age, education, church at-
caused the latter. Alternative explanations are that tendance and cigarette smoking, by means of mul-
42
PID6 5/21/04 11:29 AM Page 43
tiple logistic regression models, several significant impractical and unethical to conduct a large-scale
risk factors were identified. These included multi- randomized control trial to test the validity of this
ple sexual partners, current mate having multiple suggestion. Therefore, a case–control study was set
sexual partners, reported Trichomonas infection up which aimed to identify all children admitted
and serological evidence of herpesvirus type 2 in- to hospitals in the UK with serious acute neurolog-
fection (Table 6.3). It should be noted that there is ical illnesses of the types which it was suggested
a pronounced gradient of risk relating to increased could be caused by the vaccine and lead to perma-
numbers of partners and increased numbers of nent brain damage. For each case child reported,
partners of the mate of the woman. two control children, matched for age and sex,
A protective effect was noted from use of foam or were selected from those living in the same local
jelly as a contraceptive method (OR = 0.44), and area. The past histories of immunization and of
from the use of diaphragms (OR = 0.67) or con- other possible predisposing or aetiological factors
doms (OR = 0.53) in women who reported more were obtained for both case and control children in
than one sexual partner. These data lend support to identical manners. Of 904 cases of encephalopathy
the hypothesis that cervical cancer is due to a sex- and severe convulsions reported, 30 (3.3%) had re-
ually transmitted agent. ceived pertussis vaccine within 7 days before be-
coming ill, compared with 23 (1.3%) of 1783
control children immunized within 7 days before a
Pertussis immunization and serious acute
defined reference date, which was a significant dif-
neurological illnesses in children (Miller DL,
ference (OR = 3.3) (Table 6.4). The children were
Madge N, Diamond J, Wadsworth J, Ross EM.
followed up a decade later to determine the late
Br Med J 1993; 307: 1171–6)
outcome of their illnesses. Nearly two-thirds either
In 1975, widespread public alarm was created by had died or were suffering from significant neuro-
the suggestion that whooping cough vaccine logical dysfunction. Of 367 such children, 12 cases
might cause severe encephalopathic illnesses fol- (3.3%) were pertussis vaccine associated compared
lowed by permanent brain damage in a small but with 6 (0.8%) of 723 controls, which gives an OR of
significant number of children. It would have been 5.5. Thus, the study showed that there is a small
43
PID6 5/21/04 11:29 AM Page 44
but definite risk of serious acute neurological ill- hazards appeared to be the leather, or adhesives
nesses after whooping cough vaccine, though the used, or both.
risk was much smaller than some workers had sug-
gested from totally uncontrolled series of cases. It
Results of case–control study of leukaemia
was also clear that children who suffered from such
and lymphoma among young people near
illnesses often died or had significant long-term
Sellafield nuclear plant in West Cumbria
sequelae, though the number of such cases associ-
(Gardner MJ, Snee MP, Hall AJ, Powell CA,
ated with recent pertussis immunization was too
Downes S, Terrell JD. Br Med J 1990; 300: 423–9)
small to be certain that the vaccine was on its own
responsible for cases of permanent brain damage. Concern about levels of childhood cancers around
This study illustrates the difficulty of identifying nuclear installations and a consequent public en-
aetiological factors in extremely rare conditions. quiry led to several studies being set up. One was a
case–control study of leukaemia and lymphoma
among young people living in the vicinity of the
Perinatal deaths and maternal occupation
Sellafield nuclear plant in West Cumbria. Its aims
(Clarke M, Mason ES. Br Med J 1985; 290: 1235–7)
were to explore whether known causes or factors
Reproductive hazards are thought to exist in many associated with the nuclear site were responsible
industries. In order to explore this problem, a for the apparent excess. A total of 74 cases of
case–control study of perinatal death occurring in leukaemia and lymphoma among people born in
Leicestershire was carried out between 1976 and West Cumbria and diagnosed there at ages under
1982. Case notes were reviewed and the mothers 25 between 1950 and 1985 was identified. Risk fac-
were interviewed in all 1187 cases of perinatal tors in cases were compared with those in up to
death during this period. The control for each case eight controls matched by date of birth and sex, se-
was selected as the next live birth occurring at the lected from the same birth register as their respec-
place or intended place of delivery. All maternal tive cases. The expected association with prenatal
and paternal occupations and industries were exposure to X-rays was found, but the main find-
recorded at the interview with the mother. A total ing was of significantly raised relative risks in chil-
of 671 mothers were employed outside the home dren born near Sellafield and in children whose
at some time during pregnancy. An analysis of fathers were employed at the plant (RR = 2.4), par-
maternal occupations showed that the OR for the ticularly those fathers with high radiation dose
risk of perinatal death was exceptionally high in recordings before the child’s conception (RR = 6.4).
women employed in the leather industry (OR = 2.1 At the time, no other satisfactory explanation was
after adjustment for social class). A similar excess put forward and it was concluded that ionizing
was found in all towns within the county where radiation may be leukaemogenic to offspring. This
shoe manufacture took place. No other risk factor interpretation has been subsequently challenged
was found to account for this observation. Possible in the scientific literature.
44
PID7 5/21/04 11:30 AM Page 45
Chapter 7
Intervention studies
45
PID7 5/21/04 11:30 AM Page 46
• Individual
• Cluster: of whole groups or communities Since willingness to cooperate may not be random-
• Stratified: random selection within specified sub- ly distributed in the population, allocation should
groups be deferred until agreement to participate has been
obtained.
To avoid bias in reporting illnesses and other
possible behavioural differences, subjects should
not know to which group they have been alloca-
Individual allocation
ted. In drug and vaccine trials, this often entails the
The allocation of individuals to test or control use of a placebo treatment for controls which must
groups must be random. Other methods of assign- be presented in an identical form to the active
ment to treatment groups are to be avoided. For ex- treatment. In the case of some procedures, for ex-
ample, alternate allocation to test and control may ample provision of different types of service, blind
enable the patient or the person who assesses the allocation is not possible. A trial in which neither
outcome to guess the group to which the subject the subject nor the people assessing outcome know
has been allocated. whether the subject is receiving active treatment or
not (or which of two different treatments is being
given) is called a double-blind trial.
Cluster allocation
For practical reasons, allocation is sometimes made
Outcome
of whole groups or communities. This is because,
for example in trials of a vaccine, the spread of in- The outcome to be assessed must be specified in
fection may be inhibited in unvaccinated people if advance. It should be expressed in terms of advant-
a proportion of the population is protected, there- age to the patient or to the community, for example
by obscuring the benefit derived from vaccination. reduced incidence or severity of disease or cost to
Similarly, in recent trials of preventive advice the health service. Assessment criteria should be
against coronary heart disease, the test and control clearly defined, consistently applied and reliably
groups were workers in randomly allocated facto- recorded in order to minimize bias in the measure-
ries, in order to minimize ‘contamination’ of the ment of outcome. Misclassified cases in either test
control group with advice offered to the interven- or control group will reduce the size of difference
tion group. between them in the incidence of disease and thus
46
PID7 5/21/04 11:30 AM Page 47
47
PID7 5/21/04 11:30 AM Page 48
Table 7.1 Mild hypertension: main events in treatment and control groups.
48
PID7 5/25/04 2:21 PM Page 49
tion of individuals to treatment and control groups confidently advised to give up smoking in the ex-
would be inappropriate and unethical and, in such pectation that their health and prognosis will im-
circumstances, though less than ideal, a before- prove. However, ex-smokers are not a random
and-after intervention assessment is often used. sample of former smokers and their reasons for giv-
This study reported on the results of a campaign to ing up may be related to other factors which influ-
promote the use of vitamin D supplementation in ence their risk of developing smoking-related
Glasgow. In a precampaign survey, blood samples diseases. Nor is it certain how effective antismok-
were obtained from 189 children aged 5–17 years ing advice is in influencing smoking behaviour.
and those with biochemical evidence of rickets had Therefore, in 1968 the authors set up an RCT of
an X-ray examination of the knees. In postcam- antismoking advice in 1445 male smokers, aged
paign surveys, 2 and 3 years later, 255 children 40–59 years, at high risk of developing cardiorespi-
were similarly examined. On both occasions the ratory disease. They were allocated at random to an
children were asked about their frequency of con- ‘intervention’ group who were given individual
sumption of vitamin D supplements (in younger advice on the relationship of smoking to health
children this was checked with mothers). The re- and challenged to consider their situation. Those
sults showed a striking reduction in the prevalence who declared a wish to stop smoking were given
of rickets in children who took regular or even in- support and encouragement for an average of four
termittent vitamin D supplements, and the num- further visits over 12 months. The ‘control’ group
ber of hospital discharges of Asian children with were given no specific advice. All subjects complet-
rickets in Glasgow declined rapidly after the start ed a questionnaire 1, 3 and 9 years later. Deaths in
of the campaign. the group were monitored. After 1 year, the report-
Clearly, the decline in rickets could have ed cigarette consumption in the intervention
been due to factors other than the official vitamin group was one-quarter of that in the control group
D supplement campaign, for example increas- and over 10 years the net reported reduction aver-
ing adoption of a Western diet and lifestyle. aged 53%. However, the ‘normal-care’ group also
However, the time and place reduction in rickets reduced their consumption, reflecting a general de-
prevalence, backed by objective measures, lends cline in smoking in the population, thereby reduc-
support to an assessment of the effectiveness of the ing the apparent benefit of smoking cessation in
campaign. the intervention group over the ensuing years.
Over the first 10 years, the intervention group ex-
perienced fewer respiratory symptoms and less loss
RCT of antismoking advice: final 20 years’
of ventilatory function; their mortality from coro-
results (Rose G, Colwell L. J Epidemiol Community
nary heart disease was 18% lower than controls,
Health 1992; 46: 75–7)
and for lung cancer it was 23% lower. No further
Many studies have shown that the mortality and contact with subjects to determine changes in
morbidity of ex-smokers is less than that of those smoking habits has been attempted, but follow-up
who continue to smoke. On this basis, smokers are has been continued for a further 10 years based on
0.28
- - 13/300
- + 8/302 } 21/602 (3.5%)
49
PID7 5/21/04 11:30 AM Page 50
death certificates and cancer registrations. Com- credence by the results of two intervention studies,
paring the intervention with the normal-care both of which had methodological weaknesses. A
group, total mortality was 7% lower, fatal coronary large RCT was needed to resolve the matter. The
heart disease was 13% lower and lung cancer cases trial was conducted in 33 centres in seven coun-
(deaths and registrations) were 11% lower. It was tries amongst 1817 women known to be at high
concluded that the policy of encouraging smokers risk through having had a previous affected preg-
to give up the habit was worthwhile and should nancy. They were allocated at random to one of
not be changed. It was estimated that out of every four groups who received supplementation with
100 men who stopped smoking, between 6 and 10 folic acid and/or other vitamins or none. Of 27 ba-
were in consequence alive 20 years later. bies born to these women with a neural tube de-
fect, six were in the group who received folic acid
supplementation and 21 in the other two groups —
Prevention of neural tube defects: results
a 72% protective effect (RR = 0.28) (Table 7.2). The
of the MRC Vitamin Study (MRC Vitamin
other vitamins showed no benefit. It was conclud-
Study Research Group. Lancet 1991; 338:
ed that folic acid supplementation starting before
131–7)
pregnancy can now be firmly recommended for all
It has long been suspected that diet has a role in the women who have had an affected pregnancy.
causation of neural tube defects, one of the com- There are also grounds for public health action to
monest severe congenital malformations. The pos- ensure that the diet of all women who may bear
sibility that supplementation with folic acid or children contains an adequate amount of folic
other vitamins might reduce the risks was given acid.
50
PID8 5/21/04 11:33 AM Page 51
Chapter 8
Health information and sources
of data
51
PID8 5/21/04 11:33 AM Page 52
taken midway between the full decennial censuses. ducting deaths and emigrants from numbers
The census is conducted by the Office of National recorded in the census, and adding births and im-
Statistics for England and Wales, the General migrants. At the same time, the age distribution of
Register Office in Scotland and the Northern Ire- the people remaining is adjusted. These are known
land Statistics and Research Agency. Each census is as intercensal estimates. Unfortunately errors occur
undertaken only with the specific authority of Par- which are compounded by the passage of time. The
liament. Any individual who refuses to cooperate principal sources of error in the intercensal esti-
is liable to prosecution. mates arise through inadequate recording of immi-
The precise information that is collected varies gration and emigration in terms of numbers, age
from census to census but it invariably includes and sex. Furthermore, there is no system for ascer-
age, gender, marital state, place of birth, occupa- taining the amount of internal migration (changes
tion, number of children, usual place of residence of residence within the country). Thus, the greater
and duration of present residence. In addition, the the time that has elapsed since a census, the less
head of the household has to furnish details of the the precision of the estimate, especially estimates
residence including its type, tenure, accommoda- relating to small areas within the country. After a
tion and facilities. In recent years, it has been the census, the figures for years since the last census are
practice to ask for additional information from a recalculated, taking account of the information
sample of the population. All the information re- provided by the new census. These are called post-
lating to individuals is confidential, even within censal estimates.
government departments.
Before census day, officials deliver the appropri-
ate census form to each household and residential Population projections
institution in the country. They are collected by For planning purposes, it is often essential to have
the same official after census day. Census officers some idea of the likely size and composition of the
are available to help householders with any prob- population in years to come. The essential differ-
lems they may encounter. The data on the forms ence between population estimates and popula-
are analysed centrally. In the past, tabulations of tion projections is that an estimate is based on
census data have been published as books, some of knowledge of the births, deaths and migration that
the more detailed information only after a delay of have happened, and a projection is based upon
several years because of the time required for what is thought likely to happen. Therefore, as-
analysis and printing. Since 1981 census material sumptions have to be made about trends in mor-
has been available both as books and on computer- tality, birth rates and migration. These are arrived
readable media. Much of the material from the at by extrapolation of past trends. Unforeseen
2001 census is available on the web. Despite some changes in, for example, fertility can invalidate the
problems arising from concealment or misreport- projections.
ing of census information, and slight under-
recording because some people are not at a formal
address on census night, modern censuses are Vital events
regarded as being generally very accurate. Infor-
mation on the 2001 census is available at
General
www.census.ac.uk.
Since the early 19th century, there has been a statu-
tory requirement for all births, deaths and mar-
Estimates of population
riages in the UK to be registered. Before the Births,
between censuses
Marriages and Deaths Act (1839) most of the
The size and demographic characteristics of the records that existed were kept by the ecclesiastical
population in non-census years is estimated by de- authorities. Since the vast majority of people at
52
PID8 5/21/04 11:33 AM Page 53
53
PID8 5/21/04 11:33 AM Page 54
Figure 8.1 The General Bills of Mortality for London, 1641 and 1665.
54
PID8 5/21/04 11:33 AM Page 55
details of how far the death was investigated, the Death registration data are collated and analysed
doctor is required to state the ‘immediate cause’ of by the Office of National Statistics. The causes of
death. There is then space provided for him or her death that are analysed are normally those given as
to record the ‘antecedent causes’ (giving the ‘un- the ‘underlying cause’ rather than the immediate
derlying cause’ last) and any other significant con- cause because the former is more informative and
ditions that may have contributed to the death. As more useful for the study of disease in the commu-
far as possible, the doctor should use generally ac- nity. In the first example given above, the death
cepted terminology, such as that set out in the ICD. would be classified as due to carcinoma of the
The Registrar requires him or her to avoid the use bronchus for purposes of statistical analysis. The
of indefinite and ambiguous terms such as ‘heart tables published by the ONS must be interpreted
failure’ or ‘old age’. The completion of the certifi- with this rule in mind. They do not necessarily pro-
cate is quite straightforward in the case of an indi- vide a complete picture of mortality attributable in
vidual who has died as a result of a well-defined whole or part to specific causes.
disease that has been extensively investigated in In certain circumstances, a normal death certifi-
life, for example death by bronchopneumonia due cate cannot be issued. These are when there was no
to carcinomatosis due to carcinoma of the medical attendant during the last illness of the de-
bronchus, with chronic bronchitis as a significant ceased, when it is suspected that the death resulted
condition that contributed to death. However, in from unnatural causes, or when the death occurred
many circumstances the death certificate cannot before full recovery from a surgical operation or
be completed with such precision, for example in the administration of an anaesthetic. In such cir-
the case of an old person who has previously had a cumstances, the death must be reported to the
stroke, has diabetes, has chronic cardiac failure, is Coroner by either the attending doctor, or the
known to have bronchitis, has been bedridden for police or the Registrar. A coroner is a member of
months and who is found dead in bed one morn- the judiciary and is bound by legal processes. He or
ing. In such cases, the certified cause of death is an she has to be legally qualified but not necessarily
arbitrary opinion rather than a statement of fact. medically qualified, though some have both quali-
Generally, the precision of death certification fications. The Coroner investigates the death by
tends to diminish with increasing age of the enquiry, either directly or through his or her offi-
deceased. cers. The Coroner may order a postmortem exami-
An informant, who is usually a close relative nation and may hold an inquest, with or without a
of the deceased, a person present at death, the jury. Having established the cause of death to his or
person in charge of the institution in which her satisfaction, the Coroner will then sign a death
the person died or the person responsible for the certificate. If the Coroner has reason to believe that
disposal of the body, must register the death death was caused by the unlawful action of anoth-
with the Registrar as soon after death as possible. er person, he or she is bound to forward the papers
When doing so, he or she must give the following to the Director of Public Prosecutions. It should be
information: noted that in these circumstances it is the Coro-
• Date and place of death. ner’s job to establish the cause of death, not who
• Full name and sex of deceased. caused it.
• Maiden name of married woman. If those responsible for the disposal of the body
• Date and place of birth of deceased. wish the deceased to be cremated, an additional
• Occupation and usual address of deceased. certificate is required. The person wishing to have
The data above are recorded in the register. If the the body cremated has to complete part of a form.
Registrar is satisfied that the particulars are in order The practitioner who attended the deceased during
and that there is no need to report the death to the the last illness completes another part. This part
Coroner, he or she will issue a death certificate and has certain similarities to a certificate of cause of
authority for burial. death but the doctor must have inspected the body
55
PID8 5/21/04 11:33 AM Page 56
Figure 8.2 Death certificate (England and Wales). (Reproduced with permission of the ONS (Crown copyright).)
56
PID8 5/21/04 11:33 AM Page 57
Figure 8.2
Continued
further improvements. The commonest causes of product of conception before it has reached an age
neonatal death include congenital abnormalities when it could be expected to have an independent
and prematurity. Many of these deaths would seem life and shows no signs of life at birth. The lower
to be unavoidable, although congenital rubella is limit of fetal viability is defined as 24 weeks’
one example which it is hoped will be completely gestation.
eliminated through immunization of children. Since the Abortion Act (1967) came into force, it
Around 40% of postneonatal deaths are due to has been permissible for a pregnancy to be termi-
cot death (sometimes called ‘sudden infant death nated provided it has not progressed beyond 28
syndrome’). Associations have been shown with weeks’ gestation, if two doctors believe that the
maternal smoking, prone sleeping position, bottle continuation of the pregnancy would be injurious
feeding and season of the year. A campaign to to the physical or mental health of the woman or
encourage mothers to place their baby on their that there is a risk that the child may be born with
side or back rather than prone has led to a reduc- a disability that would prevent it from leading a
tion in the number of cot deaths. The effectiveness normal life. Under 1990 legislation, abortion is
of stopping smoking and encouraging breast feed- normally permissible only up to 24 weeks’ gesta-
ing in reducing the number of deaths from cot tion. When a termination of pregnancy is carried
death has yet to be shown. A small but increasing out under the provision of the Act (it is illegal to
proportion of deaths in the postneonatal period terminate a pregnancy other than for reasons set
are due to congenital abnormalities and condi- out in the Act) the doctors involved have a statut-
tions originating in the perinatal period, suggest- ory obligation to notify the Department of Health
ing that some infants that previously died soon (DoH). The form of notification asks for the name,
after birth are now living until the postneonatal date of birth and marital status of the woman, her
period. normal place of residence, and the number of pre-
vious pregnancies, distinguishing those that pro-
ceeded to term from those that were terminated.
Abortions
The presumed duration of the pregnancy, the
A spontaneous abortion is the expulsion of the statutory grounds for the operation and the place
57
PID8 5/21/04 11:33 AM Page 58
where it was carried out are also required. The in most people are minor, self-limiting conditions
forms are sent in confidence to the DoH where for which there is no specific treatment. Diagnostic
they are checked to ensure that the law is not being precision is unnecessary and it is a waste of time to
abused. They are collated and analysed by the attempt to discriminate between the many causes
Office of National Statistics, which publishes by complicated and expensive viral studies and
annual tabulations setting out the number of other examinations. In such circumstances the data
abortions by different criteria. that are generated may not have sufficient precision
for epidemiological studies.
In many cases, the stage at which disease is treat-
Morbidity ed depends on a complex series of factors other
than the patient’s perception of the problem.
These include the availability of health service
General
treatments, for example waiting lists, outpatient
Morbidity statistics are concerned with the appointment availability and the acceptability of
amount and types of illness that occur in the com- the treatments that the patient believes will be
munity. The sources of available data vary from offered. This will affect the morbidity recorded at
place to place and from time to time. They include, hospitals and employment sickness absence figures.
for example, attendances for primary care, hospital Finally, there is a quite separate problem in the
outpatients and admissions, as well as statutory way morbidity statistics are calculated and present-
sources and special registers for particular condi- ed. The calculation of mortality rates is relatively
tions. Most routinely collected morbidity data suf- straightforward because each individual can only
fer from serious shortcomings partly because of the die once. Thus, if there are 10 deaths in a popula-
ephemeral nature and imprecise diagnosis of many tion of 162, the death rate is 61.7 per 1000. If, how-
illnesses and partly because of inadequacies in the ever, 10 episodes of an illness occur amongst 162
information systems. Consequently, although people during a year it does not mean that 61.7 per
they should give a more complete picture of the in- 1000 population were ill — one individual may
cidence of disease in communities than mortality have had more than one episode of illness; indeed,
data, they do so with varying reliability and must all the episodes may have occurred in the same in-
be interpreted with caution. dividual. Many morbidity statistics are collected in
One of the principal problems centres around such a way that it is impossible to distinguish
the definition of illness itself. For some people, a episodes of illness from sick individuals. When
common cold or backache may represent an ‘ill- presenting or making use of rates it is important to
ness’ and justify them seeking medical help or be clear how the rate was derived. Morbidity statis-
being away from work. These people’s illnesses tics routinely available in England and Wales in-
may be recorded in one of the many routine data clude the following.
systems. For other people, symptoms that the med-
ical profession would regard as indicative or diag-
nostic of major disease may be regarded as having Morbidity statistics
no serious significance, an inconvenience to be tol- • Hospital episode statistics
erated until normal recovery takes place. Such ill- • General practice databases
nesses will not feature in any morbidity statistics • Infectious diseases
because those affected do not seek medical aid nor • Notification of episodes of STDs
allow the symptoms to alter their lifestyle. • Notification of ‘prescribed’ and other industrial dis-
Another problem is that diagnostic precision ease and accidents
• Notification of congenital malformations
varies between doctors according to their percep-
• Cancer registration
tion of the disease that they are treating. For exam- • Laboratory reports on infections
ple, influenza and upper respiratory viral infections
58
PID8 5/21/04 11:33 AM Page 59
59
PID8 5/21/04 11:33 AM Page 60
60
PID8 5/21/04 11:33 AM Page 61
Wales in 1961 after an episode in which thalido- main purposes are to provide accurate knowledge
mide was responsible for a major outbreak of limb of the incidence and prevalence of disease in a
deformities in the children of mothers who had community which will assist in the organization
taken the drug during early pregnancy. There is no and monitoring of its health services and in disease
statutory requirement on doctors or midwives to surveillance activities. Ideally, every health event
notify cases. One of the problems with these data is and every kind of health resource would be
the definition of malformations. There is little dif- recorded in a systematic and instantly available form.
ficulty in detecting a major malformation but In practice, this is neither possible nor desirable as
some minor abnormalities may not be noticed, or it would require an enormously complex and ex-
if noticed are not deemed to be of sufficient impor- pensive system which would be too slow and cum-
tance to justify notification. bersome to be of value. Most health information
systems have been developed to meet particular
needs. Nevertheless, the data are often inaccurate
Cancer registration
and the system does not always allow users’ ques-
Malignant disease has long been a major cause of tions to be answered with ease. These shortcom-
morbidity and mortality in the UK and in most ings tend to bring systems into disrepute and the
other countries. In order to study these diseases, it enthusiasm for collecting data (as well as making
is essential to know the numbers of people affected use of it) wanes.
by different forms of cancer and their survival Some of the commonly encountered problems
rates. The system of cancer registration was set up of information systems are as follows.
in 1971 specifically to facilitate research in this
field. Hospitals notify new cases of cancer to their
regional centre. Data from all regions are analysed Information system problems
further by ONS. Periodic official reports are pub-
• Lack of motivation among recorders
lished giving detailed tabulations of incidence, sur- • Design of data capture procedure
vival and mortality rates for various malignant • Inflexibility in the system
diseases at different stages. • Irrelevance of analyses
Laboratory reports
Lack of motivation among recorders Often a low pri-
The CDSC of the PHLS receives weekly reports ority is accorded to the task of record keeping. This
from microbiology laboratories in England and leads to delays in completion and poor quality of
Wales on cases of laboratory-diagnosed infections. records, for example inaccurate information, items
The amount of clinical and epidemiological data missing or no record at all. For these reasons all
reported varies depending on the infection. Al- arrangements for ‘data capture’, as it is called,
though the data are incomplete and lack denomi- should be simple to operate and create the mini-
nators which prevents their use to calculate mum amount of work.
incidence rates, they provide a useful means of
monitoring trends and detecting outbreaks. The Design of data capture procedure The type of record
CDSC also collects data related to infectious dis- needed for an information system is not always
ease from other sources, for example reports of out- compatible with that required for clinical purpos-
break investigations and immunization statistics. es. It is often possible, however, to use records
made for other purposes if they are carefully de-
signed, i.e. in standard format with provision for
Health information systems
coding, etc. This requires a degree of collaboration
Information systems are used to assemble facts and between different interests which it is often hard to
figures from a variety of sources for analysis. Their achieve.
