Vous êtes sur la page 1sur 200

PIDPR 5/21/04 11:21 AM Page i

Lecture Notes: Epidemiology and

Public Health Medicine
PIDPR 5/21/04 11:21 AM Page ii
PIDPR 5/21/04 11:21 AM Page iii

Lecture Notes
and Public Health
Richard Farmer
Professor of Epidemiology
Postgraduate Medical School
University of Surrey
Stirling House
Surrey Research Park
Surrey, UK

Ross Lawrenson
Dean of Medicine & Professor of Primary Health Care
Postgraduate Medical School
University of Surrey
Stirling House
Surrey Research Park
Surrey, UK

Fifth Edition
PIDPR 5/21/04 11:21 AM Page iv

© 2004 by Blackwell Publishing Ltd

Blackwell Publishing, Inc., 350 Main Street, Malden, Massachusetts 02148-5020, USA
Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK
Blackwell Publishing Asia Pty Ltd, 550 Swanston Street, Carlton, Victoria 3053, Australia

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

Library of Congress Cataloging-in-Publication Data

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

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.

Commissioning Editor: Vicki Noyes

Editorial Assistant: Nic Ulyatt
Production Editor: Fiona Pattison
Production Controller: Kate Charman

For further information on Blackwell Publishing, visit our website:

PIDPR 5/21/04 11:21 AM Page v


Preface, vi 13 Health promotion and health

List of Abbreviations, viii education, 96
14 Control of infectious disease, 103
15 Immunization, 114
Part 1 Epidemiology
16 Environmental health, 127
1 General principles, 3 17 Screening, 133
2 ‘Cause’ and ‘risk’ and types of
epidemiological study, 7
Part 3 Health Services
3 Descriptive studies, 14
4 Surveys, survey methods and bias, 24 18 History and principles, 143
5 Cohort studies, 32 19 The National Health Service, 153
6 Case–control studies, 38 20 Health targets, 162
7 Intervention studies, 45 21 Evaluation of health services, 173
8 Health information and sources of data, 51
9 Indices of health and disease, and
Appendices: Further Reading and
standardization of rates, 63
Useful Websites
10 Medical demography, 69
11 Evidence-based medicine, 82 Appendix 1: Suggested further
reading, 181
Appendix 2: Useful websites, 182
Part 2 Prevention and Control of Disease

12 General principles, 91 Index, 183

PIDPR 5/25/04 2:22 PM Page vi


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-

PIDPR 5/21/04 11:21 AM Page vii


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
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

PIDPR 5/21/04 11:21 AM Page viii

List of Abbreviations

AHA Area Health Authority

AIDS acquired immune deficiency syndrome
BCG bacille Calmette—Guérin (vaccine)
BMA British Medical Association
CCDC Consultant in Communicable Disease Control
CDSC Communicable Disease Surveillance Centre
CEHO Chief Environmental Health Officer
CHAI Commission for Healthcare Audit and Inspection
DHA District Health Authority
DoH Department of Health
DTP diphtheria/tetanus/pertussis (vaccine)
EBM evidence-based medicine
FHSA Family Health Service Authority
GMC General Medical Council
GPRD General Practic Research Database
HEA Health Education Authority
HES hospital episode statistics
Hib haemophilus influenzae type b (vaccination)
HIV human immunodeficiency virus
HPA Health Protection Agency
HSE Health and Safety Executive
ICD International Classification of Diseases
IHD ischaemic heart disease
IPV injected polio vaccine
ITT intention to treat
MMR measles/mumps/rubella (vaccine)
MRC Medical Research Council
NHS National Health Service
NHSME National Health Service Management Executive
NICE National Institute for Clinical Excellence
OPCS Office of Population Censuses and Surveys
OPV oral polio vaccine
PCT primary care trust
PHLS Public Health Laboratory Service
PMR perinatal mortality rates
RAWP Resource Allocation Working Party
RCT randomized controlled trial
RHA Regional Health Authority
SARS severe acute respiratory syndrome
SMR standardized mortality ratio
STD sexually transmitted disease
WHO World Health Organization

PID1 5/21/04 11:22 AM Page 1

Part 1
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

PID1 5/21/04 11:22 AM Page 4

Chapter 1 General principles

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.

Health care needs

Figure 1.1 ‘Mother with deformed child’ by Francisco José
Goya y Lucientes. (By courtesy of the Cliché des Musées Apart from its significance in day-to-day clinical
Nationaux, Paris.) practice, an unbalanced picture of disease inci-

PID1 5/21/04 11:22 AM Page 5

General principles Chapter 1

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

PID1 5/21/04 11:22 AM Page 6

Chapter 1 General principles

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.

PID2 5/21/04 11:24 AM Page 7

Chapter 2
‘Cause’ and ‘risk’, and types of
epidemiological study

have other exposures or characteristics that act

with the effects of tobacco smoking to cause the
The principal uses of epidemiology in medicine disease. Venous thrombosis is caused by a combi-
have been described in Chapter 1. These are: nation of stasis, vessel wall damage and a hyperco-
• the investigation of the causes and natural his- agulable state (Virchow’s triad). An individual may
tory of disease, with the aim of disease prevention have a disorder that results in a hypercoagulable
and health promotion; and state (for example, inherited disorders of the coag-
• the measurement of health care needs and the ulation system such as factor V Leiden) yet never
evaluation of clinical management, with the aim have a venous thrombosis because he or she never
of improving the effectiveness and efficiency of experiences the concurrence of vessel wall trauma
health care provision. and stasis necessary to produce the disease. Thus,
Both involve the important and fundamental con- the risk of deep venous thrombosis in such indi-
cepts of cause and risk. The concept of cause must viduals is measurably increased but it is not
be distinguished from the notion of association. inevitable.
Not all factors that are statistically associated with Although the cause of a disease is always statisti-
the occurrence of disease are causes. They also in- cally associated with its occurrence a statistical
clude so-called ‘determinants’, confounding vari- association cannot be taken as proof of cause.
ables and factors associated by pure chance. Sometimes an event or exposure is associated with
both the occurrence of the disease and another ex-
posure which is statistically associated with the dis-
Concept of cause
ease. This is called confounding. For example, if
• A cause is an event, characteristic or condition one were to investigate the association between
that precedes the disease and without which the alcohol consumption and coronary heart disease,
disease could not have occurred. The event may be smoking would be a confounding exposure
exposure to a microbe, chemical substance, physi- because smoking tends to be positively associated
cal trauma, radiation or other exposure. Many dis- with alcohol consumption and is also a cause of
eases do not have a single cause and thus exposure coronary heart disease. If the presence of con-
to a ‘causal agent’ does not inevitably result in dis- founding is not allowed for in such a study then it
ease. For example, smoking tobacco is a cause of might result in the misleading conclusion that
lung cancer; however, not all individuals who alcohol is directly associated with coronary heart
smoke will develop lung cancer. Those who do will disease.

PID2 5/21/04 11:24 AM Page 8

Chapter 2 ‘Cause’ and ‘risk’

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

PID2 5/21/04 11:24 AM Page 9

‘Cause’ and ‘risk’ Chapter 2

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
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.

PID2 5/21/04 11:24 AM Page 10

Chapter 2 ‘Cause’ and ‘risk’

a hypothesis. For example, in the mid-19th cen-

tury, John Snow suggested that cholera was caused
The same association between a disease and a sus- by an invisible agent in water. The epidemiological
pected causal agent should be found in studies of data were entirely consistent with the hypothesis
different populations. Failure to find consistency but the cholera vibrio and its mode of spread had
may be explained by differences in study design. yet to be discovered.
Caution is needed before rejecting a causal hy-
pothesis in such circumstances. For example, stud-
Experimental models
ies designed to test the hypothesis that carcinoma
of the breast is causally associated with exposure to The disease can be reproduced in experimental
oral contraceptives have produced conflicting re- models with animals. The fact that exposure to an
sults. Some appear to demonstrate that women ex- agent can produce a disease in animals similar to
posed to oral contraceptives over long periods of that seen in humans gives credence to a causal
time have an increased risk of breast cancer; others hypothesis. However, failure to produce the dis-
do not support this hypothesis. Careful review of ease amongst animals cannot be used as evidence
the studies reveals differences in the criteria for the to reject the hypothesis. For example, some
selection of cases and in the analytic techniques microorganisms are pathogenic in humans but not
used, which may explain the apparently conflict- usually in animals (e.g. measles virus); others are
ing results. A causal hypothesis can be regarded as pathogenic in animals but not usually in humans,
supported only when there is a general consistency and only a minority are normally pathogenic in
of findings from studies conducted in the same both.

Preventive trials
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
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.

Biological plausibility Types of risk

The association between the disease and exposure Absolute: incidence of disease in any defined
to the suspected causal agent should be consistent population
with the known biological activity of the suspected Relative: ratio of the incidence rate in the exposed
agent. Sometimes an association is observed before group to the incidence rate in the non-exposed
the biological process is identified. The fact that
Attributable: difference between the incidence rates in
there is no known biological explanation for an as-
the exposed and non-exposed groups
sociation should not on its own lead to rejection of

PID2 5/21/04 11:24 AM Page 11

‘Cause’ and ‘risk’ Chapter 2

Absolute risk in cigarette smokers = 5.16 per 1000

Absolute risk
Absolute risk in non-smokers = 0.55 per 1000
This is the most basic measurement; it is the inci- Relative risk in cigarette smokers
dence of a disease in any defined population. The = 5.16/0.55 = 9.38
denominator can be the whole population or a Attributable risk of cigarette smoking
subpopulation defined on the basis of an exposure. = 5.16 – 0.55 = 4.61 per 1000
The absolute risk in an exposed population taken
in isolation is often not a very useful index. To be This indicates that smokers were 9.38 times more
meaningful it has to be compared with the risk in likely to die during the 7-year period than non-
an unexposed population. smokers and that the additional risk of death car-
ried by smokers compared with non-smokers was
4.61 per 1000 people per 7 years. The confidence
Relative risk
with which these findings can be applied to the
This is the ratio of the incidence rate in the exposed general population is determined in part by the
group to the incidence rate in the non-exposed similarity of the two groups in respect of attributes
group. It is a measure of the proportionate increase other than their smoking habits, in part upon
(or, if the agent is protective, the decrease) in dis- whether the smokers are representative of the
ease rates of the exposed group. Thus, it makes al- whole population of smokers and in part upon the
lowance for the frequency of the disease amongst sizes of the samples investigated. If the sampling
people who are not exposed to the supposed harm- was truly representative, the proportion of deaths
ful agent. It is important to consider the relative in smokers that would be eliminated by cessation
risk in conjunction with the absolute risk. For ex- of smoking is the ratio of attributable to absolute
ample, a relative risk of 3 (people exposed have risk (4.61/5.16 = 89%). This is known as the attrib-
three times the risk of those not exposed) can cause utable fraction.
concern. However, if the absolute risk is 1 in
100 000 it is less worrying than if the risk is 1 in 100.
Types of epidemiological study
There are four broad types of epidemiological
Attributable risk
This is the difference between the incidence rates • descriptive
in the exposed and the non-exposed groups, i.e. it • cohort
represents the risk attributable to the factor being • case–control
investigated. • intervention.
The use of these measures of risk can be illustrat- They serve different purposes. None of them is en-
ed with data collected during the course of a cohort tirely clear cut and it is not profitable to try to clas-
study which compared mortality amongst ciga- sify each and every study within these classical
rette smokers with non-smokers during a 7-year types. Frequently the detailed investigation of a
period (Table 2.1). disease involves undertaking several studies of dif-
ferent types. They are defined and explained here
Table 2.1 A comparison of mortality amongst cigarette to enable the reader to understand the concepts
smokers and non-smokers. involved and to provide a framework which can be
Death rate used to identify the most appropriate study design
Number in Died within over 7 years to answer particular problems. They are discussed
study 7 years (per 1000) in greater detail, with examples, in ensuing
Cigarette 25 769 133 5.16
Non-smokers 5 439 3 0.55

PID2 5/21/04 11:24 AM Page 12

Chapter 2 ‘Cause’ and ‘risk’

studies rely on data collected in a systematic man-

Descriptive studies
ner according to well-defined procedures.
These are used to demonstrate the patterns in • In a cohort investigation individuals are selected
which diseases and associated factors are distrib- for study on the basis that they are or may be ex-
uted in populations. They aim to identify changes posed to the agent under investigation and are
in mortality and morbidity in time or to compare readily identified and ‘followed up’ for a period of
the incidence or prevalence of disease in different time. The follow-up may extend into years and
regions or between groups of individuals with dif- aims to identify the characteristics of those who
ferent characteristics (e.g. occupational groups). develop the disease (or other prior defined end
Correlations are then sought with one or more point) and those who do not.
other factors which may be thought to influence • The subjects investigated in a case–control study
the occurrence of the diseases. Studies of this type are generally recruited because they already have
may give rise to hypotheses of cause but cannot be the disease (or end point) being investigated. Their
used in isolation to explore the meaning of associ- past histories of exposure to suspected causal
ations and can rarely prove cause. This requires the agents are compared with those of ‘control’ sub-
use of the other types of study. jects—individuals who are not affected with the
disease but are drawn from the same general popu-
lation. The analysis involves discriminating
Cohort and case–control studies
between the past exposures and other relevant
These studies are observational. They are planned characteristics of the cases and those of the
investigations designed to test specific hypotheses. controls.
They aim to define the causes or determinants of The differences between these two study designs
diseases more precisely than is possible using are schematically represented in Fig. 2.1. The
descriptive studies alone. They do not involve the cohort study design is closest to the ‘ideal’ experi-
investigator in determining the exposures of mental design. Such studies tend to take longer
individuals. From the results, it is often possible to and to be more expensive than case–control stud-
suggest ways whereby the disease may be preven- ies. However, they usually yield more robust find-
ted or controlled. Both cohort and case–control ings. Case–control studies, though usually cheaper

Past history Composition of Future

risk factors study population disease

Cohort Exposed/at risk Disease

No disease
or random

Not exposed/at risk No disease

Case–control Present
Cases of disease
Matched controls
Figure 2.1 Comparison of cohort
and case–control study designs.