61
PID8 5/21/04 11:33 AM Page 62
Inflexibility in the system The need for simplicity in fied so that the data recorded will be appropriate to
records means that the number of items recorded their purpose and the collection of irrelevant data
has to be restricted. Some flexibility can be gained can be avoided.
by allowing room on the record for additional • The recording procedures should be standard-
items of local interest beyond a set of basic data ized and the data collected should be easy to ob-
required of all recorders. tain, accurate and as complete as possible in order
that reliable comparisons can be made over periods
Irrelevance of analyses Users may feel that standard of time and between different places.
analyses tell them nothing new or are unhelpful in • Data should be collected from all relevant
solving their problems. This tends to sap enthusi- sources for collation and analysis at a central point.
asm for the system. • There should be well-organized provision for
In the design of a routine information system, data storage, updating, processing and retrieval.
therefore, the following requirements should be • The system should be capable of providing an-
met. swers to enquiries within the field for which it is
• The intended uses of the system should be speci- designed, with speed and accuracy.
62
PID9 5/21/04 11:34 AM Page 63
Chapter 9
Indices of health and disease, and
standardization of rates
63
PID9 5/21/04 11:34 AM Page 64
could be grouped as less than 20, 20–24.9, 25–29.9, • Age-specific rates relate the number of events in
30–34.9 and 35 and over. The definition of groups people in a specified age group to the total popula-
is determined by the purpose of the investigation tion in that age group, for example y deaths per
and the characteristics of the population being in- 1000 men aged 45–64 years per year.
vestigated. Thus the categories of weight used in • Cause-specific rates relate cases of a specified
studies of infants will be quite different to those disease to the population at risk, for example z
used in adults. cases of stroke per 1000 hypertensive patients per
Discrete variables may also be grouped to pro- year.
duce larger numbers in each category. The class Such rates must always have a specified time
intervals between successive groups should usually dimension.
be equal but it is often convenient to group all
values at the extreme ends of a scale, which it
must be remembered distorts the frequency Special events related to total events
distribution.
Situations in which groupings are natural should Examples
be distinguished from those where they are arbi- • Stillbirths are usually expressed as x per 1000
trary; for example, ‘under 16 years’ and ‘16 years total births.
and over’ could be regarded as natural groupings in • Operative mortality can be expressed as y deaths
as much as people in the former category cannot be per 1000 operations.
married and those in the latter can. For other vari- • Case fatality rates relate the number of deaths
ables, for example blood pressure, there is no such from a particular illness to the total number of
natural division. It is possible arbitrarily to define cases of that illness.
systolic blood pressure in excess of 140 mmHg as These types of rate are not time dimensioned but
high and below that level as not high, but this does the period over which they were calculated should
not necessarily have any significance. Quantitative always be specified.
data rarely fall into natural categories.
64
PID9 5/21/04 11:34 AM Page 65
D – Died
J – Joined
L – Left
1 D
2
3
4
5
6
7 J
Subject number
8
9
10
11 D
12
13
14 L
15 L
16 J
17
18 D
19
20 L
Year A Year B Year C
b
n
c
v
n
b
ch
ril
ay
ne
ly
g
pt
t
v
c
t
Oc
De
Oc
Fe
Ja
De
No
No
Fe
Au
Ja
Ju
Ap
Se
ar
M
Ju
Figure 9.1 Morbidity and mortality M
experienced in a hypothetical factory.
65
PID9 5/21/04 11:34 AM Page 66
(or other variable being measured) in the popula- tory; cultural and social background; occupation;
tion concerned. This section sets out some of the economic constraints (e.g. paid sick leave); and
common sources of error that may affect routine availability of medical care (which is related to
health information and the steps which can be numbers of doctors, distance from doctor’s surgery,
taken to reduce their effects. number of hospital beds and appointments sys-
Errors affecting mortality and morbidity rates are tems). The effect of variations in illness behaviour
of two kinds, as shown below. is most marked in mild, non-fatal and self-limiting
conditions.
Denominator error
Numerator error
The size of population at risk often cannot be de-
The number of recorded cases of a particular dis- fined accurately and various methods of estima-
ease may be in error for many reasons including tion have to be used. Some reasons for this are:
the following. • population migration between censuses which
may increase or decrease the size of population
within an area;
Diagnostic inaccuracy
• changes in population structure within different
This is affected by: the training, skills and interests areas (e.g. age, race, occupational distribution),
of the attending physician; advances in medical due to migration, changing fertility patterns, hous-
knowledge of pathogenesis; variations in the ing and industrial decay or development; and
criteria accepted in defining a diagnosis; and dif- • changes in administrative boundaries for rea-
ferences in the availability and use of special sons that may or may not relate to health and the
investigations. For example, until the mid-20th provision of health services.
century carcinoma of the cervix was not distin-
guished from carcinoma of the endometrium in
Reduction of error
routine mortality statistics — both were classified as
carcinoma of the uterus. It is important to be aware The effects of errors such as those above can be
of changes in the precision of diagnosis and classi- reduced as follows.
fication when investigating time trends in the inci- • By use of a standard diagnostic classification
dence of disease. such as the ICD when recording mortality or
modifications of this for morbidity.
• By combination of diagnostic categories be-
Incomplete identification of cases
tween which transposition of cases may occur, e.g.
The probability that patients will consult a doctor cancer of the colon and large bowel obstruction.
or be seen at or admitted to hospital, for example, • By use of standard recording and registration
is influenced by such factors as: past medical his- procedures.
66
PID9 5/21/04 11:34 AM Page 67
• By use of denominator populations derived occurred in a designated group with that of a stan-
from similar sources and compiled by comparable dard population. It is the ratio (usually expressed as
procedures. a percentage) of the number of deaths which oc-
Errors in routine statistics can rarely be complete- curred in the designated group to the number that
ly eliminated. Therefore, caution is needed in their would have been expected if the mortality rates in
interpretation, particularly between different local- each age band of the designated group had been
ities and at different times (see also Chapter 4). the same as those of the standard population.
Thus, the death rates for each age and sex group
in the standard population (Mx) are multiplied by
Standardization of rates
the number of people of that age and sex in the
Rates calculated by using the total number of population being investigated (Px). This gives the
events as the numerator and the total population ‘expected’ number of deaths in that particular
as the denominator are called crude rates. Their age/sex group. The expected deaths for each
value is limited, particularly when comparing two age/sex group are then added to give the ‘expected’
populations with different age structures, for ex- number of deaths in the whole population being
ample mortality rates in a new housing develop- investigated. The observed number of deaths (D) is
ment with many young families and those in a then divided by the expected deaths to give the
coastal resort with a large retired population. In SMR:
these circumstances, it is essential to adjust the
observed deaths ( D )
data to take account of the age differences between SMR = ¥ 100
expected deaths [Â ( Px ¥ M x )]
the populations; this is called age standardization.
The two methods of standardization most fre-
quently used are indirect standardization and Example
direct standardization. Members of the armed forces tend to be younger
than the male population of the country as a
whole. Therefore, the fact that they have a lower
Indirect standardization mortality rate is not illuminating. It is necessary to
A commonly used method of indirect standardiza- examine the mortality of this occupational group
tion for age is to calculate the standardized mor- after taking account of the age factor. Their SMR for
tality rate (SMR). The SMR compares the mortality ischaemic heart disease (IHD) is calculated in Table
(either from a specific disease or for all causes) which 9.1. This indicates that mortality from IHD
Table 9.1 Mortality from ischaemic heart disease (IHD) in men serving in the armed forces.
15–24 0 165.03 0 1
25–34 0.06 73.24 4.39 6
35–44 0.50 42.25 21.13 22
45–54 2.01 15.93 32.02 43
55–64 6.05 4.67 28.75 76
67
PID9 5/21/04 11:34 AM Page 68
Table 9.2 Indirect standardization: mortality in males in England and Wales in 1965 compared with 1973.
68
PID10 5/21/04 11:36 AM Page 69
Chapter 10
Medical demography
69
PID10 5/21/04 11:36 AM Page 70
meet the basic needs of the majority of humankind The areas with high population growth are
within the next few generations. mainly developing countries where there are al-
Cataclysmic prophecies that humankind’s fu- ready regular famines, chronic poverty, frequent
ture is threatened in this way are not new. They epidemics of crippling diseases and declining liv-
have been widely debated since the 18th century. ing standards. The situation will only be remedied
Probably the best known writer associated with if those countries with the highest growth rates in
the problems of overpopulation is the Reverend population achieve stability and the countries
Thomas Malthus, an 18th century English clergy- with the highest growth rates in consumption of
man, who attracted attention by his essay on ‘The resources reduce their demands.
principles of population as it affects the future im- The global problem of population growth is
provement of society’. The two principles from compounded by the fact that people are not
which he argued were: ‘that food is necessary for evenly distributed on the habitable surface of the
the existence of man’ and that ‘the passion be- earth. Food shortages and disease are problems in
tween the sexes is necessary and will remain nearly some areas simply because of the local density of
in its present state’. He argued that the power of the population rather than because the area as a whole
population to reproduce was greater than power of has insufficient natural resources. It is important to
the earth to produce food. He concluded that there recognize that health depends as much upon the
must be a ‘strong and constantly operating check systems for the distribution of food and water and
on population from the difficulty of subsistence’. the disposal of waste as it does upon the quantity
This conclusion led him to recommend that there of food produced or the availability of medical
should be no extension of relief for the poor, as this services.
would artificially reduce the difficulties of subsis-
tence and lead to uncontrolled population growth!
Populations and growth rates
The time scale within which he predicted catastro-
phe was wrong, partly because he did not foresee The size of the world’s population and its growth
emigration and colonization. His contention that rate is arrived at by collating data from every coun-
difficulties in subsistence would act as a constant try. The quality of the data varies considerably
check on population growth has also been proved from country to country. Most of the richer indus-
wrong by the experience in the countries of Latin trialized countries undertake regular and detailed
America, the Indian subcontinent and elsewhere. censuses similar to those undertaken in England
At about the same time as the ideas of Malthus and Wales (see p. 51). They also have sophisticated
were being debated in Europe, similar discussions and comprehensive systems for the registration of
were taking place in China. Hung Wang Chi noted births, deaths and marriages. From these sources it
in 1793 that ‘during a long reign of peace the gov- is possible to build up a complete picture of the
ernment cannot prevent people from multiplying way in which the size and structure of the popula-
themselves, yet its remedies are few’. One of the so- tion changes.
lutions that he suggested was to legalize and en- In the poorer countries of the world national
courage female infanticide. Discussions of the censuses are conducted infrequently and tend to
problems of population have continued through- be incomplete. The additional data that are re-
out the world up to the present time but now more quired for demographic studies (e.g. the registra-
is known about the size of the world population, tion of vital events) are often defective. There are
the dynamics of growth and the potential re- particular difficulties in the most deprived sections
sources of the earth. The United Nations, through of these countries and amongst nomadic peoples
its various agencies, regards population growth as or those living in sparsely populated regions of the
one of the major world problems that will affect world with poor communications. In these latter
the quality of life, health and survival of situations, much of the data are available only on
humankind. an irregular sample basis. It is not surprising that
70
PID10 5/21/04 11:36 AM Page 71
most of the work on population growth has used impacts in some parts of the country. For example,
European data, because only in recent times has it Fig. 8.1 (p. 54) shows the ‘General Bills of Mortal-
been possible to study many of the other countries ity for London’ for 1641 and 1665. In both years,
of the world. the number of deaths greatly exceeded the number
The trends in population growth in England and of births, in 1665 by a factor of 10. It should be re-
Wales are not dissimilar to those in most European membered that Fig. 10.1 is solely concerned with
countries and can be used to illustrate the size and the resident population and that during much of
speed at which changes occurred. It has proved the period there was substantial migration. It
possible to estimate the number of residents at var- should also be noted that the scale of the figure is
ious times between 1100 and the early 19th cen- such that the recent reduction in population
tury from analysis of ecclesiastical and governmental growth rate is not apparent.
records. From the 19th century onwards formal At the same time as the population increased, its
census figures are available. The trend has been for age structure changed. Figure 10.2 compares the
the population to increase exponentially (Fig. age distribution of the population in 1821 with
10.1). The temporary decreases in population due that in 1991. In 1821, the proportion of children
to major national disasters such as epidemics of was much greater than at the present time and the
plague or war are not discernible within the scale proportion of people over the age of 50 was con-
used on the figure but at the time they had major siderably less.
The population can only increase if the number
of births exceeds the number of deaths. The
growth rate of human populations tends to be ex-
ponential because with each annual increase in
40 births the proportion of the population potentially
capable of reproduction increases. For this reason,
1 000 000s
30
the statement that there is an annual growth rate
20 of x per 1000 population (x being the difference be-
10 tween the birth rate and the death rate) gives a mis-
leading impression of the magnitude of change.
1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 The conventional way of expressing growth is the
Centuries population doubling time. This is the theoretical
period that it will take for a given population to
Figure 10.1 The growth of population in England and double, based upon the most recently available
Wales. data. Clearly, the doubling time will have to be
80+
70–79 1821 2001
60–69
50–59
40–49
Age
30–39
20–29
10–19
0–9
Figure 10.2 Comparison of the age 30 20 10 0 10 20
structure of the population of England Per cent
and Wales in 1821 with that in 1991.
71
PID10 5/21/04 11:36 AM Page 72
revised when there is a change in either birth or tive of whether they are both high or both low.
death rate. The doubling time for the population of This phase is represented in Fig. 10.3 as period A.
the UK, together with that for a number of other Typically primitive rural societies and poorly de-
countries, is given in Table 10.1. veloped urban societies tend to have high birth
and high death rates. The highest mortality tends
to be in infancy and childhood due to the com-
Demographic transition
bined effects of disease and poor nutrition.
The model of demographic transition provides a Social progress and the introduction of indus-
useful framework within which to consider the trial technology bring tangible and immediate
factors that determine changes in the size and benefits to the community. The most obvious are
structure of human populations. The population improvements in sanitation, in water supply and
is stable both in size and in age structure when the in the ability to distribute and store food. The im-
birth and death rates are equal and static, irrespec- mediate effect of these changes is that the chances
Table 10.1 Population doubling times in various countries and regions of the world. (Source: WHO, 1991.)
Population
Population Birth rate Death rate doubling Life expectancy, Life expectancy,
Country (1000s) (per 1000) (per 1000) time (years) males (years) females (years) Fertility
72
PID10 5/21/04 11:36 AM Page 73
of survival amongst the most vulnerable within societies, the desirability of large families, which is
the community, infants and children, are im- a biological necessity for survival in pretransition-
proved. Therefore the death rate begins to fall and al communities, is formalized within the belief
the community enters phase B in Fig. 10.3. During system of the group. For example, in many societies,
this phase, the crude birth rate actually rises be- the number of children a man has is perceived as a
cause the proportion of the population that is ca- measure of his virility. The next phase (C in
pable of reproduction increases and there is little Fig. 10.3) is characterized by a decrease in the
change in the age-specific birth rates. This is be- birth rate while the death rate continues to
cause people’s reproductive behaviour tends to be fall. Birth rates still exceed death rates and the
learned from their parents and it can take many exponential growth of the population, established
years to adapt fully to new circumstances. In many in phase B, continues. Again this is because, despite
a decrease in the average number of children born
to each woman, there are more women in the
reproductive age group than there were in the
previous phase.
Birth rate Population Eventually death rates stabilize (phase D) but
birth rates continue to fall. The transition of the so-
ciety is completed in phase E, when birth and
death rates are static and equal. By this time, the
size of the population is many times greater than it
was in the pretransitional phase. The size of the
new stable population is determined by the speed
Death rate of the transition.
Data from England and Wales can be used to il-
A B C D E lustrate the demographic changes discussed above.
The crude and the age-specific death rates for se-
Figure 10.3 Schematic representation of demographic lected age groups relative to the 1841 rates in Eng-
transition. land and Wales are shown in Fig. 10.4. The crude
80
40
35–44 year
30
olds
20
10
Figure 10.4 Age-specific death rates
per 10-year period for England and 1841 ‘51 ‘61 ‘71 ‘81 ‘91 1901 ‘11 ‘21 ‘31 ‘41 ‘51 ‘61 ‘71
Wales since 1841, as a percentage of Year
the 1841–50 age-specific rates.
73
PID10 5/21/04 11:36 AM Page 74
death rate is now about half what it was in the early ern world, owed more to improvements in the gen-
19th century. The greatest changes in mortality eral quality of life and to improvements in public
have been amongst the young, exemplified by the and personal hygiene than they did to any specific
5–9 year olds in the figure, which are now less than medical measures. Specific medical treatments
5% of the rates prevailing in the early 19th cen- were not introduced until long after the mortality
tury. The smallest changes have been amongst the rates from these diseases had undergone the
elderly. This is reflected in the change in life ex- greater part of their fall. It is noteworthy that many
pectancy, another way of summarizing mortality, of the lethal diseases of 19th-century Europe are
at different ages (Fig. 10.5). It is arrived at by apply- now regarded as ‘tropical diseases’. They are more
ing the prevailing age- and sex-specific mortality properly called ‘poverty diseases’. The principal
rates to the people who survive to a particular age. diseases that accounted for the high mortality and
It is clear that the greatest changes in life expect- which have now been controlled or eliminated in
ancy have been amongst the very young. Increased the western world were tuberculosis, the enteric
survival in the prereproductive age groups means fevers, cholera, smallpox, scarlet fever, measles,
that the proportion of the population capable of whooping cough and diphtheria.
reproduction increases. Thus, although each age During the 1840s, about 18% of all deaths in
group of women may maintain the same age- England and Wales were attributed to tuberculosis.
specific fertility rates as previous generations, the It is possible that some of these may have been
crude birth rates will rise. misdiagnosed carcinoma of the bronchus or some
other disease of the respiratory system, but the
numbers were so large that there can be little doubt
Reasons for the decline in mortality
that the downward trend in mortality rates shown
The reduction in mortality in England and Wales in Fig. 3.1 (p. 15) was mainly a reflection of tuber-
since the 19th century is almost entirely due to the culosis control. The decline in tuberculosis mortal-
elimination of the major endemic infectious dis- ity preceded the identification of the organism or
eases (Fig. 10.6). For most of these, mortality rates any specific treatment. The principal explanation
were highest amongst young people. It is apparent for this remarkable trend, however, probably lies in
that the virtual disappearance of these diseases improvements in diet and in consequent enhance-
from the UK, and from most countries in the west- ment of the resistance of individuals. The practice
24
70 Infectious
Deaths per 1000 population
diseases
20
60 Non-infectious
16 diseases
Expected years of life
50
12
40
1841 1979 8
30
4
20
0
10 1838 ‘39 ‘40 ‘41 ‘42 1969 ‘70 ‘71 ‘72 ‘73
Year
0 10 20 30 40 50 60 70 80
Age
Figure 10.6 Crude annual death rates from infectious and
non-infectious diseases in England and Wales, 1838–1842
Figure 10.5 Expectation of life at different ages in England and 1969–73. (Death rates from infectious diseases during
and Wales, 1841 and 1979. 1969–73 were too low to show on this scale.)
74
PID10 5/21/04 11:36 AM Page 75
of isolating cases, thereby reducing the spread of been accurate. This disease was endemic in the
the disease, probably also had an effect. 19th century (Fig. 10.9). Typically, there were su-
The enteric and diarrhoeal diseases were end- perimposed regular epidemics every 6–7 years. The
emic in the 19th century and were a particularly frequency of these epidemics was probably due to
important cause of death amongst infants and changes in population immunity. Contact with
children. Their impact began to decline in the mid- disease resulted in either death or lifelong immu-
19th century (Fig. 10.7) and seemed to be the result nity, thereby reducing the size of the susceptible
of improvements in personal hygiene and in child- population. After an epidemic, most survivors
rearing practices. A more specific measure, the pro- would be immune and this decreased the risk to
vision of a pure water supply, was responsible for the remaining susceptibles. When the proportion
the disappearance of cholera as an endemic disease of susceptibles in the population increased (by the
in the UK (Fig. 10.8). birth of children), a further epidemic occurred. Not
Because of the obvious physical signs of small- surprisingly, the majority of deaths occurred
pox, the statistics on its mortality are likely to have amongst children and infants. The elimination of
this disease was due to a specific medical measure,
the discovery of vaccination. However, it should be
noted that although vaccination became compul-
300 sory in England in 1852, it was not widely practised
280
Death rate per 100 000 population
260
240
220
200
180 100
160
100 000 population
140
Death rate per
75
120
100
80 50
60
40 25
20
1847 ‘49 ‘51 ‘53 ‘55 ‘57 ‘59 ‘61 ‘63 ‘65 ‘67 ‘69 ‘71 ‘73 ‘75 1851 ‘53 ‘55 ‘57 ‘59 ‘61 ‘63 ‘65 ‘67 ‘69 ‘71 ‘73 ‘75 ‘77 ‘79 ‘81
Year Year
Figure 10.8 Cholera mortality in England and Wales, Figure 10.9 Smallpox mortality in England and Wales,
1847–77. 1851–81.
75
PID10 5/21/04 11:36 AM Page 76
Social factors Table 10.2 First marriage rates per 1000 single women in
England and Wales. (Source: Registrar General’s Annual
By convention, child bearing and child rearing
Statistical Reviews.)
outside marriage have been discouraged in most
societies. In contemporary western societies, this Age in years
attitude has changed but in many parts of the
16–19 20–24 25–29
world, powerful taboos remain and societies
continue to censure the unmarried mother and 1938 28.1 171.6 132.2
1948 49.1 212.5 158.1
her child. Thus, marriage practices have a
1958 75.2 260.8 162.5
potent effect on the reproductive behaviour of
1968 84.6 260.9 161.4
societies.
1978 58.8 177.9 134.8
The legal minimum age of marriage is of less im- 1988 23.0 101.6 106.8
portance in most societies than the conventional
Table 10.3 Births to women aged 20–24 years in England and Wales. (Source: Registrar General’s Annual Statistical
Reviews.)
1939 252 33 93
1969 251 58 157
1988 212 30 95
76
PID10 5/21/04 11:36 AM Page 77
The legitimate birth rate to women aged 20–24 contraception. Within Christian cultures, the Hut-
was similar in 1939 and 1969 but the age-specific terite and Amish communities take the Biblical
birth rates differ considerably because the propor- dictum to go forth and multiply quite literally and
tion of women who were married changed. amongst them it has not been unusual for married
Divorce, separation and widowhood have the re- women to produce a dozen or more children.
verse effect on birth rates to those of marriage. The Some non-Christian religious groups also eschew
same conventions that discourage never-married contraception on principle. Local ethics and
women from having children discourage divorced morals may restrict the availability of the more ef-
and widowed women from reproducing. In normal ficient methods to certain groups. Thus, if sexual
times, this has little impact on birth rates but, after intercourse outside marriage is deemed wrong,
the First World War, when many women in Europe contraception for the unmarried may be seen as a
were widowed, there was a noticeable reduction in collusion with immorality. In the 1960s and 1970s,
the number of births, although there had been many clinics in England and some general practi-
little change in the size of the female population tioners would not advise unmarried women on
in the reproductive age group. contraception.
Sexual behaviour within marriage varies be- In societies where the role of women is seen
tween societies. Although taboos exist regarding mainly as child bearing and child rearing, women
the permissibility of intercourse at certain times, who limit their fertility may be rejected or may fear
for example during menstruation or religious rejection. Similar problems affect the acceptability
feasts, this has little measurable effect on birth of contraception in groups where a man’s success
rates. and strength is measured by the number of chil-
dren he fathers. During transition between high
and low mortality, fear of death of existing infants
Contraception
and children, resulting in the extinction of the
Although the possibility of contraception and family, leads to the production of more children. It
knowledge of techniques has existed for many is often difficult to convince parents in such
years (it was known to and used by the ancient societies that the survival of existing children is
Egyptians), its use varies substantially from place threatened by further enlargement of the family.
to place depending on its acceptability, availability Even if the idea of birth control is acceptable to
and efficiency. an individual, the method of contraception in-
Nowadays, in many societies, the most impor- volved may be unacceptable. Many of the simpler
tant social factor determining the patterns of re- methods require action by the male (e.g. the
production is the acceptability of contraception. In sheath or coitus interruptus), and they may detract
general, the better educated (and those who are from his satisfaction. The methods that require no
better off) are more likely to use contraception action at the time of intercourse usually require in-
than the ill-educated and poor. Its use is also deter- tervention by trained professionals (e.g. the in-
mined to some extent by religious beliefs. Mem- trauterine device (IUD) or sterilization). The choice
bers of the Roman Catholic Church are forbidden and use of methods of contraception is also
to use artificial methods of birth control. Neverthe- affected by the couple’s level of education. This is
less, the rule of the church is not universally ad- important in communities where birth control is
hered to and contraceptive practice varies amongst new and where modern techniques are not com-
Roman Catholics. It has been shown that a large mon knowledge. Most developing countries have
proportion of Roman Catholics in Europe and recognized this factor as important and are experi-
North America no longer adhere to their church’s menting with teaching methods. The most effec-
teaching. tive methods are usually the most expensive. If
The Roman Catholic Church is not the only reli- family economics mean that people cannot afford
gious group actively to discourage the practice of the new technology then in practice the method is
77
PID10 5/21/04 11:36 AM Page 78
30 30
Age at marriage
20 20
Months
Years
15 15
Interval in months
10 10
78
PID10 5/21/04 11:36 AM Page 79
between marriage and pregnancy is affected by the ent years. The 1920 and 1930 cohorts reached their
proportion of women who are pregnant when they peak birth rates at about the age of 26 years and fer-
marry. Figure 10.12 shows that in the 1960s about tility was high well into the 30s. By contrast, the
40% of women who married under the age of 20 1940 cohort reached its peak fertility at age 24
years and about 15% of women aged 20–29 years years and tended to have more children earlier in
were pregnant when they married. The propor- their lives. The 1950 cohort’s fertility was stable be-
tions fell in all age groups during the 1970s. The tween the ages of 21 and 28 years. It is probable
post-1970s changes were due to a combination of that family size of the pre-1941 cohorts was deter-
increased availability of abortion and of contracep- mined largely by the age of marriage and that,
tion to unmarried people. This hypothesis is con- within marriage, conscious control of fertility was
sistent with the fall both in the illegitimate birth haphazard, whilst the post-1941 cohorts married
rate and in the number of marriages of pregnant earlier and exercised a more precise conscious con-
women. trol over fertility.
Figure 10.13 shows the cumulative age-specific
fertility rates for cohorts of women born in differ-
Total-period fertility rate
This is a useful measure calculated from summing
the age-specific fertility rates and expressing the
40
sum of the rates as the expected number of live
35
Marriages with a birth within
Age 16–19
births per woman of child-bearing age. Thus, in
8 months (percentages)
2500
Births per 1000 women
79
PID10 5/21/04 11:36 AM Page 80
160
Terminations
140
Maternities
Total known conceptions (1000s)
120
100
80
60
40
20
Figure 10.14 Conceptions in women
0
1969 1974 1979 1984 under 20 years, distinguishing those
Year leading to maternities from those ter-
minated by abortion.
neous abortion and stillbirth rates are not signifi- for legitimate births, even after account is taken of
cant elements. Induced abortion depends upon parity and maternal age. There is a positive social
individual motivation and it affects age-specific class gradient, i.e. social class V has PMRs greater
birth rates selectively. In countries where induced than social class I. Some social class differences
abortion is legal, full statistics are published. Figure are due to reproductive behaviour. Birth weight
10.14 shows the numbers of ‘known’ conceptions is highly correlated with perinatal mortality. The
in women aged 16–19 years in England and proportion of low birth weight babies born within
Wales from 1969 to 2001, and demonstrates the a country is a strong determinant of its PMR. Also,
contribution of legal abortion to the fall in birth there is a close correlation between low birth
rate. weight and certain maternal factors, for example
Perinatal and infant mortality rates are some- parity, birth interval and maternal age.
times used as sensitive indicators of the quality of Poor maternal health can also adversely affect
health services within a country or within a dis- PMRs. Important diseases or conditions that have
trict. This is asserted because some of the causes of been shown to be associated with high PMR in-
perinatal and infant deaths are avoidable by clude the following.
medical intervention.
Three major studies of perinatal deaths in Britain
(1946, 1958 and 1970) involved following up co- Factors associated with high PMR
hort of births beyond the perinatal period to
• Hypertension
examine factors related to perinatal mortality and
• Poorly controlled diabetes
morbidity. They showed that adverse maternal ob- • Renal disease (which can also decrease fertility)
stetric factors act in a cumulative manner. Perinatal • Infection (hepatitis B, syphilis, rubella, cy-
mortality rates (PMRs) are highest in para 3 tomegalovirus and toxoplasmosis can cause fetal abnor-
women, in women at the end of their reproductive malities)
life and when the birth interval is less than 12 • Severe malnutrition
months or more than 60 months. Conversely, they • Smoking
• Alcohol can cause fetal alcohol syndrome (intrauterine
are lowest in para 1 women, in women aged 20–29
growth retardation, developmental delay and sponta-
years and when the birth interval is 18–35 months. neous abortion)
The PMR is higher for illegitimate births than it is
80
PID10 5/21/04 11:36 AM Page 81
Thus, while a large proportion of fetal and peri- reduction in mortality in developing countries
natal mortality is difficult to prevent, much can be are malnutrition, illiteracy and poverty.
done to reduce rates by appropriate antenatal and • Industrialization is inversely related to changes
postnatal care and advice. in fertility. Four explanations for this can be ad-
duced as follows:
• in urban societies children are not an eco-
Summary
nomic asset;
• Every industrialized nation has low mortality • as the infant death rate declines, the propor-
compared with non-industrialized countries. Fur- tion of children who survive to adulthood in-
ther substantial decline in mortality in industrial- creases and the number of births required to
ized countries is unlikely because the major causes attain a desired family size is smaller;
of death are associated with old age. • in urban societies, there are greater opportuni-
• There is great potential for further substantial ties for women outside the domestic environ-
reduction in mortality in Asia, Africa and Latin ment, and being committed to child rearing
America. This will be achieved by control of the restricts a woman’s activities; and
major infective diseases, especially gastrointestinal • in educated societies, the influence of secular
and respiratory infections in children and AIDS. rationality is stronger which allows readier ac-
• The principal factors acting against any quick ceptance of contraception.
81
PID11 5/21/04 11:37 AM Page 82
Chapter 11
Evidence-based medicine
82
PID11 5/21/04 11:37 AM Page 83
83
PID11 5/21/04 11:37 AM Page 84
• What are the results of the study. This often en- noted that if 20 characteristics are looked at, then
tails unpicking the results and presenting them in by chance (at the 5% level) a significant difference
a way that is more relevant to the question. between the groups is likely to be found in at least
• Will the results of this study affect practice, i.e. is one. The larger the study the more likely the
it clinically relevant? groups are to be similar. Thus big studies are to be
It is important to be aware that a statistically preferred. This will also help avoid the problem of
significant finding is not necessarily clinically a Type 1 or a Type 2 error (see below).
relevant.
Drop-out rates
Randomized controlled trials
Were all patients who entered the trial properly ac-
The principal form of evidence when considering counted for and attributed at its conclusion? If pa-
whether a treatment works or whether an exposure tients are lost to follow-up it may be that those
causes a particular outcome is an RCT. Hence it is patients who left the study had a different outcome
important to understand the principal compo- to those who were included in the final analysis.
nents of an RCT. The study should be appraised For example, if the outcome of interest is death,
critically to see whether it has been well conducted patients lost to follow-up may have had a
and can be believed. higher death rate than those who are followed up
throughout the study. This leads to an underesti-
mate of mortality in the groups studied. Similarly,
Points to look for in appraising an RCT if the drop-out rate between groups is different bias
• Randomization may be introduced. A common reason for a differ-
• Characterization of the groups ence in drop-out rates is that one treatment causes
• Drop-out rates more side-effects or is ineffective. Ideally all pa-
• Intention to treat analysis tients should receive the treatment to which they
• Blinding
were randomized, be followed up and their out-
• Sample size
comes noted.
• Results
84
PID11 5/21/04 11:37 AM Page 85
The results
Sample size
The results of an RCT are usually presented as
Small studies can sometimes be misleading. A Type
a relative risk (see p. 11). Relative risk (RR) is
1 error occurs when a study concludes that two
the absolute risk in the treated group divided by
treatments are different when in fact they are not.
the absolute risk in the untreated group (or vice
If a study is repeated 20 times, on one occasion by
versa). Another way the results can be presented is
chance a statistical difference (at the 5% level) will
by estimating the number needed to treat (NNT).
be detected. This difference occurs by chance but if
This takes into account not only the RR but also the
the results of only this one study are published it
absolute risk in the two groups being investigated.
will give a biased impression that the treatment in-
The NNT is calculated by taking the reciprocal of
vestigated is worthwhile when in fact it is not. A
the absolute risk reduction (ARR). It indicates how
Type 2 error is when the study concludes that the
many people would have to be treated with A as
treatment groups are not different when in fact
compared to B in order to prevent one additional
they are. In this case, unless a big enough study has
outcome of interest. For example, imagine 2000
been carried out the difference will not be
patients with mild hypertension are randomly al-
detected — a Type 2 error. It may be that although
located to treatment or placebo. At the end of the
there is a true difference between the two interven-
year 4 patients in the placebo group have had a
tions, the size of this difference is small and may not
stroke and only 2 in the treated group have suf-
be clinically relevant. Thus a Type 2 error is often not
fered a stroke. The RR for the treated group if 0.5.
considered such a serious problem as a Type 1 error.
Thus the treatment produces a 50% reduction in
the number of strokes. However the NNT in this
Groups are not Groups are example is 500. Five hundred people will have to
different different be treated for 1 year (and carry any risks associated
with that treatment) for one patient to benefit.
Conclude groups Correct decision Type 2 error
NNTs are probably a more relevant index to be
are not different
used for clinical practice although there is evidence
Conclude groups Type 1 error Correct
are different decision that decision makers are more likely to alter their
practice when presented with the RR.
85
PID11 5/21/04 11:37 AM Page 86
that deal with a focused question. These papers are searched, e.g. Medline, Cochrane, Embase, Cinahl,
then appraised critically to identify their strengths etc.
and weaknesses. Finally a summary of the evidence • Details of secondary references, i.e. the refer-
is reported. Well done systematic reviews are in- ences cited by the papers that were retrieved from
creasingly being accepted as the highest form of the original search.
evidence in the hierarchy of evidence. The ‘gold • Studies published in languages other than Eng-
standard’ for reviews are those done to the criteria lish. Researchers undertaking systematic reviews
set down by the Cochrane Collaboration — a group are often tempted to exclude these papers because
of researchers and clinicians interested in under- of the cost of having them translated (as well as the
taking systematic reviews of randomized trials. delay). However this may mean that some per-
Since a systematic review is a retrospective look at fectly valid studies that deal with the question are
published papers, it is important to make the not included. This again can introduce bias.
process rigorous and well defined to prevent bias • Grey literature. Studies with negative findings
and thus distortion of the findings. may be difficult to get published. If only studies
with positive results are published then the pub-
lished papers will give a positive result. The results
Points to look for in appraising a systematic of unpublished studies (the grey literature) if rele-
review vant should be included. They can be obtained by
contacting researchers known to be active in the
• Criteria for inclusion
• Sensitivity of the search field of interest. Also drug companies often have
• Method of selection unpublished studies which they may release to
• Validity of the studies researchers.
Once all the papers have been collected then
they should be appraised critically and an evalua-
tion of the overall findings made.
Criteria for inclusion
A systematic review should have clearly defined
Meta-analysis
criteria for the inclusion of studies. This usually in-
cludes the type of study (for therapy questions ide- A meta-analysis is a particular type of systematic re-
ally an RCT), the populations included in the view that uses quantitative methods to combine
studies, the treatments or exposures and relevant the results of several independent studies consid-
outcomes. The criteria should not be so restrictive ered by the analyst to be combinable. The overall
that important studies are likely to be missed. Ide- results are weighted by the size of the contributing
ally the authors should list all trials reviewed with studies. This means that the larger studies will have
a reject log and reasons for exclusion. the main influence on the outcome. The results of
a meta-analysis can be presented in a tabular or
graphical form.
Sensitivity of the search
The systematic review should demonstrate that a
Challenges to evidence-based
sensitive search strategy was adopted. Ideally it will
practice
include:
• The words that were used in the interrogation of There is a word of warning about the rational or sci-
the medical databases, which terms were com- entific approach to medicine. Firstly there is a mis-
bined and which intersected. match between the needs of patients and the
• The time period over which papers could be research agenda that provides the evidence. Re-
included. search agendas are set by those with the funds —
• The databases and other sources that have been particularly the research councils and the
86
PID11 5/21/04 11:37 AM Page 87
pharmaceutical industry. The Medical Research by published studies. Doctors have always had the
Council has been criticized for the lack of involve- clinical freedom to make judgements about the
ment of patients in setting research agendas. best course for individual patients, balancing their
Rather, it is a panel of scientists with a particular knowledge of the patient, their clinical experience
view of what research is needed that commissions and the evidence from the literature. Increasingly
studies. Their viewpoint has a major influence on with the publication and dissemination of proto-
the type of research that is funded. The pharma- cols and service frameworks the ability to balance
ceutical industry, which provides more than 60% experience with evidence-based practice is being
of the funds for medical research in the UK, natu- eroded.
rally has a different perspective. They want to find Thirdly there is the issue of the balance between
medications that will alleviate patients’ problems the patient’s experience and understanding and
and will lead to commercial success. Investigations that of the health service. Clinicians are encour-
of behavioural or population-based interventions aged to respect the autonomy of patients and to
have a much harder time attracting funds. Conse- take into account their views and experience when
quently there is more evidence on the effectiveness offering them treatment. But often these conflict
of interventions that involve drug treatment than with evidence from the published literature.
other modalities. Which should take precedence?
Studies on new drugs often exclude specific
groups for ethical or safety reasons. For example,
Strengths and weaknesses of EBM
there are few studies amongst women of child-
bearing age, children and the elderly. Practising EBM has a number of strengths and weaknesses. Its
EBM can be difficult for obstetricians, paediatri- practice requires basic skills in searching databases
cians and geriatricians. A consequence of exclud- of the medical literature, skills in epidemiology to
ing certain groups from RCTs is well illustrated by help appraise the relevant papers and skills in sta-
the recommendations for the use of lipid-lowering tistics to help interpret the results.
therapy. It has been suggested that only those who
are at high risk of heart disease and who are under
75 years of age should be treated. Part of the reason Strengths
for the age cut-off is that the trials of lipid-lowering
• Helps clinicians in their decision-making
therapy excluded older patients (above 70 or 75
• Helps ensure consistency of care offered to patients
years of age). This was not because these patients • Develops skills in critical appraisal
are unlikely to benefit but because the likelihood of • Helps clinicians keep up to date
side-effects and adverse events are higher in older • Helps in the development of evidence-based guide-
people. Consequently it makes sense to try to avoid lines
including these patients in the trials. Unfortu-
nately there is then no direct evidence from the
RCTs of the outcomes in older patients.
Weaknesses
The second problem is that EBM often clashes
with clinical experience and does not take account • Development of the evidence base has been biased
of the context in which practice takes place. It has • Ignores the benefit of clinical experience
• Does not take into account patient choice
been shown that the occurrence of an adverse
• Can be time consuming and requires the acquisition of
event when treating a patient was one of the
basic skills
biggest barriers to following treatment suggested
87
PID11 5/21/04 11:37 AM Page 88
PID12 5/21/04 11:38 AM Page 89
Part 2
Prevention and Control of Disease
PID12 5/21/04 11:38 AM Page 90
PID12 5/21/04 11:38 AM Page 91
Chapter 12
General principles
91
PID12 5/21/04 11:38 AM Page 92
tious diseases were the major causes of morbidity of these diseases is more complicated than the con-
and mortality, particularly in children and young trol of infectious diseases and therefore progress is
adults. Their control over the past 150 years owes more difficult to achieve. There has been a propor-
more to social and economic progress than it does tional increase in accidents as a cause of morbidity.
to specific medical intervention. Preventive pro- The problems of prevention of chronic diseases
grammes during this period have included such centre around their natural history, the difficulty
measures as improvements in sanitation, water sup- in identifying aetiological agents and the fact that
ply, the quantity and quality of food and the quali- many have multiple causes. Moreover, they are
ty of housing, safer conditions in the workplace and generally characterized by having a long latent
raised standards of personal hygiene. All of these period between exposure to the aetiological agent
carry obvious and immediate benefits other than and the appearance of symptoms. In many cases,
those purely related to health: they make life more the symptoms have an insidious onset and by the
comfortable and pleasant with little or no restric- time they are of sufficient severity to cause the af-
tion on personal freedom. Most of the changes were fected individual to seek medical attention, ir-
at community level and were the result of legisla- reparable damage has been done. Prevention of
tion rather than action by individuals. This made these diseases often depends on actions by the in-
them comparatively easy to institute. By contrast, dividual, rather than passively enjoying improve-
some of the more recent advances in the control ments in the environment brought about by the
and prevention of communicable diseases, such as actions of others. It demands modification of per-
the elimination of diphtheria and poliomyelitis in sonal behaviour in such matters as the use of to-
many countries and the worldwide eradication of bacco and alcohol, diet and exercise at a time in life
smallpox, required mainly medical action (immu- when the risks of contracting the disease in ques-
nization) and thus can rightly be claimed as major tion are seen as remote. It is also a fact that, even
medical achievements. The benefits of environmen- for common diseases, the absolute risks for the
tal improvements, as well as of specific immuniza- individual are indeed relatively small. In these
tion, however, will be sustained only by continued circumstances, campaigns to persuade people to
vigilance. Much modern preventive medicine is change their lifestyle require great skill and pa-
directed to this end. In the past, the presence of a tience sustained over long periods of time. These
disease in the community served as a constant re- lifestyle changes also need to be complemented by
minder of its nature and consequences. In societies public policies that promote health by, for exam-
dependent upon distant memories of childhood in- ple, the taxation of tobacco and alcohol products,
fections such as measles, whooping cough, polio the subsidizing of food production and the provi-
and tuberculosis, continuing public education is es- sion of public recreational facilities. These all re-
sential to sustain preventive activities. With the ex- quire a political will to be implemented. Despite
ception of smallpox the causal organisms have not the difficulties, prevention remains an important
been eradicated from human populations. Thus the aspiration and progress is being made in some of
diseases can recur. these diseases (e.g. in reduction of cancer morta-
The virtual elimination of the older life- lity), both by action at a political level and by per-
threatening infectious diseases has brought the suading people to change their lifestyle and habits.
non-infectious illnesses into greater prominence. In The interaction between the social and physical
modern times, despite the emergence of new infec- environment and health has also been much more
tious disease such as legionnaires’ disease, HIV and widely recognized in the last 30 years by national
severe acute respiratory syndrome (SARS), it is car- and international bodies such as the World Health
diovascular disease, malignancies, degenerative Organization (WHO). It has led to the concept of
conditions (such as arthritis) and other chronic ill- the promotion of a healthy environment and
nesses which occur amongst older people that are lifestyle being adopted in a number of cities. Ac-
the major health problems. The prevention of many knowledgement that employment, housing, bal-
92
PID12 5/21/04 11:38 AM Page 93
anced diets and a social and economic environ- progress is rapidly arrested or reversed or its conse-
ment that promotes health are all important in im- quences minimized.
proving the quality of people’s lives and increasing Useful preventive action does not necessarily re-
the length of life has meant that both government quire knowledge of the cause of a disease. There are
and local policies which affect social factors have many examples of effective prevention that pre-
to take into account the long-term consequences ceded discovery of the agent or complete under-
to health. standing of the causal mechanism. For example,
in the 18th century, Lind (Fig. 12.2) and Blane
demonstrated that scurvy in the crews of ships
Principles of prevention
could be prevented by the consumption of ade-
Disease is the result of a harmful interaction be- quate amounts of citrus fruit; this was long before
tween the host (humans), a pathogenic agent and vitamin C was discovered. In the 19th century
the environment (Fig. 12.1). Agent, host and envi- John Snow (Fig. 12.3) showed that cholera was
ronment form a dynamic system in which, in the transmitted by drinking water polluted by sewage.
healthy individual, the balance normally favours His findings led to the elimination of cholera by
the host. Thus, if the agent is locally absent or con- the provision of pure water supplies many decades
tained, or its capacity to cause disease is matched before the isolation of the causal organism. In this
by the host’s protective mechanisms, or the envi- century, Doll et al. (see Chapter 5) demon-
ronment inhibits the spread of the agent, health is strated that those who stop smoking cigarettes
maintained. Disease or injury occurs when the bal- substantially reduce their risk of contracting lung
ance is disturbed, for example owing to changes in cancer, though the carcinogenic agent in tobacco
the pathogenicity of an agent, changes in environ- smoke has yet to be identified. In general, however,
mental conditions that favour the survival and a full and accurate understanding of the causes
transmission of the agent to humans, or the break-
down or absence of human normal defence mech-
anisms. The control and prevention of disease
depends on effective intervention in the relation-
ship between agent, host and environment to en-
sure that the balance remains in the human’s
favour, or, if disease does occur, to ensure that its
Environmental
conditions:
Physical
Biological
Affects presence Social Affects human
and survival of capacity to resist
Affects exposure
agents diseases
of humans to agents
Agent properties: Human protective
Microbial mechanism:
Chemical Immunity
Physical natural (non-specific)
Psychological acquired (specific)
Behaviour
Figure 12.1 Interactions of agent, host and environment, Figure 12.2 James Lind (1716–94) author of the treatise
causing disease. on scurvy.
93
PID12 5/21/04 11:38 AM Page 94
94
PID12 5/21/04 11:38 AM Page 95
Action
High-risk individual vs.
• Primary prevention: prevents disease starting population strategy
• Secondary prevention: detects disease early
• Tertiary prevention: damage limitation Where a choice of strategy exists, the planning of a
preventive programme should take account of cer-
tain practical considerations. The most desirable
Primary prevention
approach is one that gives the greatest benefit to
This aims to prevent a disease process from start- the largest number of people. In some instances,
ing. It often calls for strategies directed at the re- this may mean that the most effective strategy is to
moval or destruction of agents but can also include target high-risk individuals. Such programmes,
environmental control, immunization, health whilst of benefit to individuals, may do little to re-
promotion and health education. duce the overall burden of disease in a population.
Sometimes a population-based approach which
confers a smaller benefit on a large number of indi-
Secondary prevention
viduals may yield greater dividends. The popula-
This aims to detect disease at the earliest possible tion strategy has the advantage that there is no
stage and to institute measures to cure or prevent need to identify a high-risk group. Everyone is tar-
its further progression. Screening programmes geted. Interventions that are simple and require
backed by effective interventions are the most im- minimal cooperation from individuals are usually
portant examples of secondary prevention. the most successful. Economic factors must also be
considered when deciding on the most appropriate
intervention strategy. Each of these strategies for
Tertiary prevention
prevention is considered in detail in the chapters
This is concerned with ‘damage limitation’ in peo- that follow.
95
PID13 5/21/04 11:39 AM Page 96
Chapter 13
Health promotion and
health education
96
PID13 5/21/04 11:39 AM Page 97
Cities can be found at the WHO website identify and remove obstacles to healthy policies
www.who.dk/healthy-cities/. At the same time the so that these become the easier choice.
role of the UK Health Education Council, which
was set up in 1968, was expanded to include public Creating supportive environments To create living
policy advice and social and environmental issues and working conditions that are safe, stimulating,
in addition to the provision and distribution of satisfying and enjoyable. To encourage communi-
health education material. ties to care for each other, and to take responsibi-
The key components of health promotion were lity for the conservation of natural resources.
defined in a charter agreed at the first Interna-
tional Conference on Health Promotion held in Strengthening community action To work through
Ottawa in 1986. This suggested a definition of effective community action in setting priorities,
health promotion and five key areas for action. The making decisions, planning strategies and imple-
Ottawa Charter stated that: menting them to achieve better health.