PID2 5/21/04 11:24 AM Page 13

‘Cause’ and ‘risk’ Chapter 2

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.

PID3 5/21/04 11:25 AM Page 14

Chapter 3
Descriptive studies

Introduction Use of descriptive studies

Often an important starting point for many epi-
demiological investigations is the description of
the distribution of disease in populations (descrip- The results of descriptive studies usually only give
tive studies). The principal advantages of descrip- general guidance as to possible causes or determi-
tive studies are that they are cheap and quick to nants of disease, for example where broad geo-
complete and they give a useful initial overview of graphical differences in prevalence are shown.
a problem that may point to the next step in its Sometimes they may be quite precise, for example
investigation. where a particular disease is very much more fre-
Usually, descriptive studies make use of routine- quent within an occupational group or only occurs
ly collected health data, for example death certifi- in a particular exposure group (e.g. asbestosis).
cation data, hospital admission statistics, collated Analysis of the data may indicate that certain
data from computerized general practices or infec- attributes or exposures are more commonly found
tious disease notifications. The main sources of amongst people who have the disease than in
routine health data are set out in Chapter 8. Some those who do not. The converse may also be
social and other variables in relation to which dis- demonstrated, namely that certain attributes are
ease data may be examined are also available from more commonly found amongst people who do
a wide variety of routine sources. The actual source not have the disease than in those who do. This
used for a particular investigation depends on the may be an equally valuable finding. It is rarely pos-
data that are required. With the exception of cen- sible to prove that an agent causes a disease from a
sus material, routine sources of social data are not descriptive study, but investigations of this type
discussed in detail in this book. will often generate or support hypotheses of aetiol-
Often the data required to describe disease distri- ogy and justify further investigations.
bution in a population and related variables are
not readily available or are unsatisfactory for epi-
demiological purposes. In these circumstances it is
necessary to collect the raw material for a descrip- Clinical impressions of the frequency of different
tive study by special surveys. These surveys are usu- conditions and their natural history are often mis-
ally cross-sectional in type (see Chapter 4). leading. The clinical impression is influenced by
the special interests of individual doctors, by

PID3 5/21/04 11:25 AM Page 15

Descriptive studies Chapter 3

events that make a particular impression and by

Variation of disease with time
the chance clustering of cases. To obtain a balanced
view of the relative importance of different condi- • Long-term (secular) trends
tions, their natural history and the factors that • Periodic changes (including seasonality)
• Epidemics
affect outcome requires data from a total popula-
tion or an unbiased (random) sample. Knowledge
of the relative frequency of different diseases is
helpful to the clinician when deciding on the most
likely diagnosis in individual patients. The proba- Long-term (secular) trends
bilities of different diseases vary at different times These are changes in the incidence of disease over
and in different situations. a number of years that do not conform to an iden-
tifiable cyclical pattern. For example, the secular
Service planning trend in mortality from tuberculosis in England
and Wales has showed a steady fall over many years
Health service planning in the past has been large- (Fig. 3.1) but recently the annual number of cases
ly based on historical levels of provision and re- has started to rise. The observation of this trend on
sponses to demands for medical care. In order to its own does not give any indication of its cause.
plan services to meet needs rather than demand, However, it is sufficiently striking to justify specific
and to allocate resources appropriately, accurate studies aimed at trying to identify the reasons for
descriptive data are required on the relative impor- the change. The inclusion in Fig. 3.1 of the times at
tance and magnitude of different health problems which various discoveries were made or specific
in various segments of the community. They are measures were introduced gives some enlighten-
also essential in order to evaluate the effectiveness ment. The overall trend seems to have been hardly
of services and to monitor changes in disease affected by the identification of the causal organ-
incidence which may indicate a need for control ism, or by the introduction of chemotherapy and
action or the reallocation of resources and bacille Calmette–Guérin (BCG) vaccination. This
adjustments to service provision. suggests that these played little part in the decline
in mortality. However, the presentation of these
Analysis of descriptive data data on an arithmetic scale (as in Fig. 3.1) disguises
an important feature of the trend, i.e. a change in
Data derived from routine mortality and morbid- the rate at which the decline occurred. When the
ity statistics (and from cross-sectional surveys) are
usually analysed within three main categories of
• time (when?)
1400 Organism discovered
• place (where?)
• personal characteristics (who?). 1200
SMR (base years 1950–52)


Time Chemotherapy and
600 BCG vaccination

Three broad patterns of variation of disease inci- 400

dence with time are recognizable. These are shown 200
1855 1875 1895 1915 1935 1955 1965

Figure 3.1 Tuberculosis mortality in England and Wales,

1855–1965 (arithmetic scale).

PID3 5/21/04 11:25 AM Page 16

Chapter 3 Descriptive studies

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)

been increasing relentlessly amongst males but the

Mass radiography rate of increase has now declined. By contrast, the
BCG vaccine
increase in mortality rates amongst women con-
tinues. The powerful correlation between mortali-
100 ty and changes in the national consumption of
cigarettes gave rise to the hypothesis that cigarette
smoking could be the causal agent, although it did
not prove causality. The hypothesis has since been
explored through large numbers of epidemiologi-
cal studies.

Periodic changes
1871 1891 1911 1931 1951 1971
These are more or less regular or cyclic changes in
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

1200 Male (deaths) 140

Cigarette consumption ¥ 109

Deaths per million

800 Cigarett
e consum
ption 80
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).

PID3 5/21/04 11:25 AM Page 17

Descriptive studies Chapter 3

800 000

600 000

400 000

Figure 3.4 Notifications of measles in

England and Wales showing periodic 200 000
variation (prior to introduction of
measles vaccination). (Reproduced
with permission of the Office of Popu- ‘42 ‘44 ‘46 ‘48 ‘50 ‘52 ‘54 ‘56 ‘58 ‘60 ‘62 ‘64 ‘66 ‘68
lation Censuses and Surveys (Crown Years (1942–68)

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
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)

PID3 5/21/04 11:25 AM Page 18

Chapter 3 Descriptive studies

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

Jan–Mar Apr–June July–Sept Oct–Dec

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
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


Number of deaths






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,

PID3 5/21/04 11:25 AM Page 19

Descriptive studies Chapter 3


700 Total sales

Sales and prescriptions (thousands)



400 Prescriptions


200 Direct sales





Figure 3.7 Sales and prescriptions of

asthma preparations compared with
deaths from asthma among people
aged 5–34 years, in England and ‘59 ‘60 ‘61 ‘62 ‘63 ‘64 ‘65 ‘66 ‘67 ‘68
Wales, 1959–68. (After Inman WHW, Years (1959–68)
Adelstein AM. Lancet 1969; ii: 279.)

with similar demographic and social charac- Broad geographical differences

teristics). However, if the strict definition of
Variations in the incidence of disease are some-
epidemic is used, it is inappropriate to use the
times related to factors such as climate, social and
term to describe secular trends in diseases such
cultural habits (including diet), and the presence of
as diabetes or malignant melanoma, since there is
vectors or of other naturally occurring hazards.
no evidence that they are temporary increases in
Although the incidence of disease does not respect
administrative boundaries between countries or
regions, these boundaries often follow broadly nat-
Place ural ecological boundaries and tend to encompass
common social and cultural groups. Much valu-
Variations in the incidence or prevalence of
able information pointing to possible causes of
disease with place can be considered under three
disease has been obtained by comparisons of rou-
tinely collected data between countries and other
administrative units. For example, various forms
of cancer and other conditions show striking
geographical difference in incidence (Table 3.2).
Variation of disease with place

• Broad geographical differences Local differences

• Local differences
The distribution of a disease may be limited by the
• Variations within single institutions
localization of its cause. Thus, if a main water sup-

PID3 5/21/04 11:25 AM Page 20

Chapter 3 Descriptive studies

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.)

Rates per 100 000

Disease England and Wales Japan

England and Wales high, Japan low

Cancer of breast (females) 47.9 6.6
Cancer of prostate 20.2 2.9
Cancer of colon 20.9 6.3

England and Wales low, Japan high

Cancer of stomach 23.0 43.8
Cirrhosis of liver (males) 5.0 21.1
Suicide 8.5 18.0

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

PID3 5/21/04 11:25 AM Page 21

Descriptive studies Chapter 3

Table 3.3 Differences in mortality amongst males between urban and rural districts in England and Wales 1969–73

Urban with populations

Conurbations Over 100 000 50 000–100 000 Under 50 000 Rural

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

PID3 5/21/04 11:25 AM Page 22

Chapter 3 Descriptive studies

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

vulnerable to disease and death than are females. Family

This is first apparent in the differential rates of still- Some diseases are especially frequent in certain
birth and early neonatal mortality, and remains families because of a common genetic inheritance,
throughout life (Table 3.4). Indeed, during later which is an intrinsic characteristic of the individu-
life, with the exception of disorders that are specif- als. The risk of disease among members of the same
ic to the female, there are few diseases which have family may also be increased because the members
a greater incidence in women than in men. In most share a common environment and culture. Cul-
societies, men are exposed to a greater number and ture affects a wide range of disease-related factors
variety of hazards than are females often because of such as type of housing, dietary habits and the way
differences in their leisure and work activities. in which food is prepared, as well as the individ-
Even when the two sexes are exposed to the same ual’s reaction to illness.
hazards for the same period of time, there is evi-
dence that women are less likely to develop disease Occupation and socioeconomic group
and that they survive better than men. Some dis- Some people are exposed to special risks in the
eases appear to vary in incidence between the sexes course of their occupation. These include expo-
only because they are more readily diagnosed in sures to dust (particularly coal dust, silica and
one sex than the other, for example gonorrhoea in asbestos), toxic substances and gases used in
men, or because they are more likely to come to industrial processes, and the risks of accident.
medical attention, for example in mothers of Some occupations influence habits such as the
young children. amount of tobacco smoked and of alcohol con-
sumed or the regularity of meals, which in turn
Ethnic group affect disease incidence.
This term tends to be used very loosely to describe When interpreting any observed correlation be-
a number of personal characteristics, including tween occupation and disease it is necessary to take
some that are strictly genetically determined, for account of the factors which determine a person’s
example skin colour, and some that have nothing choice of occupation. Some may affect the person’s
to do with genetics, for example country of birth susceptibility to disease; for example, tall and pow-
and religion. It is often difficult to disentangle erful people may choose physically demanding

PID3 5/21/04 11:25 AM Page 23

Descriptive studies Chapter 3

Table 3.5 SMRs for ages 15–64 years (England and Wales) showing trends by social class for specific causes of death.

Cause of death (ICD number) I II IIIN IIIM IV V

Malignant neoplasm of stomach (151) 50 66 79 118 125 147

Malignant neoplasm of trachea, bronchus and lung (162) 53 68 84 118 123 143
IHD (410–414) 88 91 114 107 108 111
Cerebrovascular disease (430–438) 80 86 98 106 111 136
Bronchitis, emphysema and asthma (490–493) 36 51 82 113 128 188

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.

PID4 5/21/04 11:27 AM Page 24

Chapter 4
Surveys, survey methods and bias

Introduction managed or reported and therefore do not come to

the attention of the medical profession.
Many descriptive studies make use of routinely col-
lected data. However, such data are often unsatis-
Example 1 Osteoarthritis is neither fatal nor is it al-
factory for this purpose and specifically designed
ways treated or reported. Studies of osteoarthritis
surveys are needed. The problems are shown
based entirely on the cases treated in hospital or
brought to the attention of the general practiti-
oner may be misleading.
Problems with routinely
collected data Example 2 Acne varies considerably in its severity.
In some individuals it is manifest by a few spots on
the face or back whereas in the most severe cases
there are widespread areas of pustules. Mildly
affected individuals may ignore the manifestations,
• Difficulties in ascertainment of cases or use cosmetics or preparations available from
• Variations in diagnostic criteria pharmacies without prescription. The cases seen
• Absence of records of the attributes of individuals by the general practitioner will tend to be the more
• Unsuitable format of records
severe. However, some people will be more con-
• Inconsistency in data presentation
cerned than others and thus some mild cases may
be seen by the GP. Specialist dermatologists will see
the most severe and those refractory to treatment.

Difficulties in ascertainment of cases

Variations in diagnostic criteria
The recorded number of patients with a condition
may vary for reasons that have nothing to do with These tend to vary between doctors and may
the actual frequency of the disease. For example, change with time. This may be simply a matter of
the tendency to seek medical attention and the fashion or because the facilities for accurate diag-
availability of services may vary. This source of bias nosis vary. Sometimes, there may be internation-
is of greatest importance when studying illnesses ally agreed changes in classification practices.
that are rarely fatal and therefore do not appear on
death certificates, or that are not always medically Example The ICD is revised about every 10 years

PID4 5/21/04 11:27 AM Page 25

Surveys, survey methods and bias Chapter 4

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.