97
PID13 5/21/04 11:39 AM Page 98
98
PID13 5/21/04 11:39 AM Page 99
Smoking
Strategies to reduce smoking
The UK has a long history of providing informa-
• Increase the price of cigarettes
tion about the dangers of smoking through
• Ban advertising
government-funded campaigns, advice from gen-
• Ban smoking in the work place and public places
eral practitioners and health campaigns in schools. • Identify and counsel current smokers
Punitive tax on tobacco is one public health • Provide smoking cessation clinics
policy, which has been shown to be effective in re- • Enforce the law on sales to children
ducing smoking. A 10% rise in price has been
associated with a 1% reduction in smoking. Ban-
Alcohol
ning the sale of cigarettes to children under the age
of 16 years and the prohibition of smoking in cer- Alcohol abuse is of increasing concern. It is esti-
tain public places are other examples of relevant mated that in the UK up to 40 000 deaths per year
legislative policies. The banning of advertising in are alcohol related, including a significant propor-
countries such as Canada and New Zealand has tion of the 3500 road deaths. Cirrhosis of the liver
been shown to reduce tobacco consumption, and is now four times more common in middle-aged
the UK and Europe are now following suit. men than it was in the 1970s.
Many companies and hospitals have attempted Public policies relating to alcohol include the
to create healthier environments by the introduc- imposition of excise duties and the passing of li-
tion of no-smoking policies. Some have also censing laws. The UK has among the highest rates
funded smoking cessation support for their staff. of tax on alcohol in the EU. The licensing laws were
Cinemas, airlines and some restaurants now ban introduced initially to control the ‘gin palaces’ of
smoking. In March 2004 the Republic of Ireland the 18th and 19th centuries. Paradoxically, these
passed legislation to ban smoking in public places laws are now being relaxed. Another policy inter-
such as pubs and resturants. vention aimed at reducing alcohol-related deaths
Little is done to support voluntary organizations was the passing of the drink–driving laws. This has
financially in their campaigns against tobacco. A resulted in a considerable reduction in the number
Canadian campaign involving health authorities, of deaths on the roads.
Action on Smoking and Health (ASH) and the Doctors have not always been good advocates or
Canadian Cancer Society demonstrated the effec- role models for the prevention of alcohol abuse.
tiveness of combined action in achieving a ban on The tradition of medical student drinking can lead
tobacco advertising in that country. to the development of unhelpful professional and
One of the goals that general practitioners have personal attitudes to drink. Strategies aimed at
been set as part of the National Service Framework creating supportive environments to contain the
on Cardiovascular Disease involves identifying the abuse of alcohol should include offering people
number of tobacco smokers within their practice. healthy choices, for example putting water on the
They can then refer them to smoking cessation table at mealtimes both in the home and when eat-
clinics or prescribe supportive treatment such as ing in restaurants. Offering food in pubs and other
99
PID13 5/21/04 11:39 AM Page 100
places where alcohol is served also encourages oils rather than animal fats. Whilst a population
more responsible drinking. Education includes giv- approach to nutrition is attractive, the use of a
ing people information about safer drinking levels targeted approach in certain situations is also
and publicizing the existence of help agencies. valuable. For example, preconception advice for
Often, conflicting information about the health women concerning their intake of folate will re-
benefits of moderate drinking is preferentially duce the risk of them having a baby with a neural
heard, perhaps encouraging light drinkers to drink tube defect. Perhaps more could be done to im-
more whilst doing nothing to encourage the heavy prove nutrition through the adoption of nutri-
drinker to reduce intake. tional policies. For instance, one initiative by the
Advice on dealing with alcohol abuse can be pro- Department of Health has been the ‘Five a Day’
vided to individuals. To do this those people with a programme which has been taken up by a number
problem need to be identified. Simple screening of primary care trusts and aims to get at-risk popu-
questionnaires on all at-risk patients can be used lations to eat five portions of fruit and vegetables a
both in hospital practice and in primary care. day. The Government has also launched the ‘Food
in Schools’ programme which aims to improve
school children’s knowledge about healthy
Strategies to reduce harm from alcohol abuse nutrition. This programme was launched
through the British Nutrition Foundation
• Increase the price of alcohol
• Drink–driving laws (http://www.nutrition.org.uk/).
• Make water and soft drinks easily available The other important body is the Scientific Advi-
• Only offer alcohol with food sory Committee on Nutrition (SACN). This is a UK-
• Identify and counsel problem drinkers wide advisory committee set up to provide advice
on scientific aspects of nutrition and health. This
includes advice on the nutrient content of individ-
ual foods and advice on diet as a whole including
Nutrition
the definition of a balanced diet, and the nutri-
The subject of nutrition is full of mixed messages, tional status of people. They are also consulted on
due to the paucity of consistent scientific evidence nutritional issues that affect wider public health
on the health effects of dietary change. In most policy issues including conditions where nutri-
parts of the world, malnutrition is the greatest tional status is one of a number of risk factors (e.g.
threat to health. In the developed world, obesity is cardiovascular disease, cancer, osteoporosis and/or
now a major problem. Public policy in the field of obesity). The website is http://www.sacn.gov.uk/.
nutrition has been scant and poorly coordinated.
The Health of the Nation document published by
the UK DoH in 1990 promoted a reduction in the Strategies to improve nutrition
percentage of food energy derived from fat and also • Education through the media
aimed to reduce the prevalence of obesity. Despite • No tax on healthy foods
this there has been a year-on-year increase in the • Targeted messages, e.g. folic acid for pregnant
prevalence of obesity. There are differential tax women
• Scientific advice available to policy makers
(VAT) rates on some foods, but legislation concern-
• Introduce nutrition on the school curriculum
ing food is generally aimed at minimizing known
hazards rather than supporting nutritional
objectives.
Exercise
Education about diet is widespread and often
most effectively undertaken by food manufactur- The health benefits of exercise are widely recog-
ers, for example encouraging the consumption of nized and yet its promotion is often uncoordi-
cereals, and the choice of margarine or vegetable nated. This is one area where public policy could
100
PID13 5/21/04 11:39 AM Page 101
have great influence. Some new towns in the 1970s statements by the GMC and BMA about the pre-
were designed with cycle paths and well-lit walk- scribing of the pill to girls below the age of consent.
ways to encourage healthy options for getting to The Government has a policy of providing free
and from work. The majority of local authorities contraceptive services through general practition-
have invested in sports facilities and made them ers and family planning services, but ease of access
available at subsidized rates, but many schools sold to services has to be complemented by appropriate
their sports grounds in the 1990s thus discourag- knowledge and behaviour. This is best encouraged
ing children from taking part in regular sports. through health education and by providing sup-
Recently this has been counteracted by a new ‘PE portive environments. The change in attitude to
and Sports Programme’ funded through local au- the advertisement of condoms on television and
thorities with the aim of increasing the provision their widespread availability through supermar-
and use of sports facilities. The ‘Healthy Schools kets and other retail outlets was brought about by a
Programme’ has also emphasized the importance need to promote a change in behaviour to try to re-
of physical activity to children. duce the spread of HIV. This has had an effect on
Knowledge about the benefits of exercise has in- other STDs as well as making people more aware of
creased dramatically over the last two decades. the risks of unwanted pregnancy. This example
This information is now being passed on by doc- shows how one health issue cannot always be
tors to their patients. Patients may be referred to separated from others.
rehabilitation programmes, which increasingly Some changes in health services seem to happen
emphasize the value of physical fitness. Much of by accident. Making the oral contraceptive avail-
this activity is in the form of tertiary prevention, as able only on a doctor’s prescription placed a clear
after a heart attack. However recent randomized responsibility on doctors, involving them in their
controlled trials have shown the benefit of regular patients’ sexual behaviour. General practitioners
exercise as a primary prevention strategy to reduce in particular accepted this responsibility so that
the risk of developing diabetes. now family planning advice is a major part of their
work.
The medicalization of contraception led doctors
Strategies to increase exercise to become involved in a number of other initia-
tives such as cervical screening and well women
• Healthy public policy, e.g. cycle tracks
• Increasing the provision of sports facilities clinics. The pill has thus been a very successful in-
• Sports in schools programmes fluence in reorientating doctors towards providing
• Exercise for high-risk patients, e.g. to prevent diabetes preventive rather than curative health care.
• Part of rehabilitation programmes, e.g. after a heart
attack
Ethics of health promotion
The ethics of health promotion can be approached
using the four principles often used when consid-
Sexually transmitted disease and
ering individual care.
unwanted pregnancy
Improving health through changes in sexual be-
haviour will help reduce the number of unwanted
pregnancies and sexually transmitted diseases Ethical principles
(STDs).
The laws designed to prevent underage sexual • Rights and responsibilities
• Beneficence
intercourse do little to reduce the incidence of
• Non-maleficence
teenage pregnancies. This growing problem and • Justice
the obvious need for contraception led to policy
101
PID13 5/21/04 11:39 AM Page 102
A key conflict arises between the goals of health to answers to allow it to make informed decisions.
promotion and the rights of individuals to per- Often the initiative to mount a preventive health
sonal autonomy. People working in health promo- programme is undertaken without proper consul-
tion sometimes seek restrictions on personal tation with the community. This is contrary to the
behaviour in the interests of the public good. This philosophy of health promotion, but is often due
can lead to conflict with a significant sector of the to ignorance on how to undertake community
public who wish to retain their autonomy of deci- consultation.
sion-making. Most agree that where the autonomy As far as justice is concerned, it could be argued
of others is threatened such as by drunk drivers on that funds should only be spent when there is a
the road, it is reasonable for society to intervene. good prospect of benefit to the health of the pub-
However, legislating against personal risk-taking is lic. This has been recognized by the Health Devel-
more controversial. There are no laws preventing opment Agency who have developed the HDA
mountaineering or bungee jumping, although Evidence Base so that health promotion pro-
there is legislation on the use of seat belts, which grammes of proven effectiveness can be pursued.
are only of benefit to the individual concerned. With regard to the targeting of programmes the
Similarly, the use of certain drugs is illegal al- ethics of a population-based approach must also be
though they usually only directly affect the indi- considered in the context of the needs to reduce
vidual user. Thus, the law and public attitudes on the inequities in health between the poor and the
these issues are not always consistent. rich.
In relation to beneficence and non-maleficence, These considerations suggest that all health pro-
in many situations the amount of good or the motion campaigns should at least be submitted to
amount of harm that may arise from many health an ethical review before being implemented, and
promotion initiatives is not known. This is not a that a facility should be in place to re-examine the
reason for inaction, but the community is entitled issues as the programme progresses.
102
PID14 5/21/04 11:41 AM Page 103
Chapter 14
Control of infectious diseases
Introduction Human
An infectious or communicable disease is an illness The human population is the reservoir of infection
caused by the transmission of a specific microbial in diseases such as measles and chickenpox. Were
agent (or its toxic products) to a susceptible host. these organisms to be eliminated from humans,
The agents can be bacteria, viruses or parasites. The the diseases they cause would be eradicated in the
majority of microbes are harmless to humans. same way that smallpox has been eradicated. How-
Some, although not universally pathogenic, are ever, due to their high infectivity and ease of trans-
potentially dangerous and may cause disease mission, these diseases are difficult to eliminate
in unusual circumstances. Caution is needed despite the use of mass vaccination programmes.
not to attribute a disease to an organism which In addition, some infections may be carried by
happens to be present as a commensal or non-symptomatic individuals who may transmit
contaminant. them to others. Asymptomatic carriers are often
There are many factors that determine whether difficult to identify.
or not biological agents result in the spread of dis- Human carriers are of three types: healthy,
ease in a population. They can be broadly divided convalescent or chronic.
into the presence of reservoirs of infection, the Healthy carriers are people who are colonized
method of transmission, the susceptibility of the by a potentially pathogenic organism without any
population or its individual members to the organ- detectable illness, for example staphylococcal car-
ism concerned, and the characteristics of the or- riage in the anterior nares or in the axilla, or
ganism itself. coliforms in the gut.
Convalescent carriers are people who have
recovered from the illness but who continue tem-
Reservoirs of infection
porarily to excrete the organism, for example sal-
A reservoir of infection is the site or sites in which monellae in faeces.
a disease agent normally lives and reproduces. Chronic carriers are people who, while re-
Reservoirs of infection may be classified as human, maining clinically well, may carry and excrete
other biological or environmental. organisms continuously or intermittently over a
prolonged period, for example typhoid carriers in
whom Salmonella typhi may remain in the gallblad-
der for life. Such carriers are a continuing threat to
103
PID14 5/21/04 11:41 AM Page 104
Host susceptibility
Transmission
Host factors that influence the natural history of
Infectious diseases can be transmitted by various infectious diseases include the following.
means and their mode of transmission influences
the spread of disease through a community. Inter-
Host factors
rupting the transmission of infectious agents is a
key strategy for the control of these diseases. • Age
Methods of transmission include the following. • Gender
• Nutrition
• Genetics
Transmission • Immunity: natural, acquired and population
Gender
Transmission is also affected by the conditions
which organisms require for their survival and There is some evidence that susceptibility to some
their life cycle. infections differs with gender. In general, males ex-
104
PID14 5/21/04 11:41 AM Page 105
105
PID14 5/21/04 11:41 AM Page 106
Antigenic stability
Organisms which are antigenically stable or exist
Infectivity
in only one antigenic form, for example measles
The infectivity of an organism is its capacity to virus, usually induce lifelong immunity. If the
multiply in or on the tissues of the host. This varies agent is antigenically unstable, for example in-
between microbial species, between individuals fluenza virus, or exists in many antigenic forms, for
and with the route of entry. It may also be affected example rhinovirus, humans cannot develop last-
by the presence of tissue trauma, which facilitates ing immunity. Environmental conditions, such as
the entry of organisms and provides a suitable those created by the indiscriminate use of anti-
growth medium. microbial drugs, may select out the more virulent
and resistant strains of bacteria from among
several coexisting variants.
Pathogenicity
The pathogenicity of an organism is its capacity to
The environment and infection
cause disease in an infected host (i.e. ratio of
number of cases of disease to total number of The environment is the physical, biological and so-
people infected). In the days before smallpox was cial world external to the individual. Environmen-
eradicated, nearly every infection with smallpox tal conditions interact in complex ways in
virus in susceptible people caused disease (high facilitating the occurrence and spread of infection
pathogenicity), whereas many children infected in human populations.
with poliovirus are asymptomatic (low For example, climate regulates the natural flora
pathogenicity). and fauna and the parasites that can survive and be
transmitted. If the ambient temperature is warm,
the multiplication of salmonellae in contaminated
Virulence
food is accelerated; malaria is transmitted only
Virulence is the pathogenicity of an organism in a where the climate favours survival of Anopheles
specific host. Different strains of the same agent mosquitoes.
may vary in virulence; for example, ‘wild’ strains of Similarly the quality of housing, particularly the
measles and poliovirus are virulent in humans in facilities for washing and waste disposal, influ-
contrast to the attenuated strains used in vaccines. ences the transmission of infectious diseases and
The virulence of particular organisms may vary the presence of vectors. When sanitation is poor,
over time; for example, the virulence of Streptococ- epidemics of diseases such as cholera, plague,
cus pyogenes appears to have diminished over the typhus and typhoid can soon appear. Improved
last 80 years. transportation (whether road, rail or air) between
communities has facilitated social intercourse and
the spread of infective agents. Infection which
Immunogenicity
spreads from person to person does so more rapidly
Immunogenicity is the capacity of an organism to where there is overcrowding, whether in army bar-
induce specific and lasting immunity in the host. racks, slum tenements or village communal huts.
106
PID14 5/21/04 11:41 AM Page 107
107
PID14 5/21/04 11:41 AM Page 108
Infection
of case 1 Transmission from
(primary case 1 (primary) to Transmission from
or index case) case 2 (secondary) case 2 to case 3
Incubation Symptoms Infectivity
Initial
Case 1 Case 2
exposure
Case 1 Case 2
5d 7d
Serial interval (generation time)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Days Figure 14.1 Model of infectious dis-
ease transmission. d, days.
Secondary attack rate This is the number of new gates at a medical conference (Fig. 14.2). The vehi-
cases of a disease arising within one incubation cle by which the Salmonella was transmitted in this
period after the primary case(s). It can be expressed instance was contaminated chicken pieces served
as: number of derived infections/number of sus- at a buffet lunch. The resulting gastrointestinal in-
ceptible persons in the group at risk. fections caused 196 doctors to report symptoms, of
whom 32 were admitted to hospital. Over 1600
doctor-days were lost to the NHS.
Types of epidemic
There are two main types of epidemic: common Example In 1996 the largest UK outbreak of E. coli
source and propagated. O157 food poisoning occurred in Lanarkshire in
Scotland. Over 500 cases were identified and 20
deaths resulted. The outbreak was traced to con-
Common source epidemics
taminated meat from a single butcher. The report
These epidemics result from the exposure of a into the outbreak highlighted concerns about food
group of people to the same source of infection or hygiene and the potential cross-contamination be-
noxious substance. If exposure is simultaneous for tween raw meat and cooked meat products.
all subjects, an explosive outbreak will occur one
incubation period later and the duration of the epi-
Propagated epidemics
demic will depend upon variation between indi-
viduals in the incubation period for the disease. These are due to the transmission of the infectious
Continuous or intermittent exposure of the popu- agent from one person to another, for example
lation to the causal agent produces a more measles or whooping cough. In such cases, the epi-
extended and irregular epidemic curve. The con- demic curve usually shows a gradual rise and de-
trol of such outbreaks depends on the early detec- cline, often with further waves as each successive
tion of the cause and its removal at source. generation of cases infects a new generation.
The speed at which a propagated outbreak
Example In 1986, there was an outbreak of Salmo- spreads depends on the interaction of a number of
nella typhimurium food poisoning amongst dele- factors. These include the opportunity for contact
108
PID14 5/21/04 11:41 AM Page 109
50
45
40
35
Number of cases
30
25
20 Buffet
15
10
5
Figure 14.2 Number of cases accord-
0
ing to time of onset. (From Palmer SR, 12 0 12 0 12 0 12 0 12
Watkeys JEM, Zamiri I et al. J Roy Coll 5 Sept 6 Sept 7 Sept 8 Sept 9 Sept
Phys Lond 1990; 24(1): 26–9.)
16
14 Unvaccinated
12
Number of cases
Vaccinated
10
8
6
4
2
Figure 14.3 Measles epidemic in a
primary school. (From Graham R, Bel- 1 3 5 7 9 11 13 15 17 19 21 23 25 27 1 3 5 7
lamy S, Richardson HJ. Commun Dis February March
Measles cases by date of onset
Rep 1979; number 16.)
between infected and susceptible people which is epidemic may be initiated from a common source
itself influenced both by the density of population and then continue by secondary spread from
and by the level of herd immunity. Obviously, person to person.
person-to-person spread is more likely to occur
where large numbers of susceptible people are Example An outbreak of measles occurred in a pri-
living in close proximity, particularly if there is a mary school (Fig. 14.3). After two index cases in
regular supply of new susceptible individuals early February, there were two epidemic waves at
joining the community, for example nurseries, approximately 10–14-day intervals, i.e. the median
schools, military camps, cruise ships, etc. Different incubation period for measles. The outbreak was
organisms and different strains of the same organ- modified by the fact that many of the children in
ism may vary in their virulence, the speed at which the school had been vaccinated, including some
they spread, the carriage rate in a particular com- who contracted the disease. The attack rate in un-
munity and the duration in individuals. vaccinated children was high (86%) and showed
Remote communities tend to be relatively pro- the typical wave pattern of a propagated epidemic.
tected by their isolation from some infections.
However, once infection is introduced it is liable to
The investigation of outbreaks
spread with exceptional rapidity because herd
immunity is usually low. For example, respiratory Most epidemics are public health emergencies and
infections introduced into isolated island com- require rapid and coordinated action to identify
munities can cause very high morbidity rates. An the cause and to institute effective control meas-
109
PID14 5/21/04 11:41 AM Page 110
ures. It is wise to follow a systematic procedure in subclinical infections are carried out. Phage, sero-
the investigation of outbreaks. logical and other methods of typing of organisms
may help to establish the epidemiological associa-
tion between cases and possible causes (or sources)
Outline of procedures
and to trace the paths of spread of the agent.
The steps described here are not necessarily under-
taken in the sequence given. Enquiries usually pro- Note The application of other epidemiological
ceed simultaneously with the analysis of findings techniques such as the use of case–control studies
and often with interim control measures based on may also be of value in the investigation of out-
early indications of the likely origin of the out- breaks as a means of confirming the validity of a
break. Not all the steps will be relevant in every causal hypothesis. In large outbreaks, investiga-
outbreak and the questions asked must be adapted tions can sometimes be confined to random sam-
to the circumstances. The five main stages in an in- ples of patients and people thought to be at risk.
vestigation are shown below.
Investigation of reservoirs and
vehicles of infection
Stages in investigation
110
PID14 5/21/04 11:41 AM Page 111
45
Recall of chocolate and health warning
40
202 Primary household cases
35 43 Secondary cases
30
Number of cases
20
15
March importation of
10 chocolate
111
PID14 5/21/04 11:41 AM Page 112
Table 14.1 Food poisoning attack rates for delegates eating and not eating specific foods. (From Palmer SR, Watkeys
JEM, Zamiri I et al. J Roy Coll Phys Lond 1990; 24(1): 26–9.)
112
PID14 5/21/04 11:41 AM Page 113
attention to hygiene when storing and selling their premises and equipment, and on facilities for
products. the storage and protection of food from
contamination.
113
PID15 5/21/04 11:43 AM Page 114
Chapter 15
Immunization
114
PID15 5/21/04 11:43 AM Page 115
Immunization Chapter 15
Active immunization
Active immunity to a disease is acquired naturally
after recovery from infection with the causal
organism.
Artificial active immunity can be induced by the
administration of an appropriate vaccine which
stimulates the production in the host of specific
protective antibodies similar to those induced by
natural infection. This provides complete or partial
protection, usually lasting at least for a few years
and in some cases for life. Active immunization is
Figure 15.1 Louis Pasteur (1822–95), chemist and origina- usually given as a planned procedure. It is designed
tor of rabies vaccine. both to protect individuals against infections to
which they may be exposed at some time in the
future and to control the spread of infection in
this purpose has generally been abandoned be- the community (population (herd) immunity, see
cause of the risk of anaphylaxis. The degree and du- p. 105).
ration of the protection afforded depends on the While some types of vaccine produce a prompt
amount of antibody present, but significant pro- and effective response after a single dose, the pro-
tection usually lasts no more than 3–6 months. duction of antibodies after the first dose of other
There are two main types of immunoglobulin in types of vaccine can be slow and inadequate. Mul-
use: human normal immunoglobulin and specific tiple doses at intervals of days or weeks may be re-
immunoglobulin. Human normal immunoglobu- quired to achieve protective levels of antibody.
lin is extracted from the pooled plasma of blood Further reinforcing doses at intervals may be nec-
donors. This confers short-term protection against essary to maintain immunity in later life. Such
a range of infections that are either endemic or for doses (or later natural infection) stimulate an anti-
which immunization is routine practice in the body response which is always more rapid and usu-
donor population, for example measles and hepa- ally greater and more durable than the primary
titis A. Specific immunoglobulin is prepared from response.
the serum of individuals who have recently had a
particular disease or have recently been actively
Types of vaccine
immunized against the infection. Immunoglobu-
lins of this type are prepared for varicella (chicken- Vaccines are of four main types.
pox), tetanus, rabies, hepatitis B and a number of
other infections. These tend to be in short supply
and their use is carefully controlled. This is because
115
PID15 5/21/04 11:43 AM Page 116
Chapter 15 Immunization
116
PID15 5/21/04 11:43 AM Page 117
Immunization Chapter 15
In order to reduce the number of separate injec- ing fits and irreversible brain damage. More
tions, several agents are sometimes incorporated in recently there has been concern about the measles
the same vaccine. For example, the pentavalent vaccine causing inflammatory bowel disease, and
vaccine for infants contains diphtheria, tetanus, lately MMR has been linked to the increase in
acellular pertussis, Hib, IPV vaccine whilst MMR autism. Despite scientific evidence that these risks
includes measles, mumps and rubella vaccines. are small or non-existent, the impact these scares
When giving more than one live vaccine it is con- have on immunization rates can be dramatic and
sidered advisable to give them on the same day in are a threat to the public health.
different sites (unless an approved combined
preparation is used) or to separate them by an
Anaphylaxis
interval of not less than 3 weeks to improve the
immune response. Anaphylactic shock after vaccination is much
feared and can be life-threatening, but it is very
rare. In the 3 years from June 1992 there were 87
Safety and efficacy of vaccines
spontaneous reports of anaphylaxis and no deaths.