Absence of records of attributes Cross-sectional (prevalence)

of individuals surveys
The attributes of the individuals which the study A cross-sectional (prevalence) survey is simply a de-
proposes to investigate in relation to the presence scriptive study which, instead of relying on routine
of disease may not be recorded systematically. sources of data, uses data collected in a planned
way from a defined population. The aim is to de-
Example The occupation of patients is often not scribe individuals in the population at a particular
recorded or not recorded in sufficient detail in hos- point in time in terms of their personal attributes
pital notes to allow investigation of a cancer which and their history of exposure to suspected causal
it is suspected may result from occupational expo- agents. These data are then examined in relation to
sure to a carcinogenic agent. the presence or absence of the disease under inves-
tigation or its severity with a view to developing or
testing hypotheses as described in Chapter 3.
Unsuitable format of records
The data are recorded but are not usable because Example A cross-sectional survey was carried out
the form of the record is unsuitable, or because among a multiracial workforce at worksites in New
they are governed by strict rules of confidentiality. Zealand by Scragg and colleagues between 1988
and 1990. The survey studied 5677 staff aged 40–64
Example Diagnoses may be recorded but not in a years. The subjects were asked about their age,
form or in sufficient detail to allow classification by ethnicity, past medical history, occupation and
ICD or other standard criteria. income. Their height, weight and blood pressure
were recorded and an oral glucose tolerance test to
detect diabetes mellitus was performed. The study
Inconsistency in data presentation
showed that the prevalence of diabetes increased
In the analysis of deaths, the numbers and the date with age, was more common in Maoris and that
of occurrence are indisputable in countries where approximately 50% of workers with diabetes were
death registration is standard practice. However, previously undiagnosed (see Fig. 4.1). The preva-
when analysing morbidity by time, there are sev- lence of diabetes was also significantly correlated
eral possible points of reference. Those commonly with weight and low income.
used are the date of onset of the disease, the date of
onset of symptoms, the date of first diagnosis or
the date of hospital admission. In acute diseases,

PID4 5/21/04 11:27 AM Page 26

Chapter 4 Surveys, survey methods and bias


Per cent with diabetes

40–44 45–49 50–54 55+
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.)

Survey methods lecting a sample. Some commonly used sampling

techniques are detailed below.
A number of practical and theoretical problems
can arise in the design and conduct of cross-
sectional surveys and other studies which may in- Types of sample
validate the results unless they are handled pro-
perly. The investigator needs to be aware of these • Simple random sample
• Systematic sample
potential problems and familiar with the methods
• Stratified sample
that are available to solve them or to minimize • Cluster sample
their effects. • Multistage sample

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.

PID4 5/21/04 11:27 AM Page 27

Surveys, survey methods and bias Chapter 4

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
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,

PID4 5/21/04 11:27 AM Page 28

Chapter 4 Surveys, survey methods and bias

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.)

Error and bias in numerator data Observer variation

The quality of numerator data is crucial for accu- The principal types of observer variation are as fol-
rate classification of individuals according to their lows.
personal attributes, their exposure to suspected • Failure of the same observer to record the same
causal agents and whether or not they have the dis- result on repeated examination of the same mate-
ease under investigation. In contrast to descriptive rial (inconsistency) — this is called intraobserver
studies based on routine data, special surveys offer variation.
the investigator the advantage of being able to • Failure of different observers to record the same
specify the observations that he or she wishes to be result — this is called interobserver variation. The
made, rather than being constrained by the data greater the number of different observers, the
that are collected for other purposes. Furthermore, greater are the chances of variation between them.
the investigator can prescribe the methods to be Either of the above types of error can arise for
used in examining or questioning the individuals several reasons.
involved in the study. However, the investigator • Bias induced by awareness of the hypothesis
usually only has a single opportunity to make the under investigation; for example, in a study of HIV
observations on each subject. It is therefore essen- infection, the observer may probe answers to cer-
tial that the information required is clearly defined tain questions more deeply if the subject has
at the outset and that efforts are made to ensure declared himself to be a homosexual or an
that consistent results are obtained by the instru- intravenous drug user.

PID4 5/21/04 11:27 AM Page 29

Surveys, survey methods and bias Chapter 4

• 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-

PID4 5/21/04 11:27 AM Page 30

Chapter 4 Surveys, survey methods and bias

mized by making every effort to encourage cooper-

Systematic error (bias)
ation of the potential subjects and avoiding any in-
convenience or discomfort to them. All available This is a consistent difference between the record-
means should be used to trace and persuade non- ed value and the ‘true’ value in a series of observa-
attenders to take part or continue to participate in tions. For example, if the height of an individual is
the investigation. The similarity of those who par- always measured when the person is wearing
ticipate and those who are lost from the study shoes, then the measurement will be consistent
should be checked by comparing their general at- but will have a systematic bias.
tributes such as age, marital status, sex and occupa-
tion to establish how representative they are of the
total sample.
There are several ways of handling people lost to This relates to whether a test is able either to sepa-
follow-up in the analysis phase of an investigation. rate people with a disease (or a particular attribute)
• Exclude them from both the numerator and the from those without the disease (or attribute) or to
denominator. place subjects accurately on a range of severity
• Include them up to the time that they left. This (or a scale measuring an attribute). The degree to
involves calculating units of ‘time at risk’ (see which this is achieved correctly is a measure of
p. 33). discrimination. A test with good discriminatory
• Include all those ‘lost’ for half the ‘time at risk’ power has a small range of error in relation to
on the assumption that the rate of loss was even the potential range of true results. There are two
throughout the period and on average each indi- basic characteristics of a test which measure its
vidual was present for half the time. discriminatory powers: its reproducibility and its
• Analyse the data on the assumption that all validity.
those lost developed the disease, or had the most
adverse outcome, and then on the assumption that
Reproducibility (reliability, repeatability)
none of them developed it. This will show the
range within which the true result might lie. This is a measure of the consistency with which a
question or a test will produce the same result on
the same subject under similar conditions on suc-
Assessment of error in surveys
cessive occasions. A highly reproducible test must
Some terms that are frequently used in the assess- have low random error, although it may still have
ment of error in surveys are given below. systematic error.
When reproducibility is evaluated by retesting
Common terms in survey error subjects, it is usually defined as the ratio of the
number of cases positive on both occasions to the
• Random error
• Systematic error number positive on at least one occasion. It can be
• Discrimination assessed by the following procedures.
• Reproducibility, assessed by: • Replication of tests. The results of a series of meas-
replication of tests urements by the same observer or by different ob-
comparison of test systems servers using the same test on the same group of
use of check questions
subjects (or set of specimens) under identical con-
random allocation of subject to interviewer
ditions are compared.
• Validity
• Comparison of test systems. The measurements are
repeated using a different instrument or test sys-
Random error
tem. Statistical analyses can be used to identify
This is due to the chance fluctuation of recorded whether the variation is attributable to the test sys-
values around the ‘true’ value of an observation. tem, intraobserver variation, interobserver varia-

PID4 5/21/04 11:27 AM Page 31

Surveys, survey methods and bias Chapter 4

tion or subject variation. Similar methods can be Validity (accuracy)

applied to the assessment of the reproducibility of
This is a measure of the capacity of a test to give the
questions, but there are problems because when
true result. A valid test is one that correctly detects
the same question is repeated, the subject (and ob-
the presence or absence of a condition or places a
server) may be conditioned by replies given on pre-
subject correctly on a scale of measurement. For ex-
vious occasions.
ample, glycosuria as a test of the presence of dia-
Other procedures for assessing the reproducibili-
betes has poor validity compared with a glucose
ty of questionnaires are:
tolerance test.
• the use of check questions, i.e. questions which
Validity has two components. In the case of a
seek the same information though in a different
test which divides a population into two groups,
form, for example age and date of birth; and
validity is assessed by how well it picks up those
• the random allocation of subjects to different in-
with diseases (its sensitivity) and how well it rejects
terviewers and comparison of results between
those without disease (its specificity) (see p. 137).

PID5 5/21/04 11:28 AM Page 32

Chapter 5
Cohort studies

Introduction constantly maintained (e.g. doctors, nurses, etc.);

individuals who are members of a particular insur-
Cohort studies involve the investigation of groups
ance scheme; or employees in industries with a low
of people who have no manifestations of the
turnover of workers.
disease under study at the time they are recruited.
At the time of recruitment to the study, the in-
The selected study group is observed over a period
vestigator identifies the characteristics of the sub-
of time in order to measure the frequency of
jects by the use of standard questionnaires or the
occurrence of the disease amongst people exposed
measurement of any number of biological vari-
to the suspected causal agent compared with its
ables. They are then followed up until a sufficient
frequency amongst individuals who are not ex-
proportion have reached a predefined end point
posed. Cohort studies can be used in a similar way
(usually the development of the disease being in-
to identify the determinants of disease in the study
vestigated or death). During the follow-up period,
their exposures to suspected harmful agents are
recorded. Such cohorts can be used to estimate
prevalence, incidence and risk in relation to a sus-
Types of cohort pected causal agent without recruiting an addi-
• Groups with special personal characteristics tional comparison (control) group because the
• Groups with special exposures comparison group (those not exposed to the agent
being investigated) is a subgroup of the cohort it-
self. This is called an internal control group.

Groups with special personal

characteristics Groups with special exposures
Groups of individuals who have special character- The other main type of cohort comprises groups of
istics unrelated either to their risk of exposure or to individuals who have all been exposed to the agent
their risk of disease, and which make them easy to or the experience that is being investigated. This
follow up, provide useful cohorts for the investiga- type of cohort requires the concurrent recruitment
tion of many diseases. Cohorts that have been used and study of an external control group. The con-
because they are easy to study over long periods of trol group in this instance must be drawn from a
time include, for example, those selected because population that is similar to the exposed group in
they belong to a profession of which a register is all respects other than their exposure to the agent

PID5 5/21/04 11:28 AM Page 33

Cohort studies Chapter 5

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
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-

PID5 5/21/04 11:28 AM Page 34

Chapter 5 Cohort studies

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

PID5 5/21/04 11:28 AM Page 35

Cohort studies Chapter 5

Percentage of excess deaths in cigarette smokers


100 1951–71

Figure 5.1 Age-specific excess mor-

tality in cigarette smokers in first half
of study (lower line) contrasted with
that in second half (upper line). An ex-
cess of 100% represents doubled 35–44 45–54 55–64 65–74 75–84 85–94
death rate. Bars indicate standard de- Age
viations. (From Doll et al. 1994.)



Per cent alive

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,
based on specific death rates for the 40 55 70 85 100
entire 40-year period. (From Doll et al. Age

PID5 5/21/04 11:28 AM Page 36

Chapter 5 Cohort studies

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.

Incidence rate per 1 000 000

person-years at risk
Survivors of the Hiroshima and Nagasaki
atomic explosions (Brill AB, Masanobu RR, Distance from
Heyssel RM. Ann Intern Med 1962; epicentre (m) Hiroshima Nagasaki
56: 590–609) 0000–999 1366 563
1000–1499 308 530
The second type of cohort, one which is defined by
1500–1999 42 68
the fact that the individual members have all been
2000–9999 28 37
exposed to the same experience or agent, has the
closest similarity to the laboratory experiment.
There are many instances of cohorts that have Table 5.2 Relative risk of breast cancer in women
been defined in this way. For example, the sur- subjected to regular fluoroscopies at different ages (risk in
vivors of the atomic bomb explosions in Hiroshi- general population = 1).
ma and Nagasaki comprise a unique group of Age at first exposure (years) Relative risk
people who were exposed to high levels of ionizing
<15 2.1
radiation for a short time. In this group of people
15–19 3.8
there was little difficulty in calculating the propor-
20–24 1.7
tion who developed leukaemia after, say 10 years, 25–29 1.6
i.e. the absolute risk of leukaemia. 30–34 1.2
However, in order to establish whether the inci- 35–39 0.8
dence of leukaemia in the cohort was more or less 40+ 0.9
than in a group not so exposed — the relative risk of
leukaemia — it was necessary to study a group of
1940s and 1950s while being treated for pul-
people who were similar to the exposed group in
monary tuberculosis before the dangers of X-rays
all respects except for their exposure to ionizing ra-
were fully appreciated. It has been shown that
diation. In one of the many studies of the survivors
young women in the group that were irradiated
from Hiroshima and Nagasaki, the control group
had a higher than expected incidence of breast
comprised individuals who were living in the same
cancer. In this study, the control group was all
area but outside the radiation zone. The study
other women of the same age in the population,
showed that the incidence of confirmed leukaemia
the great majority of whom, it may be assumed,
was between 50 and 100 times greater in the ex-
were not exposed to radiation in this way (Table
posed population than in the controls. Further in-
vestigations showed a clear relationship between
the distance from the epicentre of the explosion
and leukaemia incidence rates (Table 5.1), demon-
Social class differences in IHD in men
strating a dose–response effect.
(Pocock SJ, Shaper AG, Cook DG, Phillips AN,
Walker M. Lancet 1987; ii: 197–201)
Regular fluoroscopy and risk of breast
During 1979–80, 7735 men aged 40–59 years
cancer (Boice JD, Monson RR. J Natl Cancer Inst
were randomly selected from the ‘lists’ of people
1977; 59: 823–32)
registered with general practitioners in 24 towns
A cohort that experienced a different type of ioniz- in England, Scotland and Wales and were asked
ing radiation is exemplified by the group of people to participate in a long-term study. Seventy-eight
who had large numbers of fluoroscopies in the per cent of those who were approached agreed to

PID5 5/21/04 11:28 AM Page 37

Cohort studies Chapter 5

Table 5.3 Attack rates of major IHD

Cases per 1000 per annum
events during follow-up, before and Number of
after adjustment for social class Social class category IHD cases Adjusted Unadjusted
differences in risk factors.
I 21 5.6 7.4
III non-manual
} 5.5 6.0
} 6.0

III manual 169 8.2 7.7

} 7.9 7.3
} 7.5

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.