No new vaccine is released without extensive safety Over the same period 55 million doses of vaccine
tests in animals and controlled field trials designed were supplied in the UK. Thus the probability of a
to establish the level of efficacy and expected nature vaccinator encountering a case of anaphylaxis is
and frequency of adverse events after vaccination. very small. Nevertheless, adrenaline and appro-
Careful observance of specific contraindications to priate airways should always be at hand and all
each vaccine reduces the risk. Nevertheless, some doctors and nurses responsible for immunization
vaccines frequently give rise to minor reactions, for must be familiar with the management of an ana-
example local oedema at the injection site, tran- phylactic reaction.
sient fever or rash. Serious systemic reactions, espe-
cially neurological conditions, cause great concern
General contraindications to vaccination
but are very rare. To assess their significance, rou-
tine surveillance must be maintained. Careful • Immunization should be postponed if the
records should be kept of all the vaccinations given, recipient has a current acute or febrile illness.
to whom and where, with particulars of the vaccine • Immunization should not be carried out in an
used. Any serious reactions should be reported at individual who has a history of a severe local or
once to the Committee on Safety of Medicines (on a general reaction to a preceding dose.
Yellow Card). Likewise, the continued efficacy of a • Live vaccines should not be given to pregnant
vaccine in controlling a disease should be moni- women.
tored by the analysis of routine morbidity and mor- • Live vaccines should not be given to patients on
tality reports supported, where appropriate, by immunosuppressive treatment or with immuno-
microbiological data and antibody surveys. In the suppression due to disease.
UK, these studies are undertaken by the Communi- • Live vaccines should not be given for at least 3
cable Disease Surveillance Centre (CDSC) of the months after a dose of immunoglobulin or a blood
Health Protection Agency. transfusion.
From time to time the safety of a vaccine comes
under particular scrutiny. This is more likely to be
False contraindications to vaccination
an issue as the danger of the disease in question
fades from consciousness whilst concerns about • Prematurity. Infants who were born prematurely
safety become relatively more important when should be vaccinated at the recommended ages,
considering risk and benefit. Thus in 1976 there i.e. 2 months, 3 months, etc.
was concern about the pertussis component of the • A previous episode of or contact with the disease
DTP triple vaccine with reports of children suffer- concerned, for example measles or whooping
117
PID15 5/21/04 11:43 AM Page 118
Chapter 15 Immunization
cough, is not a contraindication because antibody fits of the vaccine their child is being given. Parents
testing has shown that the clinical diagnosis is fre- should be provided with written information and
quently incorrect. There is no increased likelihood given opportunities to discuss their concerns.
of complications following vaccination in those
who already have natural immunity.
Routine immunization
• Mild illness or chronic disease, for example
asthma, diabetes. The current schedule for routine immunization
• Mother or household member pregnant. recommended in the UK is shown in Table 15.1.
• A stable neurological condition. The exact timing of doses is open to variation.
• Family history of convulsions or adverse While the ages recommended for each vaccine are
reactions. considered to be optimum, it is important to en-
• History of allergy except hypersensitivity to egg. sure as far as possible that all children are vacci-
nated even if they present outside the recommended
age range, unless there are specific contraindica-
Cold chain
tions (see Immunisation Against Infectious Disease,
Appropriate storage conditions are important, par- HMSO, 1996). More up-to-date information about
ticularly for live vaccines, which need to be kept the immunization schedule can be obtained from
cold. Failure to maintain a ‘cold chain’ during trans- the website www.immunisation.org.uk.
port and storage may reduce the efficacy of a vaccine.
The most common problem is the storage facilities in
Diphtheria, tetanus, pertussis, Hib and
many doctors’ surgeries, where the constant use of
polio vaccines
refrigerators for other purposes may mean that the
required low temperatures are not maintained. In the UK it is recommended that primary immu-
nization with diphtheria, tetanus, acellular pertus-
sis, Haemophilus influenzae type b (Hib) and
Consent
inactivated polio vaccine should begin at the age of
Informed consent should be obtained before each 2 months and be completed by 4 months. This is
vaccination is given. This need not be in writing now done using a single pentavalent combination
but parents should understand the risks and bene- vaccine. This ensures protection against these
* A further routine dose of MMR at age 4 years has the advantage of boosting immunity in those who responded poorly
to the first dose and of protecting those who escaped a first dose at 12–24 months. Sometimes the second dose of MMR
is given 3 months after the first dose.
BCG, bacille Calmette–Guérin; DTP, diphtheria, tetanus, pertussis; Hib, Haemophilus influenzae b; MMR, measles,
mumps, rubella.
118
PID15 5/21/04 11:43 AM Page 119
Immunization Chapter 15
diseases as early in infancy as possible. Fears about ent as a commensal organism of the nose and
the safety of pertussis vaccine are now largely dis- throat, it can cause pharyngeal inflammation. Cer-
counted and in any case probably only applied to tain types of C. diphtheriae produce toxins, which
the whole cell vaccine that was used previously. cause the exudation of the classical pharyngeal
Reinforcing doses of diphtheria, tetanus, acellular membrane covering the fauces. The toxins pro-
pertussis and IPV should be given at or shortly duced can also cause cardiac failure and death. The
before school entry. Further doses of tetanus, bacterium is passed from person to person by direct
diphtheria and IPV are required at 15–18 years. contact or inhalation of infected droplets and is
more common in young people. Thus, children liv-
ing in overcrowded housing are particularly sus-
Tetanus
ceptible. Epidemics of diphtheria were particularly
Tetanus has been known to affect humans for cen- common in the 19th and early 20th century and
turies. The disease is caused by the circulation of caused the deaths of large numbers of infants and
neurotoxins that have been produced by the bac- young children. Prior to the Second World War,
terium Clostridium tetani. The toxins cause severe there were around 50 000 notifications each year
muscle spasms which are extremely painful and and 3000 deaths despite the fact that a vaccine
may last for a matter of seconds, or continue for made from the toxin had been available since the
many minutes. As well as causing spasm of the jaw 1920s. The death rate fell dramatically during the
muscles (hence its common name lockjaw), in- war years with the wider use of vaccine, and by
creasingly persistent spasms cause respiratory 1954 the annual number of deaths was in single
failure and death. Clostridium tetani is found as a figures. Diphtheria is no longer endemic in the UK
commensal in the large bowel of many animal and the risk of infection derives only from im-
species, including humans. The bacterium can ported cases or travellers to endemic regions.
form spores that are able to exist in a dormant state
in soil for many decades and when introduced into
Pertussis (whooping cough)
the body by means of a contaminated penetrating
wound may cause local infection with production Whooping cough was described in 1670 by
and release of neurotoxins. A vaccine derived from Thomas Sydenham who called it infantum pertus-
the tetanus toxin was developed in the 1930s and sis (violent cough of children). The Chinese de-
was administered to millions of soldiers in the Sec- scribed it as the hundred-days cough. It is caused
ond World War with great success. Today, tetanus by the highly infectious bacterium Bordetella per-
vaccination is offered to all infants, with booster tussis and is spread by droplet infection. There is a
doses at 5 years and at school-leaving age. A rein- catarrhal stage for 1–2 weeks before paroxysmal
forcing dose of tetanus vaccine may be required coughing develops. In young infants, the charac-
after certain types of high-risk injury or burns in teristic whoop may not be heard and coughing
individuals who were immunized more than 10 spasms may be followed by periods of apnoea.
years previously. Where an individual with such an Complications of whooping cough include
injury has no clear history of having completed a pneumonia, post-tussive vomiting, convulsions,
primary course of tetanus immunization, a dose of and cerebral anoxia with a risk of brain damage.
human antitetanus immunoglobulin should be Most deaths occur in children under 6 months of
given in a different site at the same time as the first age.
dose of a primary course of active immunization. In the UK in the past, whooping cough epi-
demics were seen every 3–5 years. Reduced vaccine
uptake in the mid 1970s following concerns about
Diphtheria
the safety of the vaccine led to an increase in the
Diphtheria is a disease caused by the bacterium incidence of pertussis, but this has been reversed
Corynebacterium diphtheriae. Although often pres- following much improved vaccine uptake rates
119
PID15 5/21/04 11:43 AM Page 120
Chapter 15 Immunization
and increased population immunity in the last few the first conjugate vaccine to be licensed in the UK
years (Fig. 15.2). and was introduced into the immunization sched-
The whooping cough or pertussis vaccine is a ule in 1992, with three doses given at 2, 3 and 4
component of the pentavalent DTaP, Hib IPV vac- months of age. In addition, a ‘catch-up’ pro-
cine given at 2, 3 and 4 months. It is an acellular gramme was arranged for children up to the age of
vaccine produced by inducing antigens to various 4 years. Since then there has been a rapid reduction
relevant proteins. Concern that the killed Bordetel- in morbidity and mortality due to this important
la pertussis vaccine might cause brain damage was pathogen (Fig. 15.3).
allayed following the National Childhood En-
cephalopathy Study (p. 43) which showed that
Poliomyelitis
the risk, if any, was extremely small in relation to
the risk of disease. Children who have had a severe Poliomyelitis was first recognized as a distinct dis-
reaction to a previous dose should not have an- ease in the early 19th century and became known
other dose and children with a developing
neurological illness should also not be vacci-
nated. In these situations further advice should be
sought. Hib vaccine introduced
600
500 Notificati
ons
Haemophilus influenzae type b (Hib)
400
Notifications
Immunization 92%
introduced 81%
200 000
Vaccine
uptake
Cases
150 000 2500
2000
Deaths
Cases
100 000
30% 1500
1000
50 000
500 Figure 15.2 Whooping cough notifi-
Deaths
cations: cases and deaths in England
0 0
1940 1950 1960 1970 1980 1990 and Wales, 1940–90. (Reproduced
Year (1940–90) with permission of the OPCS (Crown
copyright).)
120
PID15 5/21/04 11:43 AM Page 121
Immunization Chapter 15
as ‘infantile paralysis’ because it affected mainly in- hood should receive a primary course: no adult
fants and young children. The first epidemic was should be left unprotected against polio. Further
described in Sweden in 1887. Major epidemics oc- reinforcing doses after that given routinely at
curred in the UK during the late 1940s and early 15–18 years are not usually required except for
1950s (Fig. 15.4). The first vaccine developed travellers to countries where the disease is epidem-
against polio was inactivated virus (Salk) injected ic or endemic and for health care workers in con-
vaccine (IPV) which was introduced for routine im- tact with possible cases of the disease. In addition
munization in the UK in 1956. It was replaced by to the standard general contraindications to vacci-
the live attenuated virus (Sabin) oral vaccine (OPV) nation, polio vaccination should be postponed in
in 1962. Three types of poliovirus are included in patients with vomiting or diarrhoea.
both the oral and killed vaccines.
Polio is frequently asymptomatic but can cause
aseptic meningitis, severe paralysis and death. Measles, mumps and rubella
Paralysis may be as rare as one in 1000 infections vaccine (MMR)
in children and one in 75 in adults. Case fatality in
people with paralysis varies from one in 50 in Measles
young children to one in 10 in older patients.
The IPV (Salk) vaccine prevents the disease in Measles is an acute viral illness, which is highly in-
vaccinated individuals but is less effective than fectious in unvaccinated children. Before the vac-
OPV in creating population immunity because it cine was introduced in 1968, annual notifications
reduces but does not prevent carriage of the virus varied from 160 000 to 800 000 with peaks every 2
in the bowel. The OPV (Sabin) vaccine contains years (see Fig. 3.4). Since then, rates have declined
live attenuated virus, which provides individual with smaller and less frequent epidemics (Fig.
protection and also limits carriage and therefore 15.5). Complications occur in one in 15 reported
transmission of wild virus. Very rarely the disease cases and include convulsions and encephalitis,
has been reported in vaccine recipients or in their otitis media, pneumonia and bronchitis. Measles
non-immune contacts. Vaccine strains of po- is thus, potentially, a major cause of acute and
liovirus may be excreted for up to 6 weeks after vac- chronic ill health in children. Severe illness and
cination. For this reason, oral vaccine has been death are more common in poorly nourished chil-
replaced by IPV to immunize children. Adults who dren and those with chronic conditions, but more
have not been immunized against polio in child- than half the deaths occurred in previously healthy
children. The vaccine is usually given shortly after
the first birthday. Earlier administration is not ad-
vised because the presence of maternal antibody
may interfere with the active immune response.
10 000 Unless a very high proportion of infants are immu-
IPV nized and develop a satisfactory response, there is a
8 000
Notifications
121
PID15 5/21/04 11:43 AM Page 122
Chapter 15 Immunization
Deaths
80
300 000 MMR
60
200 000
40
Figure 15.5 Measles notifications
100 000 20 and deaths following the introduction
0 0 of mass immunization for measles in
1965 1970 1975 1980 1985 1990 1968 and measles, mumps and rubel-
Year (1965–94) la (MMR). (Reproduced with permis-
sion of the OPCS (Crown copyright).)
100 1200
CRS 1000
80
MMR
800
Terminations
60
CRS
600
40 Figure 15.6 Numbers of termina-
400 tions of pregnancies and births with
congenital rubella syndrome (CRS)
20 following the introduction of vaccine
200
Terminations for rubella for girls in 1970 and
0 0 measles, mumps and rubella (MMR)
1971 1976 1981 1986 vaccine for boys and girls in 1988. (Re-
Year (1971–86) produced with permission of the
OPCS (Crown copyright).)
sion to sterility), meningitis and encephalitis nant women and with this the number of rubella-
can occur and justify the use of vaccine to prevent associated terminations of pregnancy. As a conse-
infection. quence, the numbers of children born with
congenital rubella syndrome also declined (Fig.
15.6). However, the selective vaccination of only
Rubella
girls and women allowed continued circulation of
Whilst rubella is a mild disease, maternal rubella wild rubella virus in the community with the con-
infection in the first 8–10 weeks of pregnancy re- comitant risk that a few women who had evaded
sults in fetal damage in up to 90% of infants and immunization, or had failed to mount an adequate
multiple defects are common. The risk of damage antibody response to the vaccine, could be ex-
declines to about 10–20% by 16 weeks’ gestation posed to infection in early pregnancy. Since 1988,
after which fetal damage is rare. Rubella vaccine when MMR vaccine was introduced, both boys and
was introduced in the UK in 1970 and was recom- girls have been offered vaccination against
mended for all girls aged between 10 and 14 years measles, mumps and rubella in early childhood.
of age and for non-pregnant seronegative women This resulted in the virtual elimination of congeni-
of child-bearing age. The application of this policy tal rubella syndrome. The ultimate aim is to elimi-
over the years since 1970 has led to a fall in the nate measles, mumps, rubella and congenital
number of confirmed rubella infections in preg- rubella syndrome. The routine vaccination of girls
122
PID15 5/21/04 11:43 AM Page 123
Immunization Chapter 15
between the age of 10 and 14 years has now been ease in humans. Mycobacterium tuberculosis was a
abandoned but seronegative non-pregnant major cause of morbidity and death in the 19th
women of child-bearing age should continue to be and early 20th centuries. There were over 20 000
given single antigen rubella vaccine. Despite the deaths a year still occurring in the UK in the 1940s.
recent scare that MMR might cause autism it is be- It is an organism that usually causes infection of
lieved that the combined vaccine is safe and more the lung or associated lymph nodes (pulmonary
effective than using single vaccines. tuberculosis), although it can affect any part of the
body (extrapulmonary tuberculosis). Respiratory
infection can lead to localized disease, which is
Meningococcus group C
short lived and gives immunity to the individual,
Neisseria meningitides, the ‘Meningococcus’, is or it may cause progressive lung disease. Transmis-
commonly carried in the nasopharynx. In some in- sion of M. tuberculosis is normally by inhalation of
dividuals, for reasons which are incompletely un- air-borne droplets containing bacilli. The infection
derstood, it can become virulent, and can cause is more common when people are living in over-
septicaemia, meningitis or a combination of the crowded conditions. The disease is also more com-
two. There are a number of antigenically different mon when the population is poorly nourished or
strains of Meningococcus, the most important has a high prevalence of chronic diseases.
strains being referred to as serogroups A, B, C, The death rate in the UK from M. tuberculosis has
W135 and Y. The commonest of these in the UK been decreasing steadily since the mid-19th cen-
has always been group B Meningococcus. Group C tury, the reduction being due principally to im-
Meningococcus started to represent an increasing proved nutrition and living conditions. The
proportion of cases of meningitis, and seemed to advent of effective drug treatment and the wide-
be slightly more virulent than group B. Over a 5- spread use of BCG vaccination accelerated the re-
year period from July 1993 there were an estimated duction (see p. 16). Notifications of new cases of
3151 cases of group C meningococcal disease, tuberculosis reached a low point in 1987. Since
mainly in young children and teenagers, causing then there has been a small rise in the number of
398 deaths and 1768 ITU admissions. Most deaths new cases (in 1992 there were 5798 notifications)
and ITU admissions occurred in teenagers aged whilst the number of deaths each year is about 400.
15–19 years. Work is under way to develop a vac- The rise in tuberculosis in the UK is mainly in the
cine for serogroup B Meningococcus, but at the immigrant population and in the homeless.
time of writing none is available. Polysaccharide Developed in 1921, BCG vaccination was not in-
vaccines have been available for the other strains troduced into general use in the UK until 1953. The
mentioned for some time. These work for a rela- routine use of BCG is controversial. Studies in dif-
tively short time, are ineffective in younger chil- ferent countries have produced conflicting evi-
dren and do not prevent carriage (and therefore do dence of efficacy, the reasons for which are not
not induce herd immunity), so they are not suit- clear. As a result, whilst it is accepted for routine
able for routine use. In 1999 a new group C conju- use in some countries, others have not regarded its
gate vaccine was licensed in the UK, with none of benefits as proven and in some, where the inci-
the shortcomings of the polysaccharide vaccine, dence of tuberculosis has declined to the extent
and this is now routinely given to babies with the that it is no longer seen as cost effective, it has been
primary course of DTaP, Hib, IPV. discontinued.
In the UK, BCG vaccine is given as a routine to
school children at age 10–14 years. It is also recom-
Tuberculosis vaccine (BCG)
mended for tuberculin-negative people in the fol-
Mycobacterium tuberculosis is present throughout lowing categories.
the world, including the UK. Other Mycobacterium • Contacts of cases known to be suffering from
species are also found and occasionally cause dis- active respiratory tuberculosis.
123
PID15 5/21/04 11:43 AM Page 124
Chapter 15 Immunization
• Infants and children of immigrants in whose lation (herd) immunity and to prevent the pre-
communities there is a high incidence of tubercu- dicted epidemic. This strategy was effective in the
losis, who for this purpose may be regarded as con- short term but suggests the need to maintain a pro-
tacts. (Newborn babies who are contacts need not gramme of preschool booster immunization.
be tested for tuberculin sensitivity but should be
vaccinated without delay.)
• Health service staff. This category should include Other vaccinations
doctors, medical students, nurses and any other
staff who may come into contact with patients or Hepatitis B
infected specimens from them. It is particularly
important to test staff working in maternity and Infection with the hepatitis B virus can cause dis-
paediatric departments. The vaccine should not be ease ranging from a subclinical disturbance of liver
given to tuberculin test-positive people because of function, to acute liver necrosis and death. The
the risk of severe reactions. virus is transmitted by blood and semen. Some in-
dividuals may become chronic carriers, and these
individuals are at increased risk of hepatocellular
WHO immunization targets carcinoma. In some countries in south-east Asia
The WHO ‘Health for All by the Year 2000’ targets the virus is endemic, there are many carriers and
announced by the European Office stated that: ‘By hepatocellular carcinoma is a common cause of
the year 2000 there should be no indigenous po- death. Those infected by vertical transmission
liomyelitis, neonatal tetanus, diphtheria, measles from mother to baby, or those infected at a very
or congenital rubella syndrome in the European young age are much more likely to become carriers.
Region.’ The DoH in the UK supported this target In adults, acute liver failure is more common than
and also included mumps and pertussis. To in children but chronic carriage occurs in only 1%
help achieve this, in 1985 the Government set a of cases. Hepatitis B vaccine is produced through
national target of 90% immunization rate for recombinant DNA techniques. The vaccine is
children under the age of 2 years. The Health of the about 90% effective overall; it is slightly less effec-
Nation programme (1992) revised this to a target of tive in those over 40 years of age. The duration of
95% by 1995. Incentives were offered to general vaccine-induced immunity is thought to be 3–5
practitioners to achieve these targets which gener- years. It is recommended for doctors, dentists,
ally have been successful. Most UK regions were re- nurses, midwives, laboratory workers, mortuary
porting immunization rates of 90–95% by 1995. technicians, renal dialysis patients, the sexual part-
However, the targets are more difficult to achieve ners of hepatitis B carriers and infants whose
and sustain in inner cities, and other areas where mothers are carriers. Parenteral drug abusers, pros-
there is a very mobile population. titutes and other sexually promiscuous individuals
The Government also set a target of a 90% re- of both sexes, morticians and embalmers, inmates
duction in the number of notifications of measles of long-term custodial institutions, travellers to
by 1995 compared with around 25 000 notified areas of the world where the disease is endemic and
cases in 1989 (after the introduction of MMR in certain members of the police and other emer-
1988). By 1994, the number had fallen to around gency services judged to be at high risk may also be
10 000 cases, but the relatively low historical considered for vaccination.
immunization rates and modest but significant
vaccine failure rates left a substantial pool of
Influenza
susceptible individuals. This led to predictions of a
large outbreak in 1995. In 1994, the DoH therefore Influenza is an acute viral respiratory illness that
instituted a ‘catch-up’ programme aimed at usually occurs in epidemics during winter months.
school-aged children, to try to improve the popu- In healthy individuals, it is normally a mild illness,
124
PID15 5/21/04 11:43 AM Page 125
Immunization Chapter 15
but can cause significant excess mortality in the protection they require depends both on the coun-
elderly and other vulnerable groups. Unpre- try to be visited and also on the likelihood of their
dictable changes in the virus surface antigens, exposure. Thus, tourists staying in modern urban
which may partially or wholly invalidate immu- facilities are at much less risk from many diseases
nity acquired from exposure to earlier variants, ac- compared to an aid worker or backpacker who may
count for the irregularity of epidemics. If the be living or travelling for extensive periods in re-
antigenic shift is substantial, pandemics, some- mote parts where serious infections are endemic
times with high fatality rates, may occur. There are and living conditions are poor. Health advice
two main types of influenza virus, A and B, each of should include both general protective measures
which can independently cause epidemics. Killed and advice on specific vaccinations.
virus vaccines against both types have been shown Diseases for which vaccinations are available in-
to be protective. However, because of the antigenic clude those passed via the oral/faecal route (hepati-
instability of the influenza virus, the value of the tis A, typhoid, cholera, polio), those spread by
vaccine is variable and unpredictable. Vaccine is inhalation (tuberculosis, meningococcal diseases, in-
prepared from the latest antigenic variants of in- fluenza), those passed by mosquitoes (yellow fever,
fluenza A and B virus, issued by the WHO. These Japanese encephalitis) and others such as rabies.
are for use in the early autumn for people at special Protection against diseases passed by the
risk, such as the elderly (especially those living in oral/faecal route depends principally on good per-
residential institutions), and for those suffering sonal hygiene and the avoidance of potentially
from certain chronic diseases including pul- contaminated food and water.
monary, cardiac and renal disease, diabetes and
other endocrine disorders and conditions requir-
Typhoid
ing immunosuppressive therapy. The vaccine is
not recommended for the control of outbreaks. Vaccination is of value to those who are going to a
Live influenza vaccines are still experimental and country where they may have prolonged exposure
are not in general use in the UK. to potentially hazardous food and water. Both a
killed whole-cell vaccine and a live attenuated oral
vaccine are now available and will give 70–80%
Pneumococcus
protection. Under conditions of continued or re-
Streptococcus Pneumoniae (the Pneumococcus) can peated exposure to infection a reinforcing dose
cause pneumonia, septicaemia, meningitis or should be given every 3 years.
other infections. It is a major cause of illness, espe-
cially in the very young, the elderly, and those with
Cholera
an absent or non-functioning spleen or other
causes of impaired immunity. There are well over Cholera vaccine gives only limited protection (at
80 antigenically different strains. Two-thirds of the most 50%) and is not considered to be of value in
serious infections in adults and 85% of infections epidemic situations. Its use is therefore no longer
in children are caused by just 8–10 capsular types. recommended and it is no longer a legal require-
People at higher risk should be vaccinated. Current ment for entry to any country. The principal need
vaccines include a polysaccharide vaccine, which in cases of cholera is for adequate rehydration.
covers 23 of the capsular types, and a conjugate If properly managed, cholera is rarely life-
vaccine, which covers nine capsular types. threatening in those who are well nourished.