PID6 5/21/04 11:29 AM Page 38

Chapter 6
Case–control studies

tion is inevitable there are some diseases that result

in sudden death before the patient can be treated
The basic principle of a case–control study is to (e.g. acute myocardial infarction). In these circum-
compare the personal characteristics and expo- stances it is necessary either to accept that the find-
sures of individuals with and without the disease ings can only be applied to those who survive long
(or outcome) of interest. The individuals with the enough to be admitted or to devise a method of in-
disease are referred to as cases; those without are corporating the fatalities into the study.
controls. Both the validity and generalizability of a For many diseases hospital admission is not
case–control study are affected by the way in inevitable. It used to be standard practice to admit
which cases and controls are identified and everyone suspected of having a deep venous
recruited. thrombosis (DVT). During the admission they
would be investigated and those in whom the di-
agnosis was confirmed would be started on anti-
Identification and selection of cases
coagulant therapy before discharge. Nowadays it is
The value of a case–control study is profoundly in- not unusual to investigate the individual as an out-
fluenced by the ways in which both the cases and patient, initiate therapy on an outpatient basis and
the controls are selected. Ideally, all the cases of the then hand the long-term management over to the
disease in a defined population should be included GP. It follows that, although in the past it would
in the investigation. However, it is rarely feasible have been legitimate to recruit cases of DVT from
or indeed necessary to do this in order to reach hospital inpatients, it is no longer the case. The in-
sensible and valid conclusions. Most studies are vestigator thus has to identify an unbiased source
implicitly or explicitly concerned with a sample of of cases, that is, a clinic or process through which
cases. The source of potential cases is determined all or most genuine cases will pass. In the case of
to a great extent by the disease that is being DVT this could be anticoagulant clinics or general
investigated. practice records of patients with a diagnosis and
When admission to hospital for treatment of the anticoagulant treatment.
disease is inevitable (e.g. breast cancer, fractured In some diseases the majority of cases are treated
femur, intestinal obstruction, end-stage renal fail- by the GP with or without referral of a few to a spe-
ure) then cases can be recruited from hospital in- cialist. Lower urinary tract symptoms indicative of
patients. However, it is possible that although benign prostatic hyperplasia is an example of this
hospital admission for the treatment of a condi- situation. GPs manage many men with alpha-

PID6 5/21/04 11:29 AM Page 39

Case–control studies Chapter 6

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

PID6 5/21/04 11:29 AM Page 40

Chapter 6 Case–control studies

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:

Collection of data from cases A B A ¥ ( B + D)

RR = divided by =
and controls A+C B + D A ¥ ( A + C)

Data can be obtained by interview, questionnaire A B

AR = minus
or reference to pre-existing records or a combina- A+C B+D
tion of one or more of these sources. Whatever the
The subjects in a case–control study are identi-
method of obtaining information on the personal
fied either because they have the disease (the cases)
characteristics and exposures of the cases and con-
or because they do not have the disease (the con-
trols the same method must be used for both.
trols) that is being investigated. They are subse-
Wherever possible the field workers should be un-
quently subdivided into ‘exposed’ and ‘not
aware of the status (case or control) of the subject
exposed’ subgroups (Table 6.2). It is not possible to
about whom they are collecting the data.

Table 6.2 Division of subjects in a case–control study.

Risk in cohort and case–control
Diseased Not diseased
Suspected cause present a c
Cohort studies are designed to provide the data
Suspected cause absent b d
needed to calculate incidence rates of the disease
Total a+b c+d
amongst individuals exposed to the suspected

Table 6.1 Information available in cohort studies.

Disease present Disease absent Total

Exposed to suspected cause A C A+C

Not exposed to suspected cause B D B+D
Total A+B C+D A+B+C+D

PID6 5/21/04 11:29 AM Page 41

Case–control studies Chapter 6

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

PID6 5/21/04 11:29 AM Page 42

Chapter 6 Case–control studies

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
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
• 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
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-

PID6 5/21/04 11:29 AM Page 43

Case–control studies Chapter 6

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

Table 6.3 Risk factors for cervical cancer.

Numbers Odds ratio

Risk factor Cases Controls Crude Adjusted

Number of sexual partners of woman

<2 25 210 1.00 1.00
2–3 54 73 6.21 3.43
4–5 47 53 7.44 3.59
6–10 57 39 12.27 5.51
> 10 69 28 20.70 8.99

Number of sexual partners of mate

1 24 198 1.00 1.00
2–3 42 78 4.44 2.72
4–5 38 39 8.03 4.99
6–10 45 22 16.87 7.98
> 10 49 23 17.57 8.9

Trichomonas infections 53 21 4.61 2.10

Herpes type 2 (neutralization index > 1000) 12 6 6.57 2.70

PID6 5/21/04 11:29 AM Page 44

Chapter 6 Case–control studies

Table 6.4 Pertussis vaccine and encephalopathy in children.

All outcomes Dead or dysfunctional 10 years later

n Vaccine associated (%) n Vaccine associated (%)

Cases 904 30 (3.3) 367 12 (3.3)

Controls 1783 23 (1.3) 723 6 (0.8)
Odds ratio 3.3 5.5

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.

PID7 5/21/04 11:30 AM Page 45

Chapter 7
Intervention studies

location of individuals in a population to ‘test’ and

‘control’ groups. The intervention under trial is ap-
Descriptive, cohort and case–control studies are plied to the test group but not to the control group
used to develop and test hypotheses about the pos- and the effect is assessed in terms of the same de-
sible causes and determinants of disease. The re- fined outcome in both groups. The selection of the
sults may suggest methods of prevention or study population should be governed by the fol-
treatment which then need to be formally evalu- lowing considerations.
ated. Intervention studies are most often used for • The population under study should be represen-
this purpose and are commonly called clinical tri- tative of the population in which it is intended to
als. They are essentially experimental studies in apply the intervention being tested (this is called
which the efficacy and safety of disease manage- the target population).
ment are evaluated by comparing the outcome of • It is important to choose a stable population in
the intervention in test and control groups. The in- which there are unlikely to be heavy losses during
tervention under test is most often a new preven- the follow-up period and whose cooperation is rea-
tive or treatment regime, but the method can also sonably assured. Volunteers are usually not accept-
be used to compare different established regimes able since they tend to differ from non-volunteers
and to evaluate the effectiveness and efficiency of in important respects, such as motivation and past
different forms of service provision. Experimental history of illness.
studies in which the incidence of a disease in those • The likely frequency of the outcome being meas-
deliberately exposed to a suspected causal agent, or ured should be known, since this critically affects
protected from it, is compared with that in con- the required sample size.
trols can also be of value and provide the most • The population should be readily accessible. Tri-
convincing evidence of a causal relationship. als are often most conveniently conducted in pa-
However, for practical and ethical reasons this ap- tients attending general practice or hospital, or in
proach is rarely adopted except in animal studies. residential institutions, factories, the armed forces,
etc., even though they may not fairly represent the
general population in all respects.
Methods in clinical trials
The methodologies of preventive and therapeutic
trials have some similarities to those used in cohort
studies. The basis of a clinical trial is the random al-

PID7 5/21/04 11:30 AM Page 46

Chapter 7 Intervention studies

Allocation to test and Stratified allocation

control groups
Where the population is relatively small and non-
In principle, the allocation of patients to test and homogeneous, random selection within specified
control groups should be random: hence the term subgroups, for example age groups, may be desir-
sometimes applied to clinical trials, randomized able to ensure similarity in relevant characteristics
controlled trials (RCTs). The aim is to ensure that between test and control groups.
those treated and those untreated are similar in all
respects prior to intervention. This is necessary to
No allocation
guard against the possibility that some factor other
than the intervention could account for differ- Sometimes random allocation of treatment would
ences in outcome in the two groups. not be ethical, for example a trial of a new type of
measles vaccine in children. In this case, the com-
parison must be with past experience or that in
other populations. It is difficult in such cases to
Types of random allocation measure the extent of any benefit with confidence.

• 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
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
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

PID7 5/21/04 11:30 AM Page 47

Intervention studies Chapter 7

give a spuriously reduced apparent benefit from

Sequential analysis
the treatment. The safety of an intervention is as
important as its efficacy and the assessment of Sometimes, when a result is required urgently or
outcome should always include the frequency of when the anticipated benefits are high or the pos-
adverse effects of the intervention as well as its sible adverse effects are serious, the results are
benefits. analysed sequentially. This technique involves
continuous data analysis and allows the trial to be
stopped immediately when a significantly benefi-
cial or adverse effect has been demonstrated. Trials
Procedures for the follow-up of subjects in both can also be stopped when the results fail to reach a
test and control groups should be the same, giving previously defined level of significance, usually
particular attention to the following. that which the investigators regard as the mini-
• The data collected must be obtained and mum useful benefit.
recorded in a standard manner.
• The method used should be simple and should
Ethical considerations
be sufficiently sensitive to detect reliably all rele-
vant events in members of the study population. The ethical questions that arise in the planning
• Follow-up must be equally rigorous in both test and conduct of RCTs are shown below.
and control groups.
• Follow-up must start from the time of allocation
and continue for long enough to evaluate fully the
outcome in all subjects. • What are the possible risks of treatment and of failure
• Cooperation must be maintained at the highest to treat?
• Is it right to expose some people to possible harm from
possible level, and losses from the study popula-
an untested treatment or to withhold from others a pos-
tion for any reasons should be minimized.
sibly beneficial treatment?
• Is it right to introduce a new treatment into use with-
out first assessing its safety and benefits by a properly
conducted trial?
It is usual to use an intention to treat (ITT) analysis • To what extent should a trial be explained to the sub-
with randomized controlled trials. This method re- jects before they agree to participate?
• How can the welfare and safety of participants be
quires that study subjects are analysed according to
safeguarded while preserving the principle of ‘blind’
the treatment allocated irrespective of whether it
was taken. The reason for this is that the treatment
allocated is random and changes in treatment (in-
cluding non-compliance) are not. Therefore, there Before a new drug can be licensed there is a
is a potential for confounding if comparisons of requirement to perform a series of RCTs and
the actual rather than assigned treatment are un- demonstrate its efficacy (or relative efficacy) and
dertaken. For example, consider two treatments its safety. The trials must be carried out in accor-
where one is superior to the other: in particular, it dance with protocols set out by the licensing
acts faster. If the analysis is carried out on patients authority. The general method of the RCT is
who comply with the treatment plan then the increasingly being applied to interventions other
superior treatment may appear to be worse simply than medicines. All studies involving humans
because the successes early in the programme are must be approved by a recognized ethical
not included in the analysis. committee.

PID7 5/21/04 11:30 AM Page 48

Chapter 7 Intervention studies

group to which the patients were allocated, but the

Examples of intervention studies patients did not. This was to enable the doctor to
adjust drug dosage in those on active treatment if
MRC trial of treatment of mild necessary to achieve the target level of blood pres-
hypertension (Medical Research sure. All other management was the same in both
Council Working Party. Br Med J 1985; 291: treatment and placebo groups. Recruiting took
97–104) place over 9 years and the data were analysed se-
quentially every 6 months in order to test whether
It has long been known that people with high
any significant differences were emerging. In the
blood pressure have an increased risk of stroke and
end, 85 572 person-years of observation accrued.
other cardiovascular events and that treatment is
There was a very significant reduction in the inci-
effective in reducing the incidence of these condi-
dence of stroke in the treated group, but no differ-
tions in severe hypertension. However, the value of
ence in the rates of coronary events (Table 7.1). The
treating mild hypertension compared with disad-
overall incidence of cardiovascular events was re-
vantages of long-term therapy in otherwise
duced, but mortality from these and all causes was
healthy people was less certain. An RCT of treat-
not. It was concluded that if 850 mildly hyperten-
ment in such cases was therefore carried out by the
sive patients are given treatment for a year, about
MRC. Even though hypertension and cardiovascu-
one stroke will be prevented. On the other hand,
lar complications are relatively common condi-
this would subject a substantial percentage of pa-
tions, it was calculated that this would require a
tients to chronic side-effects, most but not all of
very large-scale trial in order to obtain a statistical-
which would be minor.
ly significant result. Subjects for the trial were
found by screening blood pressure measurements
in 515 000 people aged 35–64 years selected from
Prevention of rickets in Asian children:
the age–sex registers of 176 general practices in
assessment of the Glasgow campaign
England, Scotland and Wales.
(Dunnigan MG, Glekin BM, Henderson JB et al.
In this way 17 354 patients with a diastolic pres-
Br Med J 1985; 291: 239–42)
sure in the range 90–109 mmHg and systolic pres-
sure below 200 mmHg were identified. Patients There have been many reports of vitamin D defi-
were randomly allocated to one of four groups, two ciency leading to rickets in infants and school chil-
of which were treated with different hypotensive dren, and osteomalacia in women among the
drugs and two with placebo tablets which looked British Asian community. Theoretically, this would
identical to the active drug tablets. Randomization be easily remedied. The treatment is clear cut and
was stratified by age and sex. The target level of no trial of the efficacy of vitamin D is needed. The
blood pressure was below 90 mmHg to be reached acceptability of a prophylactic programme and its
within 6 months of entry. The study was single- effectiveness in reducing the prevalence of rickets,
blind only, i.e. the doctor knew the treatment however, needed to be assessed. Random alloca-

Table 7.1 Mild hypertension: main events in treatment and control groups.

Active treatment Placebo

Event difference Number Rate Number Rate difference

Stroke 60 1.4 109 2.6 45

Coronary events 222 5.4 234 5.5 6
All cardiovascular events 286 6.7 352 8.2 19
All cardiovascular deaths 134 3.1 139 3.3 4
Non-cardiovascular deaths 114 2.7 114 2.7 0

PID7 5/25/04 2:21 PM Page 49

Intervention studies Chapter 7

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

Table 7.2 Neural tube defects (NTDs)

Folic acid Other vitamins NTD/all babies RR
and folic acid supplementation.
+ - 2/298
+ + 4/295
} 6/593 (1.0%)

- - 13/300
- + 8/302 } 21/602 (3.5%)

PID7 5/21/04 11:30 AM Page 50

Chapter 7 Intervention studies

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
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.

PID8 5/21/04 11:33 AM Page 51

Chapter 8
Health information and sources
of data

Introduction Census data

Health is an elusive concept. The World Health Or- Most developed countries undertake regular and
ganization (WHO) defines it as ‘a state of complete detailed censuses of their populations in order to
physical, mental and social well-being’. It is, how- provide information to assist in social and fiscal
ever, difficult if not impossible to use this definition planning. Although there is evidence that many of
to measure the health of populations in any cate- the ancient empires, for example Babylon and
gorical sense. Its principal limitation is that an indi- Egypt, undertook occasional, quite sophisticated
vidual’s sense of ‘well-being’ is intimately related to censuses, it was the Romans who introduced it as a
that person’s expectations from life; these are diffi- regular administrative exercise. They did this pri-
cult to measure objectively. Therefore, in order to marily for taxation assessment purposes. Perhaps
measure and compare the health of populations, the most famous Roman census was the one which
there are few alternatives other than to make use of took the parents of Christ to Bethlehem at the time
indices of death and disease, despite the fact that of his birth. After the fall of the Roman Empire, the
these are the antithesis of health. The calculation of regular counting of populations ceased. In Eng-
death rates and disease rates requires both numera- land, the first post-Roman attempt to enumerate
tor data about the events being studied (death and the population resulted in the compilation of the
disease) and denominator data about the popula- Domesday Book in the 11th century. In common
tions in which the events take place. with most of their predecessors, the administrators
This chapter is concerned with routinely colle- at that time were concerned to identify families
cted data that can be used in the measurement of rather than individuals, and even families of differ-
health, mainly from official sources. ent status were recorded differently. From the
material that survives, it is not possible to derive a
Sources of data on the web precise figure of the population at that time.
Office of National http://www.statistics.gov.uk/ The modern system of censuses was introduced
Statistics Census in Europe during the late 18th and early 19th cen-
Health Protection Agency http://www.hpa.org.uk/ turies. In England and Wales, the first complete
Association of Public http://www.pho.org.uk/ census was undertaken at the behest of Parliament
Health Observatories
in 1801. Since then there has been a full census
Cancer Registration http://www.ociu.org.uk/
every 10 years, with the exception of 1941. In re-
Census information www.census.ac.uk cent years, 10% sample censuses have been under-

PID8 5/21/04 11:33 AM Page 52

Chapter 8 Health information and sources of data

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
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

PID8 5/21/04 11:33 AM Page 53

Health information and sources of data Chapter 8

that time were baptized in infancy, the numbers of

Birth registration
baptisms recorded in the parish registers can be
used as proxy indicators of the numbers of live • Child’s name, sex, date and place of birth
births. There was no information on stillbirths. As • Mother’s name, place of birth and usual residence
• Father’s name (if known), place of birth and
most marriages were in church, marriage rates can
also be computed from the parish records. Like-
wise, most of the population received Christian
burials and therefore the fact of death was usually
recorded. However, the presumed cause of death Not all of these data are entered on the birth
was of little interest to the ecclesiastical authorities certificate. Additional confidential information
and was not routinely noted. In the 17th century, is collected for statistical purposes. This includes
‘Bills of Mortality’ were published for some large mother’s date of birth and father’s date of birth
towns and cities. The best known are those com- (if his name appears on the register). For legiti-
piled by John Graunt (Fig. 8.1). The cause of death mate births, the following additional informa-
was arrived at by paying lay ‘searchers’, normally tion is required: date of parents’ marriage, number
women parishioners, to inspect the bodies and of previous marriages of the mother and number
form an opinion. Whereas many of the common of children born in the present marriage, distin-
causes of death left stigmata that were plain for all guishing those born dead from those born
to see, for example plague and smallpox, other dis- alive.
eases gave rise to less definite changes and there If the child is stillborn, a certificate of cause of
was doubtless considerable guesswork on the part stillbirth has to be presented to the registrar. This
of the searchers. certificate is similar to a death certificate and is
When the secular authorities made the registra- issued either by a registered medical practitioner or
tion of vital events mandatory, a government by a state-certified midwife involved with the birth.
department called the Registrar General’s Office Tabulations and analyses of birth data are pub-
was established to supervise the processing and lished annually by the Office of National Statistics
collation of records, and to report to Parliament and are available on the web (www.statistics.gov.uk).
and other government departments. Dr William They are used to study patterns of fertility and
Farr was the first medical statistician at the office. to assist in making population estimates and
His meticulous and imaginative analyses of the projections.
data set the standards for the present sophisticated
system for the registration, analysis and publica-
tion of vital events. Now the task of collating,
analysing and publishing information relating The present regulations governing registration of
to vital events is the responsibility of the Office deaths were set out in the Births and Deaths Regis-
of National Statistics (website: www.statistics. tration Act (1968). The Act requires that: ‘. . . in the
gov.uk). case of the death of any person who has been at-
tended during his last illness by a registered med-
ical practitioner, that practitioner shall sign a
certificate . . . stating to the best of his knowledge
All births must be registered by one of the parents and belief the cause of death and shall forthwith
(or someone on their behalf) with the local Regis- deliver that certificate to the Registrar.’ The certifi-
trar of Births, Marriages and Deaths within 6 weeks cate that the doctor is required to complete and
of the event. Certain of the information required at sign (Fig. 8.2) is one of cause (or causes) of death,
this time is entered in the register and is available not of fact, since the doctor is not obliged to in-
for public scrutiny. The following information is spect the body after death. After giving the de-
available for public scrutiny. ceased’s name, age, date and place of death and

PID8 5/21/04 11:33 AM Page 54

Chapter 8 Health information and sources of data

Figure 8.1 The General Bills of Mortality for London, 1641 and 1665.

PID8 5/21/04 11:33 AM Page 55

Health information and sources of data Chapter 8

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

PID8 5/21/04 11:33 AM Page 56

Chapter 8 Health information and sources of data

Figure 8.2 Death certificate (England and Wales). (Reproduced with permission of the ONS (Crown copyright).)

after death. The third part is completed by another

WHO definitions
medical practitioner who is not professionally as-
sociated with the attending practitioner nor relat- • Neonatal death A live-born infant that dies
ed to him or her or to the deceased. He or she must within 28 days
• Early neonatal death A live-born infant that dies
have been on the Medical Register for at least 5
within 7 days
years. This second doctor must inspect the body
• Late neonatal death A live-born infant that dies
and form the view that the cause of death is as stat- after 7 days but within 28
ed by the other practitioner. The final part is com- days
pleted by the medical referee of crematoria for the • Stillbirth A fetus that dies before birth
local government authority involved. He or she but after a presumed 24
has to affirm that the particulars on the other parts weeks of gestation
• Perinatal death A combination of stillbirths
of the form are reasonable and have been complet-
and early neonatal deaths
ed by properly qualified doctors.
• Postneonatal deaths Deaths from 1 month to 1
year of age
• Infant deaths Deaths under 1 year of age
Stillbirths and infant deaths
Epidemiologists are particularly interested in the
rate of stillbirths and infant deaths because they
are a sensitive indicator of the general health of the The Stillbirth (Definition) Act (1992) reduced
population and also reflect the quality of child from 28 weeks to 24 weeks the minimal gestation-
health services. Comparison of death rates be- al age by which a stillbirth is defined. This reclassi-
tween countries and the associated trends over fication led to an apparent increase in the stillbirth
time are of special interest. So that such compar- and perinatal death rates. The infant death rate in
isons can be made, agreed definitions and termi- the UK has fallen from around 150 per 1000 live
nology have been promulgated by the WHO (see births in 1900 to around 6 per 1000 in 2001. The
box). current very low rate limits the possibility of

PID8 5/21/04 11:33 AM Page 57

Health information and sources of data Chapter 8

Figure 8.2

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.
The presumed duration of the pregnancy, the
A spontaneous abortion is the expulsion of the statutory grounds for the operation and the place

PID8 5/21/04 11:33 AM Page 58

Chapter 8 Health information and sources of data

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
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

PID8 5/21/04 11:33 AM Page 59

Health information and sources of data Chapter 8

Table 8.1 Statutorily notifiable infectious diseases.

Hospital episode statistics
Under the Public Health (Control of Disease) Act
Hospital episode statistics (HES) provide data on (1984)
hospital admissions (including day cases) in Eng- Cholera Smallpox
land. For each episode the age, gender and ethnic Plague Typhus
group of the patient is recorded together with Relapsing fever
other information (e.g. length of stay, waiting
Under the Public Health (Infectious Diseases)
time, type of admission, specialty, diagnosis, oper- Regulations (1988)
ations). The system has been operating since 1989. Acute encephalitis Ophthalmia
Analyses are published annually and some tables Acute poliomyelitis neonatorum
are available on the web. Special analyses can be Anthrax Paratyphoid fever
undertaken, subject to ethical approval. The Diphtheria Rabies
website is http://www.doh.gov.uk/hes/. Dysentery (amoebic or bacillary) Rubella
Leprosy Scarlet fever
Leptospirosis Tetanus
General practice databases Malaria Tuberculosis
Measles Typhoid fever
For some years most general practices have been
Meningitis Viral haemorrhagic
using computers in their practices to assist in the
Meningococcal septicaemia fever
management of the practice and to maintain clini- (without meningitis) Viral hepatitis
cal records. There are many packages that are avail- Mumps Whooping cough
able. All of them have the facility to generate Yellow fever
prescriptions and maintain files of demographic
data, diagnoses and symptom description and re-
sults of investigations. In the late 1980s systems
were developed to harvest data from consenting
practices in order to create databases that could be phone and provide the formal certificate later. Sim-
used for research and other purposes. The data ab- ilar action will usually be taken in the case of non-
stracted from the general practices is anonymized notifiable infectious diseases (or outbreaks due to
in order to ensure patient confidentiality. The other causes, e.g. chemical poisoning) which may
main systems now available are the General require immediate investigation. This is usually
Practice Research Database (GPRD) (owned by the undertaken by the environmental health officer
MHRA) and Mediplus (owned by Intercontinental employed by the local authority.
Medical Statistics). The importance of complete and prompt notifi-
cation is not universally appreciated and therefore
many infectious diseases are under-reported. Noti-
Infectious diseases
fication is important for a variety of purposes. In
When a doctor suspects that a patient is suffering the case of some infections, such as food- and
from a notifiable infectious disease or from food water-borne disease (food poisoning, typhoid,
poisoning (Table 8.1) the law requires him or her to etc.), bacterial meningitis (particularly meningo-
send a certificate to the Proper Officer designated coccal infection), infectious hepatitis, diphtheria
by the local authority. The officer responsible for and tuberculosis, immediate action may be re-
infectious disease control is usually the local Con- quired to limit the spread of infection and to safe-
sultant in Communicable Disease Controls guard public health. Notifications are also of value
(CCDC) employed by the Health Protection in studying the aetiological factors influencing the
Agency (HPA). When it is a disease that is likely to incidence of disease in the community and in
require urgent control measures to be taken, the monitoring the effectiveness of vaccination and
doctor will normally notify the CCDC by tele- immunization and other programmes.

PID8 5/21/04 11:33 AM Page 60

Chapter 8 Health information and sources of data

in 1974. Doctors are required to inform the Proper

Notifications of episodes of sexually Officer, who in this instance may be the CEHO
transmitted diseases (STDs) (Chief Environmental Health Officer) of the local
Genitourinary medicine clinics of the NHS are re- authority or the CCDC, of the occurrence of any of
quired to make regular returns to the Department the notifiable diseases they list. Doctors must also
of Health for England (or its equivalent in other report poisoning by the following substances.
countries of the UK) of the numbers of new atten-
dances with STDs. The following are defined as
STDs for this purpose. Industrial poison reporting

Aniline Compressed air

Arsenic Chrome
Benzene (chronic) Lead
Sexually transmitted diseases Beryllium Manganese mercury
Syphilis Herpes simplex Cadmium Phosphorus
Gonorrhoea Condylomata acuminata Carbon bisulphide
Non-specific Molluscum contagiosum Also cases of epitheliomatosis, toxic anaemia and toxic
genital infection Chancroid jaundice
Trichomoniasis Lymphogranuloma venereum
Candidiasis Granuloma inguinale
Scabies Other attendances requiring
Pediculosis pubis treatment The HSE publishes the number of reported cases
annually. Although the system is of great value in
controlling these diseases, it undoubtedly gives an
underestimate of the true incidence of these condi-
For two of the above diseases, syphilis and gon- tions. Many of the diseases occur in factories and
orrhoea, the age of the patient and whether the dis- work places in which there is no medical officer
ease was contracted outside the country has to be and, even if detected by another doctor, they are
stated. In no case are data given that could identify not always reported.
an individual. Employers also have an obligation to notify the
Although this system provides a useful picture of HSE of accidents (both fatal and non-fatal) which
the overall trends in STD it has to be appreciated occur in their factories or work place. These are
that not all cases treated are seen in NHS clinics. published in the annual report.
The nature of the information that is collected A third source of data relating to industrial dis-
means that it is of limited value for all but the most eases is notifications of ‘prescribed’ occupational
basic of epidemiological studies. In particular, it is diseases, for example pneumoconiosis in coal min-
not possible to distinguish episodes from numbers ers, tuberculosis in medical laboratory workers and
of people involved. Notifications of HIV infection mesothelioma in asbestos workers. Workers who
or AIDS are not included in clinic reports. Details of have these diseases are entitled to compensation
cases of AIDS are reported separately in confidence under current regulations. Thus, notification is
to the Director of the CDSC (Communicable Dis- probably more complete than for many other dis-
ease Surveillance Centre) of the Health Protection eases since there is a potential advantage in doing
Agency, which also collates reports from laborato- so to the individual with the disease. There are cur-
ries on HIV antibody-positive cases. rently about 50 ‘prescribed’ occupational diseases.