125
PID15 5/21/04 11:43 AM Page 126
Chapter 15 Immunization
vellers. Those travelling for a short period in high- centre is also of great value. Laboratory workers han-
risk areas can be protected by passive immuniza- dling infected material should also be vaccinated.
tion using human normal immunoglobulin.
Vaccination offers good protection and should be
Rabies
offered to those staying in countries where hepati-
tis A is widespread. It may be worth testing for anti- This vaccine is usually given combined with pas-
bodies in those over 50 years of age or with a sive immunization with rabies-specific im-
history of jaundice prior to immunization. munoglobulin only to people bitten by a rabid
animal or by one thought to be infected. It may
also be given prophylactically to those with a high
Meningococcus
occupational risk or who are working in a country
The available vaccine offers protection only in which rabies is endemic.
against Neisseria meningitidis groups A and C,
whilst 70% of infections in the UK are due to group
Smallpox
B. Countries where groups A and C are endemic
and vaccination is recommended include sub- With the success of the WHO smallpox eradication
Saharan Africa, Nepal and northern India. In recent programme the vaccine was no longer thought to
years there have been outbreaks caused by a viru- be necessary. However with the recent threat of
lent strain of group W135 associated with pilgrim- bioterrorism many governments have purchased
ages to Saudi Arabia, such as the Hajj. Travellers stocks and are considering vaccinating key
on these pilgrimages should be offered vaccine personnel.
containing groups A, C, W135 and Y Meningococcus.
Malaria
Yellow fever
Each year, some 2000 cases of malaria are reported
This occurs only in parts of Africa and South in the UK in travellers. Most cases arise from failure
America. Some countries require an international to take, or poor compliance with, malaria chemo-
certificate of vaccination. Avoidance of mosqui- prophylaxis. As yet, there is no effective vaccina-
toes is the most important protective measure (as tion against malaria. It is essential for travellers to
with malaria) but immunization with the live virus areas in which the disease is endemic to take ap-
vaccine obtained from a designated vaccination propriate prophylaxis.
126
PID16 5/21/04 11:44 AM Page 127
Chapter 16
Environmental health
127
PID16 5/21/04 11:44 AM Page 128
vide a safer environment than those that are less published by HMSO in 1980. The report drew
developed. This comes about partly through better attention to the link between these persistent
environmental sanitation, good housing, clean air inequalities and the socioeconomic factors
and other physical conditions. Moreover, better influencing the material conditions of life of
education and the provision of better personal and poorer groups, especially children. Its findings
preventive health services lead to an awareness of were reviewed, updated and substantially con-
the importance of a healthy lifestyle. However, firmed by Whitehead in The Health Divide, pub-
economic development also involves industrializa- lished by the Health Education Council in 1987.
tion and urbanization. The consequences of these Further studies such as the Independent Inquiry
go beyond possible damage to the physical envi- into Inequalities in Health Report chaired by Sir
ronment. They may lead to disruption of old cul- Donald Acheson have failed to demonstrate any
tures, weakening of family ties and the creation of marked narrowing of the divide. The issue was
communities where support for the less competent raised again in the Chief Medical Officer’s report
members has to be provided by welfare services for 2001 available through the DoH website.
rather than through an integrated community
support system.
Causes of pollution
Within any society, the poorest tend to be the
least healthy. The consequences of poverty, such as
poor standards of nutrition, housing, medical ser- Pollution
vices and education, favour high disease rates. The
• Air pollution
converse also applies: those who suffer from dis-
• Water pollution
ease, such as the physically and mentally disabled
• Sewage and waste disposal
and those with chronic ailments, have the least • Ionizing radiation
earning capacity. Persistent disease in an individ- • Industrial accidents
ual can lead to the phenomenon of downward ‘so-
cial class migration’. Since the individual is unable
to retain the more demanding types of job they
Air pollution
may be forced to live in progressively poorer cir-
cumstances in which they are exposed to greater Air pollution in industrial areas arises mainly from
environmental hazards and risks of disease. This combustion of hydrocarbon fuels. The two princi-
can give a further downward twist in a cycle of dep- pal sources are power stations and motor vehicles.
rivation. Urbanization in general leads to the cre- A number of pollutants have been identified as
ation of wealth and in most western countries is causes of ill effects among exposed individuals and
reflected in the better health of the majority. How- populations. These include the following.
ever, the large populations who come to live close • Sulphur dioxide from the burning of coal or
to industrial installations are often exposed to a va- heavy oils. These were the principal sources of the
riety of related health risks. Again, it is the poorest historic London smogs.
and most disadvantaged who are often forced by • Suspended particulate matter. This can be identi-
circumstance to live in these unhealthy environ- fied through filtration methods and is produced by
ments. This affects their health and that of their both vehicle exhaust fumes (mainly diesel) and in-
children. dustrial processes.
Contrary to hopes and expectations, since the • Lead from petrol fumes has been of concern for
inception of the NHS there is little sign that the in- some years, leading to the wider use of unleaded
equalities in health status between social groups in petrol in some countries and prohibition of leaded
the UK is decreasing. Indeed, in some cases they fuel in others.
may be increasing. The facts were documented in a • Hydrocarbons in the atmosphere from both ve-
report, Inequalities in Health (the Black Report), hicle exhausts and industrial processes. The poten-
128
PID16 5/21/04 11:44 AM Page 129
tial carcinogenic action of the complex hydrocar- December 1962, London again experienced atmos-
bons that replaced lead in petrol may be a cause for pheric conditions similar to those in 1952 (tem-
concern. perature inversion). The excess number of deaths
on this occasion was about 700. Whilst the UK now
has few such problems smog is still a cause of ill
Weather conditions
Occasionally, weather conditions arise in which
there is temperature inversion, i.e. a warm air blan-
ket covering a layer of cold air at ground level. In 1000
Deaths
cities, this leads to the trapping and rapid accumu-
lation of pollutants known as ‘smog’. Such high
750 0.75
concentrations of pollutants can cause epidemics
SO2 ppm
Acute health effects
Smoke (mg/m3)
SO2
500
400
300
SO2
Concentration
(mg m3)
200
Concentration
100 (mg m3)
Smoke
Emission (1000 tons)
Figure 16.2 Changes in the emission
0
of smoke and sulphur dioxide and 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968
their concentrations in London air, Year (1958–68)
1958–68.
129
PID16 5/21/04 11:44 AM Page 130
130
PID16 5/21/04 11:44 AM Page 131
131
PID16 5/21/04 11:44 AM Page 132
amounts of ionizing radiation than the general methyl isocyanate gas caused over 2000 deaths and
population. has led to over 500 000 claims for compensation.
The acute effects of exposure to high doses of ra- This was an example of an industrial conglomerate
diation include radiation burns, radiation sickness siting a factory close to a residential population in
and death. The long-term effects following expo- a developing country. Having suffered the horrors
sure to high doses have been shown to include can- of poisoning from the accident, the local popula-
cer (including lung, bone, thyroid and breast tion had neither the medical resources to deal with
cancer) as well as leukaemia, non-Hodgkin’s lym- the disaster, nor the legal resources to seek appro-
phoma, congenital abnormalities and thyroid dis- priate compensation for the accident. Smaller-
ease. Information about ionizing radiation comes scale accidents happen frequently around the
from special events such as by following exposed world and threaten local communities. Prevention
cohorts from Hiroshima, Nagasaki and Chernobyl, in these circumstances not only relies on high stan-
or from people with occupational exposure. In ad- dards in the workplace but also depends on sensi-
dition, the exposure of large numbers of patients to ble planning strategies, which site hazardous
high dosages of X-rays has provided information industrial processes away from residential popula-
about long-term effects. Examples of medical ex- tions.
posure include 40 000 children who in the 1940s
had ringworm treated with X-rays to their scalp
Global health
until their hair fell out, and tuberculosis patients
who had large numbers of chest X-rays. Both The concerns of ecologists about the depletion of
groups showed an excess risk of death from the ozone layer and acid rain have already been
cancer. mentioned. In addition, the increasing proportion
Nowadays in the UK, physicians are interested in of carbon dioxide in the atmosphere seems to be
the effects of ionizing radiation on the general leading to an increase in the global temperature,
population, on people living near nuclear power which potentially could cause melting of the polar
installations or weapon factories and on those at ice caps and a raising of the oceans’ levels. This will
risk due to their occupation. A cluster of cases of threaten many island communities. Global warm-
leukaemia and non-Hodgkin’s lymphoma around ing will also have potential adverse effects on
the nuclear power installation at Sellafield gen- agriculture, which may further exacerbate the
erated particular interest. The cluster has been nutrition problems of many developing countries
investigated using both a case–control study and a causing a deterioration in the health of the world
cohort study, but despite the high relative risks for population. Global warming and other global is-
those children living within 5 km of Sellafield and sues were the focus of a 1992 WHO conference in
for children whose fathers worked at Sellafield a sat- Rio de Janeiro. This led to an acceptance that ac-
isfactory causal explanation has not been found. tion is required by all member countries to reduce
the use of fossil fuels and to stop deforestation and
for joint action to protect the environment. A
Industrial accidents
follow-up Earth Summit in Johannesburg in 2002
The general public are not only at risk from acci- reaffirmed the need for sustainable development as
dents that lead to nuclear radiation exposure but being a central element of the international
are also at risk from accidents involving the trans- agenda. However, the conference was seen by
port or storage of a wide range of chemicals. The ac- many as a failure, with there being few gains in the
cident at Bhopal, in India, involving the release of 10 years since the Rio de Janeiro Conference.
132
PID17 5/21/04 11:45 AM Page 133
Chapter 17
Screening
133
PID17 5/21/04 11:45 AM Page 134
Chapter 17 Screening
The detection of the organism in such people will that carcinoma in situ, the condition that the
be of no benefit to them since they suffer no ad- screening process detects, commonly progresses to
verse consequences. However, it is often in the in- invasive carcinoma. The second is that invasive
terests of the people with whom they come in cervical carcinoma is invariably preceded by a
contact and the wider community that they phase of carcinoma in situ. If either of these as-
should be identified. Ideally once identified they sumptions is invalid, the rationale of the pro-
should be treated, but in some circumstances it is gramme fails. Moreover, it is impossible, for
not possible to eliminate the organism, for exam- obvious ethical reasons, to carry out the long-term
ple typhoid carriers. When treatment is not pos- studies that would be required to test them. Thus,
sible, it may be advisable to isolate the affected the benefits of some screening programmes are
individuals from situations that may expose others theoretical rather than proven, and in future it will
to danger. For example, in an outbreak of methi- be desirable to test the effectiveness of screening
cillin-resistant Staphylococcus aureus wound infec- programmes with randomized controlled trials be-
tions on a surgical unit it would be reasonable to fore their introduction.
screen all the operating theatre and ward staff in an Sometimes, the early detection of disease serves
attempt to identify any healthy carriers. Once only to extend the period of awareness that it is
identified, such carriers would be taken off clinical present without improving the prognosis. Further-
duties until such time as they were proven to be more, in any screening programme, cases with a
clear of infection. long and relatively benign natural history are more
Screening has become increasingly recognized as likely to be detected than those with a rapidly pro-
a major tool in improving population health. This gressive and fatal outcome. The dividends from
has led to the formation of a UK National Screen- screening in these circumstances can be disap-
ing Committee whose remit is to advise ministers pointing, unless the interval between successive
on: examinations is carefully timed to take account of
• the case for implementing new population variations in the natural history of the disease in
screening programmes not presently purchased question.
by the NHS within each of the countries in the Before embarking on any screening programme
UK; it is necessary to consider three further important
• screening technologies of proven effectiveness points.
but which require controlled and well-managed
introduction; and Ethics In contrast to clinical practice, which in-
• the case for continuing, modifying or withdraw- volves the patient asking for the doctor’s aid to
ing existing population screening programmes, in treat established symptoms, in screening pro-
particular, programmes inadequately evaluated or grammes apparently healthy people are invited to
of doubtful effectiveness, quality, or value. present themselves for examination. They have
Their website address is www.nsc.nhs.uk. the right to assume that this will benefit them, or at
The use of screening in disease control involves least will do them no harm.
some important assumptions. Some programmes,
for example, rest on the assumption that a patho- Cost Screening large numbers of people is expen-
logical process can be detected reliably before it is sive and can divert both staff and financial re-
clinically manifest and that, if it is so detected, it sources from other health services. It is essential
can be reversed, arrested, retarded or alleviated therefore to evaluate screening programmes ade-
more readily than if treatment were delayed until quately before they are introduced and to weigh
the patient presented with symptoms. For in- the potential dividends both for the individuals
stance, the cervical cytology screening programme screened and for the health of the community
depends on two assumptions neither of which has against the gains from alternative uses of the same
ever been scientifically proven. The first of these is resources, the so-called ‘opportunity cost’.
134
PID17 5/21/04 11:45 AM Page 135
Screening Chapter 17
135
PID17 5/21/04 11:45 AM Page 136
Chapter 17 Screening
use because up to 90% of people will see their gen- before overt disease is apparent. Without knowl-
eral practitioner over a 2-year period, so that it is a edge of the full natural history from first detection
cost-effective way of reaching a large proportion of by screening to the adverse outcome to be pre-
the population. vented, it is impossible to know what proportion of
those screened positive and treated would have
progressed to clinical disease.
Criteria for screening programmes
Before the introduction and design of a screening
Population to be screened
programme, certain criteria should be considered.
Attention should be paid to the way in which indi-
viduals are recruited to a screening programme.
Criteria for screening Ideally all ‘at-risk’ individuals should be identified
The disease Severity and frequency, natural
and a systematic effort should be made to screen
history them all. This may be possible where relevant lists
The population Identification of risk groups, exist. For example, all newborn babies are known
attitudes to screening and can be screened for phenylketonuria. Those
The test Sensitivity and specificity of the who respond to an ‘open’ invitation to attend for
test, acceptability of the test screening tend to come mainly from self-selected
The treatment Effectiveness of early treatment,
‘health conscious’ groups who are often at least risk
availability and acceptability of
treatment
(low-yield groups) but may also attract those who
The evaluation The cost of the programme, for one reason or another have delayed seeking ad-
screening participation rates vice about existing symptoms (high-yield groups).
Frequently, however, it is individuals in highest-
risk groups who have the poorest response rates
which, unless it can be overcome, limits the poten-
Importance of the disease
tial effectiveness of the programme.
Diseases for which a screening programme is pro-
posed should be important in respect of the seri-
Characteristics of the test
ousness of their consequences or their frequency or
both. Thus, breast cancer is an important disease No screening programme is possible without a sim-
because it is both a common cancer and has a high ple, safe and inexpensive test which can reliably
case fatality rate. Successful intervention would be discriminate between those who have a high or
expected to have a significant impact on mortality low risk of disease. The range of ‘normal’ findings
and morbidity within a population. Another ex- by the test must be known. It should be quick and
ample is congenital hypothyroidism which is a easy to use because the object is to test large num-
rare disease but is worth detecting early both be- bers of people in a minimum time and at a reason-
cause of its serious consequences if untreated and able cost. Unlike clinical practice in which a
because it is eminently treatable. diagnosis and a decision to adopt a particular treat-
ment is normally based on the history, the findings
from physical examination and the results of labo-
Natural history of the disease
ratory investigations, screening is primarily a sort-
The natural history of the disease must be known ing process which depends on the results of a
in order to identify the points at which the disease single test. This imposes particularly heavy de-
is potentially detectable by screening and at which mands on the test.
active intervention is likely to be effective: this The purpose of screening tests is to divide indi-
should be before irreversible damage has been viduals into two distinct groups: test positive and
done. Ideally there should be a long latent period test negative. However, test positive does not al-
136
PID17 5/21/04 11:45 AM Page 137
Screening Chapter 17
ways mean that the individual has the disease or essary mastectomy. The false-negative category
predisposing condition and conversely test nega- presents different problems. Clearly, the individu-
tive does not always mean that they are free from als concerned derive no benefit from the test itself.
the disease or unlikely to contract it. Convention- Furthermore, they may be falsely reassured that
ally, the characteristics of a test are measured in they are disease free, however carefully the test re-
terms of its sensitivity and specificity (Table 17.1). sults are reported to them, and may delay seeking
Sensitivity is the probability that the test will be medical aid when symptoms subsequently appear.
positive if the disease is truly present: a/(a + c).
Specificity is the probability that the test will be
Predictive values
negative if the disease is truly absent: d/(b + d).
In order to measure the sensitivity and specifi- Knowing the false-positive and false-negative rates we
city of a screening test, it is desirable to conduct can ascertain the predictive values of a test:
• Positive predictive value is the probability of truly
follow-up studies over a period of time amongst
having the disease when a screening test is positive: a/a
people who have been assigned to the positive or +b
negative categories by the test but have not been • Negative predictive value is the probability of being
treated. In some diseases, the presumptive evi- disease free when the screening test is negative:
dence of disease in test-positive individuals is so d/c + d
strong, and the potential consequences of failure
to offer prompt treatment are so grave, that it may
be unethical to conduct such an investigation.
Acceptability of the test
However, if a screening programme is initiated
without full knowledge of the test characteristics, The acceptability of a test is an important factor in
problems will arise. Although false negatives will the success of a screening programme. Symptom-
become apparent in due course, these diminish the less patients are less amenable to uncomfortable,
programme’s community benefit. Some of the false time-consuming and potentially harmful investi-
positives will be identified by subsequent investi- gations than those who are seeking medical aid for
gations which precede definitive treatment, but a problem or potential problem that they them-
those that are not so identified and therefore selves recognize.
treated will tend to exaggerate the benefits of the
programme. They will also waste resources.
Effectiveness of early treatment
The problems for patients of being falsely as-
signed to the positive category are that they may be There is no value in detecting a disease early unless
subjected unnecessarily to time-consuming, un- there is an effective treatment that improves the
pleasant and potentially harmful further investiga- prognosis compared with treatment at a later stage.
tions. Occasionally, they may be submitted to Consequently, clinical trials of the proposed inter-
unnecessary and harmful treatments, e.g. women vention are required, particularly because the fre-
in a mammography programme having an unnec- quency of spontaneous regression in the early
stages of disease is often not known. The reversion
of an observation in the presumed pathological
Table 17.1 Measurement of test sensitivity and specificity. range to one in the normal range must not be con-
fused with successful treatment. Furthermore,
Disease status treatments must be assessed in a group that is sim-
Present Absent Total ilar to that which it is proposed to screen. For ex-
ample, if it is demonstrated that early treatment of
Test positive a b a+b
mild hypertension reduces morbidity in a group of
Test negative c d c+d
Total a+c b+d men aged 45–54 years, it cannot be assumed that it
will benefit men aged 55–64 or 65–74 years who
137
PID17 5/21/04 11:45 AM Page 138
Chapter 17 Screening
have similar blood pressures, nor that men in the both direct and opportunity costs of a screening
45–54 age group with higher blood pressures will programme, must therefore be assessed before its
enjoy the same improvement in prognosis. introduction. The calculated cost of a screening
programme to the health services should include
the costs of all the screening tests performed (both
Availability and acceptability
manpower and consumables), the cost of further
of treatment
investigations to discriminate between the true
Clearly, there is little point in the early detection of and false positives, the total treatment costs of
a disease unless the patient is willing to accept and, the positive cases, and the total treatment costs
where appropriate, to sustain treatment at this of the false negatives. The benefits include the
stage. When a patient has symptoms and believes savings on the treatment of cases if they had been
that medical intervention will bring relief, he or allowed to present in the normal way, as well as
she is more likely to accept the treatment and even the social benefits related to potentially lost
endure some side-effects. In offering treatment in earnings or the loss of a parent and the ‘value’ of
the absence of symptoms, the doctor is in a diffi- pain and suffering that would have been incurred.
cult position. Long-term treatment for chronic dis- These are difficult to quantify. It is of course
orders which cause no obvious and immediate unreasonable to initiate a screening programme
disability, for example hypertension, may not al- unless there are sufficient resources (trained
ways be successful because of non-compliance. manpower, hospital beds, technical equipment,
This non-compliance may be because of a misun- etc.) to meet the treatment needs identified by the
derstanding on the part of the patient, or because programme.
of unacceptable side-effects or forgetfulness. For-
getfulness is probably the greatest problem, as pa-
Participation rates
tients have no symptoms to remind them of their
condition. Many screening programmes are only worthwhile
Sometimes, delay in seeking medical aid in the if there is a high acceptance rate amongst those in-
presence of symptoms may be because the patient is vited to participate. Reasons for low uptake can be
fearful of the disease itself or of the treatment which that the screening test is not acceptable to many
he or she thinks may be offered. For example, some people. For example, cervical screening, especially
women may delay seeking advice about breast when carried out by a male doctor, will be avoided
lumps because they perceive mastectomy as a more by some women. This may show up through eth-
immediate and frightening prospect than the con- nic or social class variations in the uptake rate of
sequences of the disease, or because they see the di- screening. Other influences on the success of a pro-
agnosis as a deferred but inevitable death sentence. gramme include the level of knowledge concern-
The success of screening programmes for such con- ing the disease being screened for, the manner of
ditions may also be limited for similar reasons. the invitation (letters from the person’s general
Termination of pregnancy following antenatal practitioner have proved most successful) and the
screening presents a stark example of an interven- accessibility of the screening venue.
tion being absolutely unacceptable to some
women. If a woman would not consider termina-
Appropriate intervals for screening
tion in any circumstances, screening for fetal ab-
normality is useless and should not be carried out. The first round of screening in a population (the
prevalence screen) will have a higher detection rate
and be more cost effective than any subsequent or
Cost of screening
repeat screening (incidence screen). Judging the
Health services increasingly have to recognize that most appropriate interval for repeat screening re-
resources of all types are finite. The cost, including quires detailed research.
138
PID17 5/21/04 11:45 AM Page 139
Screening Chapter 17
There are two important forms of bias that can ethical issues. They were highlighted by the con-
be introduced into screening programmes. cerns regarding the spread of HIV. It was argued
that routine screening of certain groups would
help both in understanding the dynamics of the
Lead time bias
transmission of HIV and in its control. On the
This is the apparent lengthening of survival other hand, as there was no effective early treat-
achieved by earlier diagnosis rather than by ment for HIV infection many believed that the
efficacious intervention. Clearly, early treatment pursuance of such a policy represented an unrea-
will always increase survival time by at least the sonable and unacceptable intrusion on the privacy
length of the interval between the presymptomatic of individuals.
diagnosis and symptomatic recognition: the so- Once a decision has been made that the public
called ‘lead time’. To demonstrate that an inter- good justifies unsolicited invitations for screening,
vention is effective, age-specific death or illness then a number of other ethical issues need to be ad-
rates must be improved. Increases in survival dressed. People who participate in a screening pro-
time can be very misleading when used in isolation gramme have a right to information concerning
as a measure of effectiveness of a screening the conduct of the programme. They should be
programme. aware of the potential disadvantages as well as the
expected benefits and they should be free to enter
or withdraw without coercion. Some programmes
Length bias
can cause unnecessary worry to participants, par-
Interval screening is more likely to identify slowly ticularly if they have a positive test. This is some-
progressive cases whose prognosis is significantly times called the ‘labelling effect’. In addition, some
better than individuals with aggressive disease. individuals, including some who are falsely la-
Consequently, cases identified by screening will belled positive, may suffer harm from either the
appear to have a better prognosis than those who screening test or subsequent treatment. Finally, it
have been identified following the appearance of is necessary to know whether a specific screening
symptoms. In such circumstances, the overall mor- programme is the best way to spend scarce re-
tality in the population may be unaltered because sources. This is a matter of judgement that must be
the screening programme has missed many of the based on good information ideally using a
people with aggressive disease. cost–benefit analysis that takes into account all the
costs and benefits to both the patient and society.
Resources spent on a screening programme may
Ethics
mean that less is available for the provision of
The wider application of screening in the interests health care to others. All of the above ethical ques-
of the public health (whether in an attempt to con- tions should be considered by health staff involved
trol the spread of disease or in order to understand in screening programmes whether they be doctors,
the pathways by which it is spread) raises difficult nurses or managers.
139
PID17 5/21/04 11:45 AM Page 140
PID18 5/21/04 11:46 AM Page 141
Part 3
Health Services
PID18 5/21/04 11:46 AM Page 142
PID18 5/21/04 11:46 AM Page 143
Chapter 18
History and principles
143
PID18 5/21/04 11:46 AM Page 144
144
PID18 5/21/04 11:46 AM Page 145
the standards and nature of care offered within the ing the plague in the 17th century, when most
hospitals. physicians left London together with other mem-
bers of the upper classes. Then, by default, the
apothecaries adopted their new role. By the begin-
The medical profession
ning of the 19th century, the apothecaries were
well established as doctors to all but the upper
classes and were the forerunners of the general
Doctors
practitioner of today. They were not, however, ap-
Until the middle of the 19th century there were three pointed to the honorary staff of the voluntary
types of medical practitioners in the UK: hospitals.