Industrial diseases and accidents Congenital malformations

In order to improve personal safety at places of A national scheme for the notification of congeni-
work, the HSE was established as a statutory body tal malformations was instituted in England and

PID8 5/21/04 11:33 AM Page 61

Health information and sources of data Chapter 8

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.

PID8 5/21/04 11:33 AM Page 62

Chapter 8 Health information and sources of data

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.

PID9 5/21/04 11:34 AM Page 63

Chapter 9
Indices of health and disease, and
standardization of rates

Introduction Ordinal variables

The health of a community can be measured by the These are used to rank the quality of characteristics
appropriate use of basic morbidity and mortality in order of severity, importance, etc.; for example,
data, including those from sources described in the pain might be classified as 0 = none, 1 = some pain,
previous chapter. To be intelligible, however, the 2 = severe pain, 3 = very severe pain. The analysis of
crude numbers need to be presented in a form that such variables requires different statistical tech-
allows valid comparisons to be made between niques to quantitative variables.
groups, between years and between different areas.
There are certain conventions in the handling and
Continuous quantitative variables
presentation of data. There are four main types of
variable as shown below: These measure attributes that can occur at any
• qualitative (categorical) point on a scale, for example height, weight or
• ordinal blood pressure. The degree of precision to which a
• continuous quantitative continuous variable is measured depends upon
• discrete quantitative. its intended use in a particular investigation
and the discriminatory power of the measuring
Types of variable

Discrete quantitative variables

Qualitative variables
These measure attributes that can occur only as
These are descriptive of a fixed attribute, for exam-
whole numbers (integers), for example the number
ple gender, religion, occupation and nationality.
of children born to a woman or the number of
Such data are sometimes labelled for convenience
deaths in a year.
by using numbers, e.g. 1 = male, 2 = female; or 1 =
Church of England, 2 = Roman Catholic, 3 =
Methodist, etc. These numbers have no meaning Grouping of data
other than as labels.
For convenience of handling and presentation,
continuous variables may be grouped as if they
were discrete. For example, body mass index (BMI)

PID9 5/21/04 11:34 AM Page 64

Chapter 9 Indices of health and disease

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
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.

Incidence and prevalence rates

Rates In order to demonstrate how incidence and preva-
lence rates are derived, the mortality and morbi-
It is rarely useful to state numbers of events alone.
dity experience of the employees in a hypothetical
These can be interpreted only when they are re-
factory is shown schematically in Fig. 9.1.
lated to a denominator, i.e. expressed as a rate; for
example, it is not helpful to say that the number
of deaths from pneumoconiosis is greater in coal Prevalence rates
miners than in, say, farm workers without relating
the figures to the numbers of people employed in The prevalence of a disease is the number of cases
the two occupations. of that disease in a defined population at a particu-
Two types of rate are frequently used: firstly lar point in time (point prevalence) or during a
events related to a population and secondly special specified period (period prevalence). Both are ex-
events related to total events. pressed as a rate (x per 1000 population). From Fig.
9.1, the point prevalence rate at the beginning of
the year was four per 18 people and at the begin-
Events related to the population ning of August it was two per 18 people (one had
died and one had left since the beginning of the
Examples year and two had joined). The period prevalence
• Birth rates are usually given as x per 1000 total for the year was 12 per 18 people (by convention
population per year. the denominator is the mid-year population). The

PID9 5/21/04 11:34 AM Page 65

Indices of health and disease Chapter 9

D – Died
J – Joined
L – Left

1 D
7 J

Subject number
11 D
14 L
15 L
16 J
18 D
20 L
Year A Year B Year C








Figure 9.1 Morbidity and mortality M
experienced in a hypothetical factory.

period prevalence approximates to the sum of the

Cumulative incidence
point prevalence at the beginning of the period
and the incidence during the period. Another method of expressing incidence is cumu-
lative incidence which takes the number of people
at the beginning of the study period as the denomi-
Incidence rates
nator and the number of events that occur in that
The incidence of a disease or other events is the population as the numerator. It is a useful measure
number of new cases that occur during a specified when estimating the likelihood that someone will
period in a defined population. The most accurate contract or die from a particular disease.
way to estimate this is by calculating the denomi-
nator of each individual’s time at risk. This can
Case fatality rate
be approximated as the number of people in the
population at the mid-point in time multiplied by An important concept, which is similar to that of
the period of the study. Thus, from Fig. 9.1, in year cumulative incidence, is case fatality. Here the de-
B, the incidence of illness was 8. (The first illness in nominator is the number of people with the dis-
subject 11 and the illnesses in subjects 14, 19 and ease and the numerator is the number dying from
20 started before the beginning of the period that disease. The period at risk does not need to be
specified and are therefore discounted.) The specified but for some diseases such as meningo-
incidence rate is 8 per 18 people per year, i.e. the coccal meningitis or myocardial infarction it is
mid-year population is used as the denominator. usually relatively short. Case fatality rates for can-
It should be noted that the incidence for a specific cers are measured over longer periods.
period is only valid for that period. Thus, in the 6
months January–June of year B, the incidence of
Error in health information
disease was 5; it is clear that it cannot be multiplied
to give an incidence of 10 during a 12-month The value of data ultimately depends on how accu-
period. rately they reflect the true frequency of the disease

PID9 5/21/04 11:34 AM Page 66

Chapter 9 Indices of health and disease

(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

Errors affecting mortality and morbidity data

The recording system
Those affecting the numerator, e.g.
The completeness and comparability of different
• diagnostic inaccuracy
• incomplete identification of cases sources of data may be affected by: the doctor’s
• variability of the recording system view of the value of records; the simplicity and ef-
ficiency of a records system; and changes in the
Those affecting the denominator, e.g.
• population migration conventions for coding and classification of dis-
• changes in population structure ease or rules for selecting priorities among multiple
• changes in administrative boundaries diagnoses.

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.

PID9 5/21/04 11:34 AM Page 67

Indices of health and disease Chapter 9

• 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.

Death rates from IHD Population of

in England and Wales armed services Expected deaths
Age group (years) (per 1000) (Mx) (1000s) (Px) (Mx ¥ Px) Observed deaths

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

Total 85.79 148

observed 148 ¥ 100

SMR = ¥ 100 = = 173
expected 85.79
Note: SMRs for occupational subgroups are usually confined to people aged 15–64 years because the working population
is confined to this age group.

PID9 5/21/04 11:34 AM Page 68

Chapter 9 Indices of health and disease

amongst men in the armed forces after adjusting

Direct standardization
for age distribution is higher than the national ex-
perience by a factor of 1.73. Direct standardization for age involves calculating
Another example of how standardization can be the age-specific death rates in the study population
helpful is in comparing mortality in different and applying them to the same age groups in a
years. The age structure of the population of Eng- ‘standard’ population. This can be real or hypo-
land and Wales has been changing for many years thetical. In this way, the number of deaths that
and therefore crude death rates can give a mislead- would have occurred in the standard population,
ing impression of changes in mortality. The SMR had it experienced the same death rates as the
gives a clearer indication of the true picture (Table study population, can be computed and compared
9.2). This indicates that mortality in males in Eng- with other groups. The method of direct standard-
land and Wales declined between 1965 and 1973. ization is shown in Table 9.3.

Table 9.2 Indirect standardization: mortality in males in England and Wales in 1965 compared with 1973.

Death rate per 1000 Male population in

males in England and England and Wales Expected deaths
Age group (years) Wales, 1965 (Mx) (1000s), 1973 (Px) (Px ¥ Mx) Observed deaths

<1 21.8 355.3 7746.0

1–4 0.9 1561.7 1406.0
5–14 0.5 4037.3 2019.0
15–24 1.0 3534.2 3534.0
25–34 1.1 3337.5 3671.0
35–44 2.5 2877.6 7194.0
45–54 7.4 3033.6 22 449.0
55–64 21.4 2643.1 56 562.0
65–74 53.0 1855.9 98 363.0
75+ 118.4 639.7 75 740.0
85+ 242.4 112.8 27 343.0

Total 306 026.0 296 546

observed 296 546

SMR = ¥ 100 = ¥ 100 = 97
expected 306 026
Table 9.3 Direct standardization: standardization of mortality in England and Wales in 1949 against the 1979
population. The age-standardized 1949 death rate (against the 1979 population) is 339 396.5/24 002 = 14.14 per 1000.
This can be compared directly with the crude death rate for 1979 which was 12.41 per 1000.

Age group Population, Deaths, Death rate, Population, Expected

1949 (a) (1000s) 1949 (b) 1949 (b/a) 1979 (c) (1000s) (c ¥ b/a)

0–9 3417 17 643 5.16 3339 17 231.3

10–19 2869 2345 0.82 4063 3331.7
20–29 3339 5031 1.51 3534 5336.3
30–39 3189 6839 2.14 3326 7117.6
40–49 3178 16 062 5.05 2020 14 241.0
50–59 2335 32 097 13.75 2924 40 205.0
60–69 1727 60 580 35.08 2257 74 661.6
70–79 957 77 127 80.59 1384 111 536.6
80+ 228 42 218 185.17 355 65 735.4

Total 21 239 260 278 24 002 339 396.5

PID10 5/21/04 11:36 AM Page 69

Chapter 10
Medical demography

Europe would have been insufficient to support so

large a population. Furthermore, the technology to
Despite the presence of many serious endemic create a safe urban environment, with pure water,
diseases and the occurrence of major epidemics adequate sanitation and means for the bulk trans-
and wars, the populations of most European port of food, did not exist until recently.
countries increased substantially between the 16th The populations of most other parts of the world
and 20th centuries. In recent years there has been began to increase more recently and their rate of
a reduction in the rate of growth of populations increase has reached that prevailing in Europe in
in many European countries. It seems likely that the 18th and 19th centuries only during the last
in the foreseeable future either they will remain century. An important difference between the
stable or there might even be a modest contemporary situation in many of the poorer
decrease. developing countries of the world and Europe in
The growth of the European resident population previous centuries is that there are no longer large,
since the 17th century underestimates both the ex- sparsely populated continents rich in natural re-
tent to which the numbers of European people in- sources that can be colonized and in which people
creased and the rate at which the increase took can thrive. Thus, population growth, which in pre-
place. Over several hundred years people emigrat- vious generations was regarded as a national prob-
ed in large numbers mainly to the Americas, Aus- lem, is now a world problem. It is forecast that if
tralasia and parts of Africa. The migrations were the prevailing rates of growth are sustained, the
prompted by economic hardship, social pressures world’s population, now about 6.4 billion people,
and religious persecution as well as for trading rea- will reach 9 billion by 2050. The earth’s mineral
sons and fortune hunting. A large proportion of and energy resources are finite and the rate at
the present populations of North America and Aus- which they are being consumed, particularly by
tralasia are descendants of these migrants. Their the industrialized countries, is increasing. In
numbers now exceed those of the parent many parts of the world, there is a hopeless
(European) populations. inability to meet local needs, and resources are
Whether or not the European population would inadequate to enable them to import essential
have increased in size to the extent that it has with- commodities. It is predicted that, unless there are
out migration and dispersal throughout the world reductions in both the rate of population growth
can only be a matter of speculation. It is unlikely and the rate at which natural resources are
that it would have done, as the natural resources of consumed, there will be a catastrophic failure to

PID10 5/21/04 11:36 AM Page 70

Chapter 10 Medical demography

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

PID10 5/21/04 11:36 AM Page 71

Medical demography Chapter 10

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

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

70–79 1821 2001

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.

PID10 5/21/04 11:36 AM Page 72

Chapter 10 Medical demography

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 Birth rate Death rate doubling Life expectancy, Life expectancy,
Country (1000s) (per 1000) (per 1000) time (years) males (years) females (years) Fertility

Kenya 25 905 47.0 11.3 19.8 56.5 60.5 7.0

Cambodia 5729 41.4 16.6 28.3 47.0 49.9 4.7
Mexico 81 141 29.0 5.8 30.2 62.1 66.0 4.2
Argentina 32 609 21.7 7.9 50.6 65.5 72.7 3.0
Singapore 2705 17.8 5.0 54.5 68.7 74.0 2.0
New Zealand 3380 17.8 7.8 69.6 71.9 78.0 2.2
India 844 324 29.9 20.3 72.5 55.4 55.7 4.0
Canada 25 309 15.3 7.3 87.0 73.0 79.8 1.8
USA 248 710 16.3 8.6 90.4 71.8 78.6 2.0
Ireland 3523 14.9 8.9 115.9 71.0 76.7 2.2
Russian Fed. 143 585 14.6 10.7 178.1 64.2 74.5 2.0
Sweden 8635 14.3 11.0 210.4 74.8 80.4 2.1
Japan 123 611 9.9 6.7 217.0 75.9 81.8 1.6
UK 57 367 13.8 11.3 277.6 72.4 78.0 1.8
Spain 39 025 10.2 8.5 408.1 73.2 79.1 3.7
Belgium 9844 12.0 10.6 495.5 70.0 76.8 1.6
Denmark 5148 12.4 11.9 1386.6 71.8 77.7 1.6
Italy 57 052 9.8 9.3 1386.6 73.2 79.7 1.3

Population doubling time

World region Population (millions) Birth rate (per 1000) Death rate (per 1000) (years)

Africa 662 45.0 15.0 23.5

Latin America 457 29.0 7.0 31.9
Asia 3171 28.0 9.0 36.8
Oceania 27 19.0 8.0 63.4
Former USSR 291 18.0 11.0 99.4
North America 278 15.0 9.0 115.9
Europe 500 13.0 11.0 346.9

World 5385 27.0 10.0 41.1

PID10 5/21/04 11:36 AM Page 73

Medical demography Chapter 10

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

100 65–74 year olds

Death rate per 10-year period (percentages)


70 5–9 year olds

60 Crude death rate


35–44 year

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.

PID10 5/21/04 11:36 AM Page 74

Chapter 10 Medical demography

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

70 Infectious
Deaths per 1000 population

60 Non-infectious
16 diseases
Expected years of life

1841 1979 8
10 1838 ‘39 ‘40 ‘41 ‘42 1969 ‘70 ‘71 ‘72 ‘73
0 10 20 30 40 50 60 70 80
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.)

PID10 5/21/04 11:36 AM Page 75

Medical demography Chapter 10

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
Death rate per 100 000 population

260 for a further 20 years.

220 Other infectious diseases that ceased to be a
200 major cause of mortality included scarlet fever,
160 which was endemic and had regular superimposed
120 epidemics. Its eventual elimination could have
100 been due to the advent of more successful treat-
60 ments for the complications of the disease or to
20 changes in the virulence of the organism.
Many of the measures that have achieved the
1851 ‘55 ‘59 ‘63 ‘67 ‘71 ‘75 ‘79 ‘83 ‘87
Year control of the infectious diseases are to a large ex-
tent by-products of improvements in the quality of
Figure 10.7 Mortality rates from enteric fevers, England life and, more recently, relatively simple medical
and Wales, 1851–89. measures. All should be applicable and are being
applied in poorer countries of the world at the pres-
ent time with consequent accelerating reductions
300 in their mortality levels.
Death rate per 100 000 population

180 100
100 000 population

Death rate per

80 50
40 25
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.

PID10 5/21/04 11:36 AM Page 76

Chapter 10 Medical demography

age of marriage. It is, however, used as a means of

Factors affecting fertility in
reducing population growth in some countries,
notably China. Although conception may still take
It has been shown that reductions in mortality place below the minimum age for marriage, the
have been achieved either as by-products of tangi- pregnancy is stigmatized as illegitimate. The con-
ble and universally acceptable improvements in ventional age of marriage tends to be several years
the environment or from certain specific medical greater than the legal minimum. In 19th-century
measures, such as vaccination, which reduce the Sweden, the conventional age of marriage was the
risk of contracting diseases. By contrast, reductions middle to late twenties. This convention was im-
in the fertility of a population require the consent posed in rural communities by obliging a man to
and cooperation of individuals together with demonstrate his ability to support his wife before
changes in their personal attitudes to child bearing marriage could take place and by his living apart
and child rearing. In pretransitional populations, it from women during the period he was becoming
is necessary for women to bear large numbers of established. Although the age-specific legitimate
children in order that there will be sufficient sur- fertility rates of Swedish women at this time were
vivors to maintain the size of the community and close to the maximum possible, the effective dura-
to provide a work force to sustain essential activi- tion of the fertile period was reduced. In Ireland,
ties. Strong social customs and belief systems have there has always been a tendency for late marriage.
grown up to support this need and therefore In England and Wales there have been significant
changes in fertility depend on changes in social changes in the age of first marriage during the past
customs and ethics. Next, it is necessary to 50 years (Table 10.2). The effect of the proportion
promote and make available fertility control to of women who are married on age-specific birth
individuals. rates is obvious from Table 10.3.

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
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

Legitimate births per Percentage of women Births per 1000

1000 married women who were married women (married and single)

1939 252 33 93
1969 251 58 157
1988 212 30 95

PID10 5/21/04 11:36 AM Page 77

Medical demography Chapter 10

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
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

PID10 5/21/04 11:36 AM Page 78

Chapter 10 Medical demography

not available to them. The problem of cost is great-

Some recent changes in the
est in countries with the greatest problems.
patterns of fertility in England
The efficiency of a particular method of contra-
and Wales
ception is assessed by the number of conceptions
per women-years of use. The assessment should be For about 50 years until the 1980s, there was a ten-
made in a group similar to that in which the dency for women to marry earlier. Since then there
method will be used. Table 10.4 shows estimates of has been a steady rise in the age of first marriage.
relative efficiency of some of the current methods. The mean interval between first marriage and the
These estimates were made in married women who birth of the first child fell until the early 1970s
were likely to have regular intercourse and to be when it began to increase (Fig. 10.11). The increase
motivated to use the method correctly. Some cou- in the interval between first marriage and first
ples use contraceptive methods incorrectly. For ex- pregnancy was associated with an increase in the
ample, there is some evidence that some women use of efficient contraception, particularly oral
are erratic in their use of oral contraceptives which contraception and the IUD. The mean interval
alters the apparent effectiveness of the method.
In 1969 and again in 1995, a great deal of public-
ity was given to the risk of venous thromboem-

Packets of oral contraceptives (1000 000s)

Total births per 1000 women aged 15–44

bolism associated with oral contraceptives and a

large number of women precipitately stopped 94 General 28
fertility rate
using them. They did not appear to use alternative 90 24
methods and consequently the decline in birth 86 20
rate in England and Wales was temporarily halted
82 16
(Fig. 10.10).
78 12

Table 10.4 The relative efficiency of different methods of 74 Oral 8

contraception. contraceptive
70 2
Pregnancies per 100
Contraceptive used women-years of use 1964 ‘65 ‘66 ‘67 ‘68 ‘69 ‘70 ‘71 ‘72 ‘73
Oral contraceptives 0.15
Intrauterine device 2.00
Figure 10.10 General fertility rate and oral contraceptive
Diaphragm 2.40–5.00
sales, showing the effect of the 1969 ‘pill scare’.

30 30

Age at first birth

25 25

Age at marriage
20 20

15 15
Interval in months
10 10

Figure 10.11 Average age of women

1938 ‘43 ‘48 ‘53 ‘58 ‘63 ‘68 ‘73 ‘78 ‘83 at marriage and average age at birth
Year (1938–83) of first legitimate child in England and

PID10 5/21/04 11:36 AM Page 79

Medical demography Chapter 10

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
sum of the rates as the expected number of live
Marriages with a birth within

Age 16–19
births per woman of child-bearing age. Thus, in
8 months (percentages)

30 Table 10.1, we can see the UK total-period fertility

25 rate is 1.84 which is below the replacement level of
20 Age 20–24 2.0. In fact, because some children die before they
15 reach reproductive age, the replacement total-
period fertility rate is about 2.1 in the UK, and in
Age 25–29 countries with high infant and child mortality the
rate will be even greater.
1952 ‘57 ‘62 ‘67 ‘72 ‘77 ‘82 ‘87
Year (1952–87)
Fetal loss and infant mortality
Figure 10.12 Trends in known illegitimate conceptions in Fetal and infant survival rates are amongst the
England and Wales. most important factors influencing demographic
change. Fetal loss during pregnancy occurs in three

Births per 1000 women

2000 Fetal loss

1500 • Spontaneous abortion

Years women born • Induced abortion
1000 1920 • Stillbirths
500 1940
0 In developed countries, 15–25% of known con-
15 20 25 30 35 40
Age ceptions spontaneously abort. The true rate may be
as high as 40%. Sixty per cent of spontaneous abor-
Figure 10.13 Cumulative fertility of birth cohorts in tions have abnormal chromosomes. In the process
England and Wales. of demographic transition, changes in sponta-

PID10 5/21/04 11:36 AM Page 80

Chapter 10 Medical demography

Total known conceptions (1000s)






Figure 10.14 Conceptions in women
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

PID10 5/21/04 11:36 AM Page 81

Medical demography Chapter 10

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-
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.

PID11 5/21/04 11:37 AM Page 82

Chapter 11
Evidence-based medicine

fits and risks of available treatments (including no

treatment). Benefits and harm can be assessed in
One of the guiding principles behind the new NHS terms of either cost, or some measure of health
is that health care should be based on evidence. outcome. Much medical practice is based on
The idea of evidence-based practice was introduced anecdotal evidence and ‘experience’ which may be
into the UK relatively recently. The first workshop unreliable and biased. EBM promotes the concept
was organized by Professor David Sackett at Oxford that clinical (and health care) decision-making
in 1995. Medicine has not always been evidence should be based on the best patient/population-
based — that is why evidence-based medicine based studies. It uses a hierarchy of evidence, with
(EBM) has been regarded as a new discipline. Tradi- the highest quality normally coming from ran-
tionally the teaching of medicine has been based domized controlled trials (RCTs). These are not al-
on an apprenticeship-type system. Undergradu- ways possible and in such circumstances evidence
ate students observed their teachers as they prac- from other studies is used.
tised, learnt from them and emulated them. This
same tradition was continued in postgraduate
Hierarchy of evidence
EBM is the application of population-based re-
search to the care of patients. Increasingly patients • Systematic review of randomized trials
and those responsible for paying for health care ex- • Randomized controlled trial
• Cohort studies
pect evidence that the treatments offered are of
• Case–control studies
proven benefit. They want to know if a drug is pre- • Case series or case reports
scribed that it not only lowers the cholesterol level
or reduces the blood pressure but that it also pro-
longs life or reduces the risk of heart attack or
stroke. Many illnesses or diseases have a range of EBM came about because of the revolution in in-
treatments, all of which are effective. Some treat- formation technology. The rapid growth in the
ments are of unproven value and others although availability of electronic databases of the medical
effective can cause harm. For other conditions literature allowed original research papers to be
there are no effective treatment. When evaluating identified and retrieved rapidly. This can be done
interventions in an environment of limited re- without leaving the consulting room. Before the
sources it is desirable to compare the relative bene- advent of computerized databases many doctors

PID11 5/21/04 11:37 AM Page 83

Evidence-based medicine Chapter 11

relied on medical textbooks, supplemented by structure their questions. (P is for population, I

browsing the journals that crossed their door or for intervention, C for comparison and O for
picking up new ideas from observing the practice outcome.)
of consultant colleagues. Unfortunately the text- P—How are a group of patients similar to the one
books are usually out of date and many of the jour- in the clinical scenario described? Patients in-
nals were little more than medical newspapers, cluded in the published studies may differ in
which published articles that were newsworthy some respects to the patient or clinical scenario
rather than of scientific merit. Consultant practice encountered. Key factors such as age, gender,
was followed because it was believed that consult- diagnosis, ethnicity and so forth need to be
ants were better informed of new developments considered.
and better equipped to appraise new studies criti- I—Which main intervention, treatment (or expo-
cally. Whilst this was often true, there can be much sure or prognostic marker) is being considered?
variation in consultant practice and this can lead C—What is the main alternative with which this
to uncertainty in decision-making. manoeuvre or treatment should be compared?
Many studies on new therapies are compared
against placebo. This is required by the licens-
Practising EBM
ing authorities to demonstrate efficacy but
EBM is used by clinicians to help their decision- clinicians want to know whether the new
making. It utilizes a structured approach, involv- treatment is more effective than the treatment
ing five key steps. These include taking a clinical usually offered.
scenario and from this identifying the key ques- O—What can be accomplished with the patient —
tions that are needed for the management of the e.g. increased length of life, shorter hospital
patient. That question should then be formulated stay, less pain, etc? The outcome must be
in such a way that it can be answered through use measurable. There are objective measures of
of the medical literature. The structured question health or disease that can be used to assess
produces key words which are used to help formu- one treatment or group against another.
late the search strategy that identifies the relevant
papers. The papers then have to be critically ap-
Searching for the evidence
praised and the evidence synthesized and used to
help clinical decision-making. The key search terms are found from the structure
of the question, particularly the intervention/ex-
posure of interest and the outcome of interest. The
search strategy should be as sensitive as possible —
Structured approach to EBM it should identify all the relevant papers to help
answer the clinical question. Specificity is obtained
• Clinical scenario
• Structured question by combining search terms and by using the rele-
• Search for the relevant papers vant filters for the different types of question. (For
• Critically appraise the evidence example, a filter can be developed that restricts the
• Use evidence to help decision-making search to randomized trials.)

Critically appraising the evidence

Forming structured questions
Once a relevant paper has been identified it is im-
Forming an answerable question from a clinical portant to be able to appraise it critically. This is
scenario is the first step in practising EBM. It is a done by asking three simple questions.
discipline that requires practice. Practitioners of • How valid is the study? Is it well designed and
EBM often use the acronym PICO to help them carried out in an appropriate population?

PID11 5/21/04 11:37 AM Page 84

Chapter 11 Evidence-based medicine

• 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).