• Physicians
• Surgeons
• Apothecaries
The GMC
Nearly all of these worked almost entirely outside
hospitals. In 1858, physicians, surgeons and apothecaries
were placed on a common register maintained by
the General Medical Council (GMC) which was
Physicians charged by law with control over training and
qualifications. The GMC still regulates the profes-
These were university graduates (in England sion and is responsible for setting standards of
usually from Oxford and Cambridge) who had education, the registration of medical practitioners
then qualified for a diploma of the Royal College of and dealing with complaints about a doctor’s fit-
Physicians (founded in 1518). Their background ness to practice. Only graduates who have under-
was upper class and their practice was mainly gone the prescribed training and passed the
among the upper and merchant classes. They at- appropriate exams can be registered. Members of
tended the voluntary hospitals on a charitable the public who are dissatisfied with the conduct or
basis. performance of any registered medical practitioner
may complain to the GMC who will then investi-
Surgeons gate the complaint and may take action against the
doctor, which can include the removal of the doc-
These originally belonged to the Guild of Barber tor from the register.
Surgeons. After 1745 the surgeons split from the
barbers and in 1800 formed the College of Sur-
geons. The College awarded membership to those Hospitals
who passed their exams. Surgeons usually under-
took an apprenticeship, attended lectures in anato-
Voluntary hospitals
my and walked the wards in one of the teaching
hospitals before qualifying. A few hospitals were established in England by reli-
gious orders during the Middle Ages. These include
St Bartholomew’s and St Thomas’s Hospitals in
Apothecaries
London. They were founded as practical demon-
The third group, who coexisted with the physi- strations of Christian charity to provide care for
cians and surgeons, were the apothecaries. Strictly, the destitute. By 1700, there were fewer than 12
they were apprentice-trained tradesmen whose such hospitals in the whole country, most of which
qualification was in making medicines rather than were in London. The period during which the
in diagnosis and prescribing. Although they did act greatest number of hospitals was built was in the
as doctors, they were breaking the law in pursuing late 18th and 19th centuries. The voluntary hospi-
such activities. They extended their activities dur- tals were supported at first by church funds, chari-
145
PID18 5/21/04 11:46 AM Page 146
table contributions and endowments. They were small population units and in order to produce a
founded principally as places of asylum and rest for viable system, groups of parishes combined to ad-
the physically sick and chronically disabled. They minister the Poor Law legislation. These groups
were staffed by unpaid doctors (consultants) and, were called parish unions. Boards of guardians ap-
in the teaching hospitals, by doctors in training. pointed by the unions were responsible for the
Their location was determined in part by local day-to-day administration of the institutions. The
need and in part by the availability of ‘private prac- Poor Law infirmaries were for the destitute sick and
tice’ for the honorary staff, which was their only were quite unlike hospitals as we know them
source of income. Outside the main teaching cen- today. At first, they did not have any medical staff:
tres, there were other types of voluntary hospital, nursing care was provided by the non-sick inmates
including cottage hospitals, funded locally and of the workhouse. Over the years, the infirmaries
staffed by local general practitioners on a part-time improved, although there was considerable varia-
basis. tion in standards. A feature of much of the Poor
The advent of more sophisticated medical treat- Law legislation and the legislation governing mat-
ments and diagnostic techniques, developed ters of public health was that, although it gave
largely in the London teaching hospitals, made the local authorities discretionary powers to improve
voluntary hospitals become more selective in their the standards and scope of care, it did not place a
admissions. They tended to admit patients who duty on them to do so. In this lies one of the rea-
were of good repute, whose stay was likely to be sons for the maldistribution of health care re-
short, and they avoided admitting the chronically sources in the UK. The Poor Law infirmaries were
sick. The destitute were admitted to the workhouse made over to local government authorities in
where rudimentary medical care was provided. By 1929. They then became municipal hospitals.
the end of the 19th century, because of the increas- From then until the outbreak of the Second World
ing costs of providing a service, they had to intro- War a concerted effort was made to improve stan-
duce a system of payment for those who could dards and staffing. In 1939, the municipal hospi-
afford to pay. The charitable funds were used for tals were grouped with the voluntary hospitals in
those who could not. Most hospitals employed regions as part of the Emergency Medical Service.
‘lady almoners’ whose job it was to establish who
should be subsidized and to what extent. Despite
Other hospitals
the introduction of a semi fee-paying system, the
costs of maintaining these hospitals rose faster There were two other types of public hospitals dur-
than their incomes and they became increasingly ing the first half of the 20th century: fever hospi-
financially embarrassed. At the outbreak of the tals and lunatic asylums. The fever hospitals were
Second World War the Government set up the established to protect the public from infection.
Emergency Medical Service in order to meet Only later were they able to offer treatment.
the needs of the large number of military and Among them were large numbers of tuberculosis
civilian casualties that were expected. This guaran- sanitoria. These were built between the two World
teed money to the voluntary hospitals to meet the Wars and are testaments to the high prevalence of
predicted need. After the war, lack of a secure in- that disease and to increasing faith in its treatment.
come made a return to their former independent The lunatic asylums had a chequered history. Until
status impossible. Most of them were incorporated 1890, the mentally disturbed were cared for in pri-
into the NHS in 1948. vate mad houses (some with appalling reputations)
or in prison or in workhouses (not the workhouse
infirmary which was established for the physically
Municipal hospitals
sick). The Lunacy Act of 1890 placed a duty on
The Elizabethan Poor Law enabled parishes to at- county authorities to provide asylums for those of
tach infirmary wards to workhouses. Parishes were unsound mind. The London County Council built
146
PID18 5/21/04 11:46 AM Page 147
Domiciliary nursing
Domiciliary health services
At the beginning of the 19th century there were
few trained nurses. The need for home nursing was
National Health Insurance Act appreciated by the middle of the century and in
The Poor Law Commission (1909) demonstrated 1887 the Queen’s Institute of District Nursing was
that a lack of early medical advice often resulted in established. The Institute set and maintained stan-
prolonged sickness and consequent poverty. Its dards of practice and coordinated local voluntary
findings led to the introduction of the National committees.
Health Insurance Act in 1911. The important pro-
visions of this Act were as follows.
Mothers and infants
The extremely high maternal and infant mortality
The National Health Insurance Act (1911) in the 19th century led social reformers to look for
• Free medical treatment from a general practitioner ways of preventing this waste of life. Important
whom the insured person was free to choose (provided landmarks were as follows.
the doctor had agreed to participate in the scheme) • Foundation of the Manchester and Salford
• Doctors who participated in the scheme were paid on Sanitary Association, 1862. This organization em-
a capitation basis, i.e. so much per year per person regis-
ployed women to give instruction and guidance to
tered. This was advantageous to the general practitioner
mothers on child rearing. The scheme eventually
as it guaranteed him a regular income for the first time
• Weekly payments to insured persons while sick to en- developed into what is now known as health
able them to maintain minimal living standards visiting.
• The Midwives Act, 1902. This prohibited un-
trained women from practising midwifery.
The scheme was restricted to working men whose • The Maternity and Child Welfare Act, 1918. This
income was below a specified minimum amount. obliged local authorities to provide a medical ser-
It did not include retired persons, the wives of vice for expectant mothers, nursing mothers and
working men or their children. The scheme was children under 5 years of age.
administered by approved Friendly Societies. In • The Midwives Act, 1936. This made local
subsequent years, the National Insurance scheme authorities responsible for ensuring that there
was extended and by 1945 covered the majority of were sufficient midwives to meet the population’s
the population. need.
147
PID18 5/21/04 11:46 AM Page 148
148
PID18 5/21/04 11:46 AM Page 149
149
PID18 5/21/04 11:46 AM Page 150
150
PID18 5/21/04 11:46 AM Page 151
151
PID18 5/21/04 11:46 AM Page 152
Health in England (Acheson, 1988), recommended a cial training in the epidemiology of infection, mi-
return to the old title of ‘public health medicine’ crobiology and clinical infectious diseases. These
and doctors specializing in this field are now called specialists now work for the Health Protection
public health physicians. The report redefined Agency. They are responsible for surveillance of
their role, as outlined below. communicable diseases, investigation of outbreaks
and the instigation of control and preventive
measures.
Role of public health physicians (1988) • Research in epidemiology and public health Re-
search in public health involves a multidiscipli-
• To enquire into all matters which affect the health of
nary approach, which incorporates statistics, social
communities or population groups
• To measure health care needs sciences, health economics and information
• To plan, administer and evaluate services, with par- technology. Specialist areas of research are also
ticular reference to the prevention of disease developing such as health services research,
• To promote health in the community pharmacoepidemiology and global health, as well
• To provide relevant advice to health authorities, cen- as in the more traditional areas concerned with
tral Government and other bodies
study of the causes and prevention of disease.
Most of this work is undertaken in academic
departments.
In addition, it was recommended that there should • Promoting the health of the population This in-
be a cadre of public health physicians with special cludes the design, management and evaluation of
training in communicable disease control. health promotion activities. Health promotion is
Today, public health physicians have four major often initiated at either a national or regional level,
areas of responsibility: and public health physicians with a special interest
• To advise on the provision of health services These are usually involved in the identification of issues,
doctors usually work for primary care trusts and are the design of appropriate programmes and in ar-
responsible for assessing the health care needs of ranging their evaluation.
populations, advising on the purchasing of appro- Public health physicians are often in the fore-
priate services and the evaluation of their effective- front of changes in health services and their roles
ness and efficiency. This requires skills in and responsibilities can change rapidly. For exam-
managing, analysing and interpreting information ple, in the UK the abolition of the district health
and health statistics. Many of these doctors have authorities and the transfer of many of their re-
areas of special expertise such as the provision of sponsibilities to primary care trusts has meant that
acute or special needs services. At the same time, public health physicians have had to rapidly adjust
an understanding of the demographic and social to working in new ways. Whatever the future, it
structure of populations and the dynamics of seems certain that a population perspective of
change is extremely important. health and the expertise of those trained in the rel-
• The control of communicable diseases This is the evant specialties will always be essential in a public
responsibility of public health physicians with spe- health service.
152
PID19 5/21/04 11:47 AM Page 153
Chapter 19
The National Health Service
153
PID19 5/21/04 11:47 AM Page 154
154
PID19 5/21/04 11:47 AM Page 155
• environmental health and a number of other func- control. As a result of a series of enquiries and re-
tions relating to the control of infectious disease. ports by advisory groups in the 1960s, a major re-
organization of the NHS occurred in 1974. The
most important aspects of that reorganization
Early problems
were that the country was divided into a number of
In its early years, the NHS experienced many diffi- regional health authorities (DHAs) within each of
culties and shortcomings. The most significant of which there was a number of area health author-
these were as follows. ities (AHAs), each of which was in turn subdivided
• There had been a gross underestimation of the into districts (DHAs). The authorities were respon-
cost of the service. The estimated first-year cost of sible for the provision of all services other than the
the NHS was £179 million. It actually cost £400 independent contractor services within their geo-
million. graphical boundaries. They thus took over many of
• The NHS inherited many old and small hos- the responsibilities that had been left with the
pitals, which had been built under the Poor Law local authorities in 1948. Where possible, the geo-
provisions. After the Second World War, the graphical boundaries of the health authorities were
Government’s first priority was to build houses aligned with those of the local government au-
rather than hospitals. As a result there was almost thorities. The independent contractors (general
no new hospital building for the first 20 years of medical practice, general dental practice, pharma-
the existence of the NHS. ceutical service, ophthalmic services, etc.) became
• The division of administration of the service be- the responsibility of family practitioner commit-
tween three bodies (hospitals, general practitioners tees. The DHAs had a mainly strategic planning
and public health) resulted in lack of coordination and financial control role, the AHAs planned and
and cooperation. For example, many hospitals managed some of the specialist services whilst the
served several different local authority areas; and DHAs were responsible for the day-to-day manage-
all three divisions of the service were involved in ment. In 1975, the Resource Allocation Working
maternity services. Party (RAWP) was appointed to address some of the
The NHS had failed to correct the long-standing inconsistencies of funding between regions but
inequalities in service provision between different not to advise on the total level of funding for the
parts of the country and between different types of service. Prior to the 1974 reorganization, expendi-
service. The most neglected services were the care ture per person in some regions was only 55% of
of the aged, the mentally ill and people with learn- that of the richest region. RAWP’s main objective
ing disabilities, together with services for the was to ensure that ‘there would eventually be equal
chronically sick and disabled. The northern re- opportunity of access to health care for people at
gions of the country were poorly provided with equal risk’. RAWP did not take into account gen-
hospitals but the areas in and around London had eral practitioner services, local authority services or
an historical, relatively overgenerous provision. those provided by the private sector. Following
The continued geographical maldistribution of fa- RAWP there was substantial redirection of re-
cilities was at least partly due to the fact that there sources to certain less well provided areas.
was inadequate capital investment in new hospi- In 1982, the AHAs were abolished. Some of their
tals in underprovided regions. responsibilities were transferred to the DHAs and
others to the DHAs. In the 1980s, there was a ren-
aissance of the philosophy that optimum effi-
Changes in the 1970s and 1980s
ciency within an organization was best obtained
The original tripartite structure of the NHS was by exposing the organization to market forces. At
seen as a hindrance to the achievement of an inte- the same time, there was a move away from the
grated and balanced service throughout the coun- principle of state ownership. However, the State re-
try. Equally frustrating was the lack of financial tained responsibility for politically sensitive areas
155
PID19 5/21/04 11:47 AM Page 156
156
PID19 5/21/04 11:47 AM Page 157
• transmitting policy decisions to the regions; and provide general practice services, community serv-
• monitoring progress in their achievement. ices such as district nursing and health visiting
The Secretaries of State for Scotland, Wales and (previously provided by community trusts), and
Northern Ireland have similar responsibilities and public health and health promotion, which were
are supported by equivalent administrative depart- previously provided by DHAs. Many are also re-
ments (http://www.doh.gov.uk/). sponsible for managing cottage hospitals. Perhaps
the most important responsibility of PCTs is the
purchasing of most of the secondary care services
NHS Management Executive for the people in their area. Secondary care services
are the specialist services provided by hospital
The NHS Executive is concerned with:
trusts.
• Regional liaison matters
• NHS manpower
In order to purchase services on behalf of their
• NHS finance communities PCTs must understand their health
• NHS support services (building design, maintenance, needs. Consequently they have developed demo-
equipment, etc.) graphic profiles, and local information on morta-
lity and morbidity rates to help them prioritize
their services. PCTs are not restricted to purchasing
services from local providers. In theory, they can
Strategic Health Authorities
purchase services from whomsoever they wish.
These cover an average population of 1.5 million. Some services may be provided by private practi-
The main functions are: tioners or an alternative acute trust. Indeed some
• to support PCTs and NHS trusts in delivering the have explored, with government approval, the
NHS plan in their area; concept of purchasing health care from other
• in building capacity, i.e. ensuring that there are European states. Other services can be provided ei-
adequate staff and facilities available; and ther by specialists or by primary care. For example,
• encouraging improvement in performance by many chronic conditions such as diabetes, care of
the local health agencies. the elderly or care for people with learning disabil-
The chief executive of a strategic health ities can be managed either by primary care teams or
authority is responsible directly to the Department from hospital-based clinics. Decisions on where to
of Health for the performance of the service. He or purchase care, and the balance between types of
she advises on planning and strategic matters and services can vary according to current priorities,
is supported by directors of public health, finance, perceived quality of care, overall outcomes and cost.
planning, personnel and information technology.
The directors of public health of the StHAs are re-
NHS trusts
sponsible for ensuring that there is a public health
network. The network consists of a range of public Most specialist services are provided by NHS trusts
health practitioners, and as a body they help to en- which, although part of the NHS, have consider-
sure that the roles previously carried out by DHAs able autonomy within broad guidelines. Each trust
can continue. has a board of trustees and a chief executive officer.
There are two main types of trusts.
• Acute care trusts. These manage large district hos-
Primary Care Trusts
pitals (or groups of hospitals) and sometimes re-
PCTs are free-standing, legally established, statuto- gional or tertiary services. Some also manage
ry NHS bodies that are accountable to their strate- ambulance services.
gic health authority. They are organizations that • Mental health trusts. These provide mental health
serve a geographically discrete population usually services, including psychiatric inpatient units,
of between 100 000 and 250 000 people. The trusts community mental health services and in some
157
PID19 5/21/04 11:47 AM Page 158
cases forensic services. In some districts, mental the one body. These include the NHS public health
health services are run by acute trusts. staff responsible for the control of infectious dis-
eases, the Public Health Laboratory Service, the
Centre for Applied Microbiology and Research and
National Institute for Clinical
the National Focus for Chemical Incidents. Its
Excellence
functions are:
The National Institute for Clinical Excellence • to advise the government on public health
(NICE) was set up as a Special Health Authority for protection policies and programmmes;
England and Wales in 1999. It is part of the NHS, • to deliver health protection services;
and its role is to provide health professionals and • to provide information and advice to health pro-
the public with authoritative, robust and reliable fessionals;
guidance on current ‘best practice’. The guidance • to respond to new threats to public health; and
covers both individual health technologies and the • to improve knowledge through research and de-
clinical management of specific conditions. The velopment and training.
organization have prioritized their activities to Its current website is http://www.hpa.org.uk.
look specifically at expensive and new technolo-
gies and provide national guidance for health pro-
fessionals. This is one of the ways that the problem Local authorities
of inequalities can be tackled by ensuring there are
Local authorities such as borough and county
national guidelines against which the health ser-
councils have purchasing powers similar to those
vices can be monitored. Their website is
of the health authorities, particularly in respect of
http://www.nice.org.uk.
services for the elderly, some mental health
services and services for those with learning
Commission for Health disabilities.
Improvement
The Commission for Health Improvement (CHI) Special care groups
was established as an independent body that
would review the care provided by the NHS in Eng-
Care of older people
land and Wales (Scotland has its own regulatory
body, the Clinical Standards Board). CHI’s stated Elderly patients often reach the point where they
aim was to address unacceptable variations in NHS are unable to look after themselves at home. This
patient care by identifying both notable practice can be precipitated by an acute illness or injury.
and areas where care could be improved. It also Support for older patients may involve the provi-
published the NHS performance ratings. On 1st sion of a home care worker or admission to shel-
April 2004 the Commission for Healthcare Audit tered accommodation or a residential home — or
and Inspection (CHAI) came into being. It has for those with a serious chronic illness admission
taken over the responsibilities of CHI plus some of to a nursing home may be needed. Long-term
the functions of the Audit Office relating to health. care for older people is now a joint responsibility
The website is www.chai.org.uk. between the health services and the social care
department of the local authority.
Older people who have suffered an illness and
Health Protection Agency
are in need of social support will have both their
The Health Protection Agency (HPA) is a new or- health and social care needs assessed. The social
ganization set up to provide an integrated ap- care departments are responsible for finding ap-
proach to protecting public health. It has brought propriate levels of care for patients being dis-
together a number of different organizations under charged from hospital. Patients are then ‘means
158
PID19 5/21/04 11:47 AM Page 159
tested’ and may have to pay for the social care com- living at home or in accommodation subsidized by
ponent of their care package. The PCT funds their the council. Most health districts have one or more
health care needs. If a patient’s discharge from hos- special teams that liaise closely with the social ser-
pital is delayed after they have been assessed be- vices staff, educational authorities and voluntary
cause a suitable care package is not available in the organizations in order to plan and provide ade-
community the social care department has to re- quate services for this group.
imburse the acute trust until the patient leaves
hospital.
The cost of the NHS
All employed people in the UK pay compulsory
Mental illness
weekly or monthly National Insurance contribu-
The care of mentally ill people requires the provi- tions, which partly finance the NHS. However,
sion of both short- and long-term accommodation most of the cost of the NHS is met from general tax-
and community mental health services. Under the ation. Other finance comes from charges to users,
Community Care Act (1990) these are a joint re- which include dental charges, prescription charges
sponsibility of the local authority and health. The and charges to private patients in NHS hospitals.
responsibility for health services is met through The level of charges to users and the income they
the PCT purchasing appropriate services from yield varies from time to time. In 2003 the total ex-
health care providers. The local authority has a role penditure on the NHS amounted to £63.5 billion.
in purchasing social care in the community such as Most of the expenditure is on hospital services.
the provision of accommodation and day care for About 65% of total hospital expenditure is on
people with a chronic mental illness. salaries. This leaves little room for financial ma-
noeuvre because the numbers of doctors, nurses
and other professional staff cannot easily be ad-
People with learning disabilities
justed to meet short-term changes in need, and
(‘mental handicap’)
cuts in this direction usually lead to a decline in
About four per thousand of the population have services. The differences in the costs of hospitals of
learning disabilities. Under the age of 25 years the different types are largely due to variations in the
majority live at home. In the past, those who were numbers of staff needed to provide the services re-
severely disabled or whose families were no longer quired by various patient groups, for example the
able to provide total care were looked after by clinical staff directly involved in the care of the pa-
health authorities in long-stay hospitals. The NHS tient, and the specialist and technical staff who are
and Community Care Act (1990) transferred to necessary to enable the clinicians to function ade-
local authorities responsibility for maintaining a quately (radiologists, pathologists, radiographers,
register of people with learning disability and for scientists, laboratory technicians, operating the-
the provision of appropriate accommodation. atre staff, intensive care staff, etc.). There are also
They can do this either through local authority differences between hospitals of different types in
hostels or, more commonly, by purchasing accom- the amount of capital investment required in in-
modation and care from private and voluntary or- struments and machinery. As manpower accounts
ganizations. This has meant that the hospitals for the major proportion of hospital costs, the
which traditionally provided care for this group weekly costs are only marginally affected by
have stopped admissions of new long-stay pa- whether or not a bed is occupied or the appropri-
tients. Consequently, the number of people in pri- ateness of its use. Thus, a chronically sick person
vate care far outweighs those accommodated in being cared for in an acute bed costs almost the
hospital. Much of the cost for care of people with same as an acutely sick person in the same bed. Ex-
learning disabilities has been shifted from the tensive misuse of hospital facilities can, if habitual,
health sector to local authorities whether they are prove very wasteful.
159
PID19 5/21/04 11:47 AM Page 160
160
PID19 5/21/04 11:47 AM Page 161
solute manpower deficiencies can arise from a • Planners have miscalculated and have made in-
shortfall in national training programmes, from sufficient numbers of training places available to
net emigration of personnel, or from a need to in- cope with future needs.
crease the staff available to meet rapid advances in • Some areas of the country are more desirable to
diagnostic and therapeutic technology. In the de- live in than others; because of this there may be an
veloped world, absolute deficiencies are uncom- overprovision in some districts and severe deficit
mon; manpower problems result mainly from poor in others.
distribution. This occurs because:
• Certain specialties may be less attractive to a
Facilities
young graduate than others. For example, it has al-
ways been easier to recruit general surgeons and The availability of sophisticated equipment can re-
physicians than it has been to attract people to strict the development of services even if man-
geriatrics and psychiatry. Usually, people tend to power and finance are adequate. This is particularly
train for specialties that interest them rather than important for planning the responses to new tech-
for those that are most needed. nological developments.
161
PID20 5/21/04 11:48 AM Page 162
Chapter 20
Health targets
162
PID20 5/21/04 11:48 AM Page 163
Nation. Similar documents were produced for the targeted lung, breast, colorectal, testicular, cervical
other countries of the UK. In 1998 these targets and skin cancer.
were updated in the White Paper Our Healthier Na-
tion. The targets related to coronary heart disease,
Lung cancer
cancer, accidents and mental health were retained
but the target related to sexual health was given
less prominence. Frequency and trends
Lung cancer is the most common cause of cancer
death in the western world. In the UK, there are
Our Healthier Nation (1998)
38 000 deaths per year. The death rate is higher
The key subjects chosen for action were: amongst men than women. It increases with age
• Cancer (Fig. 20.1). The rates of both registrations of
• Heart disease and stroke
new cancers and deaths in women are increasing
• Mental health
• Accidents
in contrast to the rates in men, which are
decreasing.
6000
5000
Male
4000 Female
Death rate
3000
2000
1000
163
PID20 5/21/04 11:48 AM Page 164
Professional
Managerial
Non-manual
skilled
Social class
Manual skilled
Partly skilled
Unskilled
1600
1400
Death rates per 100 000 in 1998
1200
1000
800
600
400
200
0
4
9
–3
–3
–4
–4
–5
–5
–6
–6
–7
–7
–8
–8
35
40
45
50
55
60
65
70
75
80
85
164
PID20 5/21/04 11:48 AM Page 165
Strategies
Strategies
In the UK, women aged 50–64 years are invited to
attend breast cancer screening every 3 years. Over In March 1998 the Government announced a colo-
one million women are screened annually and rectal screening pilot to assess whether it would
9000 new cancers will be detected. Randomized be worthwhile introducing a national screening
controlled trials indicate that when a high propor- programme for colorectal cancer. Prospective pa-
tion of the eligible women attend screening, up to tients are sent a faecal occult blood testing kit and
a 30% reduction in mortality can be achieved. those who test positive are invited to have a
Consideration is being given to increasing the cov- colonoscopy. The results of the pilot study are ex-
erage by including women up to 70 years of age. pected shortly.
1600
1400
Male
Death rates per 100 000
1200
Female
1000
800
600
400
200
Figure 20.4 Age-specific death rates 0
per 100 000 for men and women due 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84 85–89
to colorectal cancer (England and Age
Wales, 1998).
165
PID20 5/21/04 11:48 AM Page 166
3000
2500 Number
Number
2000
1500
1000
1953 1963 1973 1983 1993 Figure 20.5 Number of deaths from
Year carcinoma of the uterine cervix in
England and Wales, 1953–98.
166
PID20 5/21/04 11:48 AM Page 167
Male
26
Death rates per million
21
Female
16
11
167
PID20 5/21/04 11:48 AM Page 168
25 000
20 000
Death rates per million
15 000 Male
Female
10 000
5 000
30
Death rate per 100 000 population
25
Rate of decline
20
slowing down
15
10
5
Figure 20.8 Death rate per 100 000
0
from stroke in England, 1969–96 (3
1970 1975 1980 1985 1990 1995
Year years’ average adjusted rates). (Repro-
duced with permission of the ONS.)
168
PID20 5/21/04 11:48 AM Page 169
Strategies
United States
The health target set by Our Healthier Nation is to re-
duce the death rate from heart disease and stroke
United Kingdom
and related diseases in people under 75 years by at
least two-fifths by 2010. Intermediate goals have
Sweden
been set by the National Service Framework (NSF)
Japan Mortality per 100 000 for CHD. These have tended to be less specific and
focus on the development of policies based on
France local need. The NSF has encouraged the health
services to develop policies for reducing smoking,
Australia promoting healthy eating and physical activity,
and for reducing overweight and obesity. They
0 100 200 300 400
have also encouraged the targeting of high-risk
populations such as people originating from south
Figure 20.9 Age-standardized mortality rates in coronary
heart disease in different countries.
Asia and those from lower social classes to try and
reduce health inequalities.
Goals for reducing smoking include the intro-
duction of specialist smoking cessation clinics and
4.0
3.5 increasing the number of smokers provided with
Relative risk
169
PID20 5/21/04 11:48 AM Page 170
has remained relatively static (Fig. 20.11). Mental community mental health services. Other sug-
illness is a major cause of morbidity and utilizes gested strategies include encouraging general prac-
considerable health resources both from primary titioners to recognize and treat a higher proportion
care and from the specialist mental health services. of depressed patients, to treat depression with anti-
Around 1% of the population have a major func- depressants at full therapeutic dosages, to elicit an
tional psychosis at any one time and one in seven alcohol history from patients and, when appropri-
people see their general practitioner in any one ate, to take steps to reduce excessive drinking.
year with neurosis (principally anxiety and Many of the issues relevant to improving mental
depression). health are outlined in the National Service Frame-
Prescriptions of antidepressants and anxiolytics work for Mental Health.
are major items within the pharmaceutical budget.
There is great scope to improve the efficiency and
Accidents
effectiveness of prescribing in this area. Currently,
up to 75% of prescriptions for tricyclic antidepres- In England, accidents result in 10 000 deaths per
sants are at subtherapeutic dosages. Thus, patients year and are the most common cause of death in
are exposed to their side-effects with a reduced like- people under 30 years of age. There is substantial
lihood of benefiting from the treatment. The key variation in the numbers and types of accident
target for mental health is to reduce the death rate with age, sex and social class: for example the an-
from suicide and undetermined injury by at least a nual death rate in males aged 15–44 years is four
fifth by 2010 — saving up to 4000 lives in total. times that of females in this age group, whilst chil-
dren from poorer backgrounds are more likely to
die as the result of an accident than are those from
Strategies
better-off families (Fig. 20.12).
The first priority in achieving the goals for mental The pattern of accidents varies enormously with
health will be to improve the local and national environmental conditions and personal factors.
collection of data and to introduce standardized For example, road accidents occur most frequently
assessment procedures. It is hoped that the devel- in the hours of darkness and in winter months,
opment of comprehensive local services based on whereas drownings occur most frequently in the
local joint planning and purchasing arrangements daytime in summer. Alcohol can be a significant
will ensure continuity of health and social care. factor in both. Nearly half of all deaths in children
This includes plans to ensure 24-hour access to are the result of road traffic accidents. Other im-
180
160 Male
140
Death rates per million
120
100
80
60
Female
40
20
0
1940
1944
1948
1952
1956
1960
1964
1968
1972
1976
1980
1984
1988
1992
1996
170
PID20 5/21/04 11:48 AM Page 171
Professional 16.5
Unskilled 82.9
Figure 20.12 Mortality from injury
and poisoning in children 0–15 years 0 10 20 30 40 50 60 70 80 90 100
by social class, 1989–92. (Roberts I, Death rate per 100 000
Power C. BMJ 1996; 313: 784–6).
250
200 Male
Death rates per million
Female
150
100
50
0
Figure 20.13 Age-specific death 5–14 15–24 25–34 35–44 45–54 55–64 65–74 75–84 84+
rates per million in men and women Age
from transport accidents (1998).
portant causes in children include suffocation, (transport, police, local government, etc.) as well as
burns and scalds, falls and poisoning. There is a the DoH in order to achieve substantial change.
sharp peak in the incidence of road accident deaths
in males in the 15–24 years age group (Fig. 20.13).
This peak in road accidents is much less dramatic Main strategies to help reduce accidental
in females. In people over the age of 55 years, the deaths
frequency of accidental deaths increases, particu-
• To reduce the death rate from road accidents
larly as a result of falls in females. Statistics such as • To reduce the death rate from accidents in the home
these help to identify areas of risk in which there is • To reduce the death rate from accidents in the work
a special need for preventive effort. The Our Health- place
ier Nation target is to reduce the death rates from
accidents by at least one-fifth and to reduce the
rate of serious injury from accidents by at least one-
Road accidents
tenth by 2010.
Accident prevention is one area which is a cross- Every year around 3500 people in the UK are killed
government concern. It requires concentrated ef- in road traffic accidents and 40 000 are seriously in-
forts from a number of Government departments jured. Although these figures are better than many
171
PID20 5/21/04 11:48 AM Page 172
of our European counterparts there is still room for ness, drug abuse and domestic violence can all lead
improvement. The UK Government published a to injury and death. One aspect of home safety that
Road Safety Strategy in 2000, which aimed to re- has been highlighted has been fire safety. The fire
duce the number of people killed or seriously in- service has been encouraged to spend more time
jured in road accidents by 40% and to reduce the on prevention of fires and in preventing deaths
number of children killed or seriously injured by through the use of smoke alarms. Another strategy
50%. The strategy includes improvements in road area to reduce accidents and deaths in the home is
design, increased driver training and improved ve- by reducing falls in the elderly. This can be
hicle safety. At the same time the wider health is- achieved by encouraging safer footware, the fitting
sues raised by transport are also being tackled by of handrails and environmental assessments of the
trying to reduce congestion and by promoting safe home of at-risk elderly patients.
walking and cycling.
Safety at work
Home safety
Around 6–7% of all accidents occur in the work
More people die from accidents in the home than place. The targets are to reduce the number of
in any other location. Whilst home safety has been working days lost to work-related injury and ill
improved by better design of domestic appliances health by 30% and to reduce death and major in-
(especially electrical and heating appliances), jury from accidents at work by 10% by 2010. A
flame-proof children’s clothing and good housing third of work place deaths in the UK occur on
design (especially the protection of stairs and bal- building sites and the number of deaths in the
conies) more could still be achieved. Many home building industry has been increasing. The respon-
accidents are due to unsafe behaviour rather than sibility for this lies with the Health and Safety
an inherently unsafe environment. Thus drunken- Executive rather than the Department of Health.
172
PID21 5/21/04 11:49 AM Page 173
Chapter 21
Evaluation of health services
173
PID21 5/21/04 11:49 AM Page 174
Needs assessment
When estimating the need for health services, it is
useful to look first at the prevalence and incidence
Needs of diseases within the population. This coupled
Demand
with the demographic data is the minimal baseline
information required to estimate need. Interpreta-
tion of epidemiological data on need should take
into account factors such as age, gender and eth-
nicity. It is necessary also to take into account
whether an effective intervention exists, and the
availability of the necessary facilities and resources
to meet identified needs.
Supply
Unperceived needs
An individual who is aware of his or her need for
Figure 21.2 A schematic approach to needs assessment.
medical intervention has symptoms or signs that
he or she associates with illness. However, the pro-
fessional worker may detect signs of disease that
take action themselves, for example by going to are amenable to treatment in the absence of such
bed for 2 days because of influenza, or take advice symptoms. This is sometimes incidental to exami-
from a friend or relative. Once they decide that nation for another reason or may come to light
they require medical intervention they make a from screening or health examination surveys.
demand on the health service. The doctor who
sees the patient may or may not accept that
Demands
the problem will benefit from his or her skills. The
only type of ‘need’ that can be measured without The work load of a health service is affected by the
special study is that which creates a demand on incidence of acute diseases and the prevalence of
the service. chronic diseases for which care may be required
174
PID21 5/21/04 11:49 AM Page 175
over a long period. Demand is measured either by continual improvement has been adopted. This re-
monitoring the workload of the service or by spe- quires the structures and processes involved in
cial surveys. Demands on services are not always health care to be continually modified and careful
an accurate proxy for need. They are affected by: monitoring will verify the improvement in out-
• knowledge of the existence of services; comes. This entails the application of the so-called
• local availability of services; and ‘quality cycle’ in which a standard of care is set, the
• sectional pressure for intervention. process of care given is monitored and the out-
Without knowledge of the existence of a facility come is measured, then new standards are
or treatment an individual will not make a demand adopted. Measurement of quality can involve every
for that service. For example. a patient with a patient treated, as is applied, for example, to renal
painful hip who is not aware of the benefits of a dialysis patients and those undergoing chemother-
total hip replacement may perceive a need to have apy, or it can involve a sample of patients.
the pain relieved, yet take no action. The publicity
given to a particular service, for example by a tele-
Clinical governance and audit
vision documentary, inevitably increases the per-
ceived need and therefore increases demand. Clinical governance was a new initiative proposed
Similarly, demand is likely to be greater where a in the 1997 White Paper. It is the way in which
particular specialist facility is available locally and NHS organizations quality assures their services. It
this is known to the local population and their involves putting in place the information, meth-
doctors. ods and systems to ensure good quality care is
Demand for health services can also be thought being provided. It has been described as the frame-
of in terms of health need and supply. Many de- work through which NHS organizations are ac-
mands are for services where there is also need, countable for continually improving the quality of
such as for emergency care or maternity services. In their services and safeguarding high standards of
other cases the demands may be present but the care by creating an environment in which excel-
needs are less obvious, e.g. for tonsillectomy in lence will flourish. This includes action to ensure
young children, circumcision of male infants or that risks are avoided, adverse events are rapidly
breast augmentation in women. Finally there may detected, openly investigated and the lessons
be demand and need but no supply. In some devel- learned. Good practice should be rapidly dissemi-
oping countries facilities such as renal dialysis or nated and systems ought to be in place to ensure
coronary artery bypass grafting are simply not continuous improvements in clinical care. Al-
available. In other circumstances there may be a though given a new name, much of the work car-
need but the technology does not exist. However, a ried out in the name of clinical governance is audit.
scientific breakthrough can rapidly change this, Audit involves a cycle (Fig. 21.3) where firstly the
creating new demands on services. The develop- standard of care to be achieved is agreed and the
ment of sildenafil for the treatment of erectile dys- services are then measured against this standard.
function might be an example. Suggestions are then made as to how this care can
be improved, changes are made and a further audit
is carried out to ensure the predicted changes have
Quality in health care
been achieved. Clinical audit examines the total
Quality is a nebulous concept. It is a function of package of care offered to patients. This may in-
both the service provided and the expectation of volve assessment of the structure and process of
the customer. Thus, as expectations rise, patients’ care as well as outcomes. It may review not only
perception of the quality of care they are receiving medical care, but also nursing care, the physical
is likely to fall. The maintenance of quality has environment and the organization and manage-
often been focused on the elimination of bad or ment of services. Health care can be measured in
unacceptable practice. Increasingly, the concept of terms of seven key parameters.
175
PID21 5/21/04 11:49 AM Page 176
Access
Aspects of care Access involves the assessment of barriers to care in
• Efficacy: does it work? order to ensure that people obtain the treatment
• Effectiveness: how well does it work? they need when they need it. Barriers can include
• Efficiency: is this the best way of doing it? cost, waiting lists, location of the service or the
• Equity: is it fair? need to convince a general practitioner of need.
• Accessibility: can everyone use the services? Often these barriers are only identified by asking
• Acceptability: is it what they want?
patients directly.
• Appropriateness: is it what they need?
Acceptability
Efficacy
Some services may not be used because of the way
Efficacy is the measure of the capacity of an inter- they are provided. Issues such as privacy, the gen-
vention to produce a desired effect. der or attitude of staff, and the setting of the ser-
vice can influence the utilization of health
services. These factors are often only discovered
Effectiveness
through patient questionnaires.
Effectiveness involves assessing clinical outcomes
of health care such as mortality rates and survival
Appropriateness
times. A treatment must show an improvement in
clinical outcome, ideally through use of random- Any assessment of health services must measure
ized controlled trials, in order to be considered whether the needs of the population are being met.
effective. This requires constant assessment of need and
audit of the structure of health services as well as
monitoring such indicators as waiting lists.
Efficiency
Efficiency involves the assessment of the costs of
Accreditation
services. The most efficient service will produce the
desired outcome at the lowest cost. In some countries, the concept of accrediting or-
ganizations that meet certain quality standards is
being adopted within the health service. Accredita-
Equity
tion is common practice, for example in the food
Equity involves assessing differences in the needs industry, and is sometimes applied to hospital lab-
of those receiving care, differences in their treat- oratories. The assessment includes standards of
176
PID21 5/21/04 11:49 AM Page 177
practice (including the training of staff), adherence outcome. Examples of outcomes that can be com-
to protocols, validity and reliability of diagnostic pared include the cost per patient successfully
testing, safety standards, etc. The concept of ac- treated or cost per life saved. The disadvantage of a
creditation is applied to the whole range of hospi- cost-effectiveness analysis is that it may not reveal
tal and community services in the USA and other positive or negative effects of compared
Australasia: accreditation provides purchasers with treatments other than those that have been
an assured quality standard which is taken into ac- recorded and which are the subject of the analysis.
count when negotiating contracts. In many cases,
accredited hospitals are rewarded by being paid a
Cost–benefit analysis
higher rate for the services they provide. UK
purchasers use similar processes to monitor the Cost–benefit analysis compares two or more treat-
quality of services provided by contractors. ments or services, by placing a value (usually mon-
etary) on all the accrued costs and on all the
benefits. Thus, when considering a treatment, the
Health economics
benefits may include added years of life. This is as-
The cost of health services has been one of the signed a monetary value often based on future po-
Government’s primary concerns since the begin- tential earnings of the individual. It may also
ning of the NHS. The demand for health services include the costs of continuing care or treatment.
can be almost infinite but resources are always lim- This form of economic analysis allows purchasers
ited. In this circumstance choices have to be made. to compare many different treatments to help
Health economics can help decide which are the them decide which is the best buy. The disadvan-
best choices. Often in the past management of tage is that it tends to discriminate against the eld-
health care has focused on the principle of cost erly and those with a low earning potential such as
containment. This has the disadvantage that con- the physically disabled and those with learning
tainment may actually reduce efficiency. Attempts disabilities because the benefits are not easily
to save money by undertaking fewer hip replace- measurable in monetary terms.
ments does not reduce the fixed cost of the the-
atres, or the need to pay the nurses and surgeons.
Cost–utility analysis
So whilst overall costs may reduce, the cost per hip
replacement will increase. Economists look at ways Cost–utility analysis accounts for all the costs of
of better using the money that is allocated to the comparable treatments, but measures the benefits
health services. They use the term ‘opportunity in a common unit (other than money). One
cost’ — in a cash-limited health service, money common unit that has been used is the quality-
spent on one aspect of care means that another adjusted life year (QALY). A QALY combines the
service cannot then be funded. Health economists quantity of life gained with an adjustment for qual-
also use several methods of comparing costs and ity of life. This allows direct comparison between
outcomes. treatments but is less discriminatory against those
with limited earning capacity than cost–benefit
analysis. Other measures of the benefits of medical
Cost effectiveness
care have been developed; for example, in 1993 the
Cost effectiveness measures the cost of one or more World Bank adopted a new and more sophisticated
treatments or services in comparison to a single unit, disability-adjusted life years (DALY).
177
PID21 5/21/04 11:49 AM Page 178
PIDAPP 5/21/04 11:50 AM Page 179
Appendices
Further Reading and Useful Websites
PIDAPP 5/21/04 11:50 AM Page 180
PIDAPP 5/21/04 11:50 AM Page 181
Appendix 1
Suggested further reading
Armitage P, Berry G. Statistical Methods in Medical Joint Committee on Vaccination and Immunisa-
Research. Oxford: Blackwell Science, 2001. tion. Immunisation Against Infectious Diseases.
Ashton J, Seymour H. The New Public Health. Open London: HMSO, 1996.
University Press, 1990. McKeown T. The Role of Medicine. Oxford: Basil
Beaglehole R, Bonita R, Kjellstrom T. Basic Epidemi- Blackwell, 1980.
ology. World Health Organization, 1993. Naidoo J, Wills J Health Promotion Foundations for
Benenson AS. Control of Communicable Disease in Practice, UK. Second edition. Bailliere Tindall, 2000.
Man. American Public Health Association, 1997. Pereira-Maxwell F. A–Z of Medical Statistics. Oxford
Bland M. An Introduction to Medical Statistics. University Press, 1998.
Oxford: Oxford University Press, 2000. Rose G. The Strategy of Preventive Medicine. Oxford
Detels R, McEwen J, Beaglehole R, Tanaka H. Oxford Medical Publications, 1992.
Textbook of Public Health. Oxford University Sackett DL, Straus S, Richardson S, Rosenberg W,
Press, 2002. Haynes RB. Evidence-Based Medicine. How to Prac-
Donaldson RJ, Donaldson LJ. Essential Public tice and Teach EBM, 2nd edn. London: Churchill
Health Medicine, 2nd edn. Petroc Press, 2000. Livingstone, 2000.
Drummond MF, Maynard A. Purchasing and Provid- Scambler G, ed. Sociology as Applied to Medicine.
ing Cost-Effective Health Care. Churchill Living- London: W.B. Saunders, 1997.
stone, 1997. Townsend P, Davidson N. Inequalities in Health
Greenhalgh T. How to Read a Paper. BMJ Publica- (the Black Report). London: Penguin Books,
tions, 2001. 1982.
181
PIDAPP 5/21/04 11:50 AM Page 182
Appendix 2
Useful websites
182
PID INDEX 5/21/04 11:23 AM Page 183
Index
A association
abortion 57–8 chance, distinguishing causes and determinants
induced (terminations) 80, 138 from 8–10
following antenatal screening strength of 9
138 Association of Public Health Observatories 182
spontaneous 79–80 attributable risk, in cohort studies, calculation 11
Abortion Act (1967) 57 audit 175–6
Abortion Act (1990) 57 autonomy, personal, rights to 102
acceptability of service 176
accessibility of service 176 B
screening test 138 Bandolier 182
treatment following screening 138 barrier contraceptives and cervical cancer 42–3
accidents 170–2 BCG vaccination 15, 114, 116, 123–4
industrial see industry beneficence 102
prevention 170–2 Bentham, Jeremy 149
accreditation 176–7 Beveridge Report 153–4
accuracy of test 31 bias 84
acid rain 130 avoidance 29–30
Action on Smoking and Health 99 in cohort studies 33
acute care trusts 157 in denominators 29–30
adjuvants, vaccine 116 in numerator data 28–9, 66
aetiology see cause in sampling 27–8
age 21 in screening programmes 139
death rates related to 67–8, 73–4 systematic 30
fertility rates related to 76–7 births
standardization of rates for 67–8 control methods see contraception
structure of population 71 information 53
susceptibility to infection and 104 rates 64, 76–7
AIDS/HIV virus 9, 104 illegitimate 79
notifications 60 seasonality and mental illness 18
screening 139 stillbirths 22, 53, 56–7, 64
air pollution 128–30 Births and Deaths Registration Act (1968) 53
alcohol abuse, health promotion Black Report 5, 128, 151
99–100 blinding 85
Alma Ata Declaration on Primary Care blood pressure, elevated see hypertension
151 body mass index 63–4
anaphylaxis after vaccination 117 Bordetella pertussis see pertussis (whooping cough)
animals immunization
as models of disease 8 breast cancer 136, 164–5
as reservoirs of infection 104, 110 case–control study
antenatal screening, termination following case selection 38
138 confounding variables 41
antibodies fluoroscopy and 36
passive immunization with 114–15 oral contraceptives and 10
to vaccines 114–15 British Heart Foundation 98
antidepressants 170 British Nutrition Foundation 100
antigenic stability of pathogen 106 bronchial carcinoma
apothecaries 145 mortalities 16
appropriateness of service 176 smoking and 8, 41–2
area health authorities 155 see also lung cancer
asbestos 9 buildings, hospital, expenditure 159
183
PID INDEX 5/21/04 11:23 AM Page 184
Index
184
PID INDEX 5/21/04 11:23 AM Page 185
Index
capture procedure 61 E
grouping 63–4 Earth Summit 132
numerator, errors and bias in 28–9, 66 education see health education
presentation, inconsistency 25 effectiveness
routinely collected, problems with 25 cost-effectiveness 177
web sources 51 screening 135
see also health information; records efficacy
databases, general practice 59 definition 176
deaths see mortalities (rates) vaccines 117–18
demands on services 174–5 efficiency of NHS 176
demography, medical 69–81 elderly
fertility see fertility (rates) accidents 172
fetal loss and infant mortality 79–81 care, local authority responsibilities 158–9
mortality, reasons for decline 74–5 enteric infections see gastrointestinal infections
populations and growth rates 70–2 environment
transition 72–4 health services concerned with 127–32
denominator error 66 infection and 106
Department of Health, responsibilities workplace safety 172
156–7 Environmental Health Officers 113, 150
Department of Statistics 182 epidemics 18–19, 107–11
deprivation, cycle of 128 common source 108
derived infection 107 herd immunity affecting periodicity 105
descriptive studies 14–23 investigation 109–11
data analysis 15–23 propagated 108–9
use of 14–15 types 108–9
determinants 7 epidemiological studies see studies/surveys
chance association and, distinguishing equity 176
8–10 errors (in surveys) 28–31
descriptive studies in discovering 12 assessment 30–1
developing countries, demographics 72–4 avoidance 29
diabetes 25 in health information 65–7
prevalence 26 random 30
diagnosis/detection 5 systematic 30
criteria used, variations 24–5 Escherichia coli 157 food poisoning 108, 112
diet ethics
heart disease and 168–9 clinical trials 47
promoting healthy 100 health promotion 101–2
see also nutrition screening programmes 134, 139
diphtheria 119 ethnicity 22
diphtheria/tetanus/pertussis (DTP) vaccination 117, evidence-based medicine 82–7
118–20 challenges 86–7
direct standardization 68 critical appraisal 83–4
discreet quantitative variables 63 systematic reviews 85–6
discrimination 30–1 randomized controlled trials 84–5
disease/ill-health search strategies 82–3
cause see cause strengths and weaknesses 87
distribution 9 exercise 100–1
indices of 63–8 expenditure on NHS 159
prevention exposure (to agents or experience)
intervention strategies 94–5 radiation
principles 93–4 leukaemia and 36, 44, 132
see also specific diseases lymphoma and 44
divorce and fertility 77 special, groups with 32–3
doctors/physicians
history of profession 145 F
public health 151–2 family planning see contraception
see also general practitioners fertility (rates) 76–9
Doll, Richard 34 factors affecting 76–8
domiciliary health services 147, 172 patterns, changes in 78–9
double-blind trial 46 total-period 79
185
PID INDEX 5/21/04 11:23 AM Page 186
Index
186
PID INDEX 5/21/04 11:23 AM Page 187
Index
187
PID INDEX 5/21/04 11:23 AM Page 188
Index
188
PID INDEX 5/21/04 11:23 AM Page 189
Index
189
PID INDEX 5/21/04 11:23 AM Page 190
Index
190
PID INDEX 5/21/04 11:23 AM Page 191
Index
191
PID INDEX 5/21/04 11:23 AM Page 192