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

Intention to treat analysis

In reality there are always some patients who do not
The method of randomization can introduce bias receive the treatment to which they were random-
and influence the generalizability of the findings. ized. By analysing the results using intention to
It is therefore important to know how the random- treat analysis any bias due to unplanned drop-outs
ization was carried out in order to be able to assess or cross-overs will be avoided. (A cross-over is when
whether this is likely to have influenced the re- a patient is allocated to treatment A but actually re-
sults. It is particularly important to check that the ceives treatment B.) If on intention to treat analysis
staff involved in recruiting subjects to the study there is still a treatment effect then this is likely to
were not also responsible for the randomization. be a true effect. Analysis by actual treatment group
is also usually worthwhile and is more likely to
show a statistically significant difference between
Characterization of the groups
groups. With placebo-controlled trials it has been
In the paper there should be a table showing the shown that compliant patients who take their
characteristics of the treatment groups being com- placebo have a better outcome (up to 30% better)
pared. Sometimes by chance, particularly in small than the non-compliant patients. Including drop-
studies, the groups may be unequal (e.g. more men outs in the placebo group can introduce bias. Again,
in one group) and this can cause bias. It should be intention to treat analysis removes this bias.

PID11 5/21/04 11:37 AM Page 85

Evidence-based medicine Chapter 11

Blinding • An estimate of the response rate in one of the

groups — if the outcome of interest is a rare event,
In a single blind randomized trial the patient is un-
then a larger sample size will be needed.
aware which treatment they are receiving. This is
• Level of statistical significance — this is usually at
important when the assessment of the outcome of
least the 95% level to avoid a Type 1 error.
interest is subjective — such as pain, anxiety, etc. In
• The value of the power desired — this is usually
a drug trial this can be achieved by giving one
set at the 80% level because missing a small but
group of patients a placebo. In a double-blind trial
true difference is less important than identifying
both the patient and the investigator should not
a spurious positive association. Greater power re-
know which treatment they received. This then re-
quires a bigger sample size and will incur addi-
moves possible bias in both the reporting and
tional costs.
recording of the outcome of interest.

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.

When designing a study it is important to ensure

Critical appraisal of
that Type 1 and Type 2 error is avoided.
systematic reviews
The size of the sample required is determined by:
• Difference in response rates to be determined — A systematic review involves identification of all
this should be a clinically significant difference. the relevant primary papers in human populations

PID11 5/21/04 11:37 AM Page 86

Chapter 11 Evidence-based medicine

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
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
• 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

PID11 5/21/04 11:37 AM Page 87

Evidence-based medicine Chapter 11

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.
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

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

expect a long and healthy life. Nowadays if illness

occurs it is assumed that modern medicine can or
The health of a population depends both on the ought to be able to restore the sufferer to normal
provision of health care for the sick and on public health. These changed expectations have been
health services to promote health and prevent the brought about to a large extent by the publicity
spread of disease. Until the middle of the 19th cen- given to the more dramatic advances in medical
tury the state accepted little responsibility for knowledge and treatments and by the evident suc-
health. The first attempts to improve public health cess of modern medicines in reducing mortality,
in the UK involved legislation, beginning with the particularly during infancy and childhood. The
Nuisances Act of 1846, which aimed to remove public also feels a sense of ownership of the health
sewage and offal from the streets. This was fol- service and expects ready access to it when needed.
lowed by a series of further public health acts. At Although it is true that during the past 50 years the
the beginning of the 20th century the Govern- scope and effectiveness of medical treatments have
ment turned its attention to personal medical care. been extended greatly, it is also true that many of
David Lloyd George introduced insurance-based the diseases which commonly affect humans are
health care for workers in 1911. The majority of self-limiting and that medical treatment does little
health care, however, was still obtained privately, to alter their natural course. Furthermore, few of
or through friendly societies and charitable institu- the diseases that result in death or major disability
tions. The state took no major role as a provider of can be cured. The main impact of modern medi-
health care until the inception of the NHS in 1948, cine has tended to be to allow people to live longer
which promised access to free health services for and more comfortably with their diseases. The
all. Today, the Secretary of State for Health is re- public often fails to appreciate this.
sponsible to Parliament for the work of the Depart- For many of the major diseases, it is both logical
ment of Health (DoH) whose aim is to improve the and desirable to take steps where possible to prevent
health and well-being of people in England (see their occurrence. Even if a treatment eventually be-
DoH website at http://www.doh.gov.uk/). Separate comes available, a strategy of prevention would
arrangements are in place for Scotland, Wales and usually be more cost effective in improving both
Northern Ireland. public and personal health. In future, it is hoped
Earlier generations tended to accept ill health that medical research and practice will give greater
and premature death as unavoidable hazards of attention to the means whereby health can be pro-
human existence. Over time people have come to moted and diseases prevented. Historically, infec-

PID12 5/21/04 11:38 AM Page 92

Chapter 12 General principles

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-

PID12 5/21/04 11:38 AM Page 93

General principles Chapter 12

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

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)

Figure 12.1 Interactions of agent, host and environment, Figure 12.2 James Lind (1716–94) author of the treatise
causing disease. on scurvy.

PID12 5/21/04 11:38 AM Page 94

Chapter 12 General principles

Strategies related to the agent

If the agent can be identified, it may be possible to
remove or destroy it at source. For example, by
ceasing to use asbestos as an insulating material,
the incidence of mesothelioma has been reduced;
the control of bovine tuberculosis in humans was
achieved by eradication of the disease from milk-
ing herds and pasteurization of milk.

Strategies related to the environment

These include attention to general environmental
factors such as standards of housing, nutrition,
working conditions, water supplies, sewage dispos-
al and the control of environmental pollution.
Environmental measures directed at the specific
causes of individual diseases are also important
and people may be protected from potentially in-
jurious agents by the construction of barriers be-
Figure 12.3 John Snow (1813–58) epidemiologist who tween them and the source of harm. Examples of
studied the transmission of cholera. such measures include the prevention of transmis-
sion of food-borne infection by hygienic food
production methods; elimination of vectors, for
example action to prevent the spread of malaria or
of diseases and of the factors that determine the
yellow fever by mosquito control; and the use of
balance between agent, host and environment is
machine guards in industry to reduce the risk of
helpful in order to construct appropriately directed
preventive and control programmes. Epidemiolog-
ical studies are used to identify the causal agents
and those elements in the environment or in peo- Strategies related to humans
ple’s behaviour and personal characteristics that are
There are three strategies involving individuals.
key determinants of the natural history of disease.
• The enhancement of general or specific resist-
ance to disease, i.e. by improved nutrition or
Intervention strategies immunization.
• The modification of personal behaviour, i.e. by
Based on the knowledge gained from epidemiologi-
encouraging people to adopt healthier lifestyles
cal studies three main types of intervention
by not smoking, moderating alcohol intake, im-
strategy may be adopted.
proving diet, avoiding obesity, exercising regularly,
• The use of screening to detect predisposing con-
ditions or the early stages of disease when action
Intervention strategies
can be taken to prevent its onset or control its
Strategies related to: progress, for example tuberculin testing for tuber-
• Agent
culosis, blood pressure measurement to identify
• Environment
hypertension, or mammography for breast cancer
• Humans

PID12 5/21/04 11:38 AM Page 95

General principles Chapter 12

ple with manifest disease by modifying continuing

Preventive action
risk factors such as smoking and by the implemen-
Action is usually classified as follows. tation of effective rehabilitation.

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.

PID13 5/21/04 11:39 AM Page 96

Chapter 13
Health promotion and
health education

Introduction The new public health

A new public health initiative was heralded by the
The terms health promotion and health education
Lalonde Report for the Canadian Government
are sometimes confused. Both are strategies aimed
(1974), which incorporated health promotion as
at improving the public health, but while the
an integral part of the government strategy to im-
concepts are complementary they are not
prove public health. Lalonde identified four main
influences on people’s health.
Health promotion involves the empower-
ment of the community in improving its health
through education, through the provision of pre-
ventive health services and by improvement of the Lalonde’s four health factors
social, physical and economic environments. 1 Genetic and biological factors
Health education is the empowerment of in- 2 Behavioural and attitudinal factors—the so-called
dividuals through increased knowledge and under- lifestyle factors
standing, but does not involve the political 3 Environmental factors, which include economic, so-
advocacy necessary in health promotion. cial, cultural and physical factors
4 The organization of health care systems
The health strategies that emerged during the
19th century were in some ways similar to those
that we now term health promotion. Thus, Medi-
cal Officers of Health worked for local authorities A growing awareness of the factors that influ-
with the aim of improving the environment, en- ence health encouraged people with an interest in
couraging healthy public policies, introducing pre- prevention to involve organizations and institu-
ventive strategies (e.g. sanitation and vaccination) tions not usually primarily concerned with health.
and encouraging better health through education. This led to the concept of Healthy Cities, which
Another step in the development of health promo- also originated in Canada and was subsequently
tion was the Peckham Pioneer Health Centre proj- embraced by the World Health Organization
ect, which began in south London in the 1930s. It (WHO), spreading throughout the world. In the
provided conventional health care and health edu- UK, many health promotion initiatives were co-
cation together within an environment that sup- ordinated under this umbrella, first in Liverpool
ported community development through the and later in Manchester, Newcastle, Camden, Belfast
provision of recreational and sports facilities. and Glasgow. More information about Healthy

PID13 5/21/04 11:39 AM Page 97

Health promotion and education Chapter 13

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.

Health Promotion is the process of enabling

Developing personal skills To support social and per-
people to increase control over, and to improve,
sonal development through the provision of infor-
their health. To reach a state of complete physi-
mation, health education and the development of
cal, mental and social well-being, an individual
individual skills.
or group must be able to identify and to realize
aspirations, to satisfy needs and to change or
Reorientating the health services To encourage
cope with the environment. Health is therefore
health service providers to look beyond their man-
seen as a resource for everyday life, not the ob-
date for clinical and curative services and ensure
jective of living. Health is a positive concept em-
that health services are aimed at the pursuit of
phasizing social and personal resources, as well
health rather than only the cure of illness.
as physical capabilities. Therefore, health pro-
motion is not just the responsibility of the
The principles of the Ottawa Charter were
health sector, but goes beyond healthy life-styles
adopted in various ways by many countries
to well-being.
throughout the world, but the initial enthusiasm
It also proposed that: ‘health promotion should seems to have waned. The UK adopted health
focus on equity in health and reducing differences targets in line with ‘Health for All by the Year 2000’
in health status by ensuring equal opportunities in 1990, and in 1999 a new set of goals were
and resources to enable all people to achieve their outlined in Our Healthier Nation. These targets are
fullest health potential’. The five areas for health aimed primarily at action by the health services
promotion action were as follows. without a commitment to changes in public policy.
They include targets to improve health outcomes
in relation to cancer, coronary heart disease and
The Ottawa Charter stroke, accidents and mental health. There are a
1 Building healthy public policy
number of difficulties in adopting the health pro-
2 Creating supportive environments motion approach. The long interval between the
3 Strengthening community action adoption of preventive strategies and measurable
4 Developing personal skills improvements in health means that organizations
5 Reorientating the health services see little short-term return on their investment.
The processes of community consultation, health
education and altering public policies are time
Building healthy public policy To encourage policy consuming, and are often politically controversial.
makers in organizations and government to place Many health promotion programmes have been
health on their agenda. This may include efforts to initiated without a clear commitment to evaluate

PID13 5/21/04 11:39 AM Page 98

Chapter 13 Health promotion and education

their outcomes. Given the limited health budget,

Health promotion in the UK
it is not acceptable to institute unproven in-
terventions, whether they involve conventional The Health Development Agency is the Depart-
medical treatment or a health promotion pro- ment of Health’s health promotion arm and suc-
gramme, unless they are rigorously and scientifi- ceeded the Health Education Agency in January
cally tested. 2000. Its website is http://www.hda-
The emphasis that many politicians and others online.org.uk/. The Agency is a special health au-
have placed on personal responsibility for health thority. Its aim is to identify the evidence of what
has been criticized because it ignores the economic works to improve people’s health and reduce
and social influences. This can be illustrated by health inequalities. Then, in partnership with pro-
considering smokers who suffer ill health. They are fessionals, policy makers and practitioners, it will
blamed for the outcome of their voluntary action develop guidance and work across sectors to get
whilst the advertising of tobacco products in evidence into practice. Members of the Board of
many countries continues to be permitted and the Authority are appointed by the Secretary of
the companies who promote them take no State for Health and include leading figures from
responsibility for the adverse outcome. Similarly, health, associated professions, the media, educa-
children who grow up in impoverished homes, tion and related fields.
lacking education and with little hope of employ- Primary care trusts are also charged with improv-
ment, have bleak futures and may be unable to re- ing the health of the population for which they are
spond to the admonition of those from more responsible. Most of their budgets are committed
privileged backgrounds to change their ways. to the provision of personal health services, but
(These issues were discussed in the Black Report some of their resources are allocated to health pro-
referred to on p. 5.) motion. Often this is through specialist health
Another issue relating to the effectiveness of promotion staff. These staff use a combination of
health promotion programmes concerns the health education and community support to target
dilemma of whether to adopt a population strategy particular issues. They tend to concentrate on
or a targeted strategy. The former involves at- high-profile issues such as cervical cancer, HIV or
tempting to achieve health gain through actions heart disease.
involving the whole population while the latter fo- Voluntary bodies, such as the Royal Society for
cuses efforts on particular risks associated with the Prevention of Accidents, the British Heart
specific conditions. Both approaches have their Foundation, Cancer UK or environmental groups
adherents, but scientific evaluation of their com- such as Greenpeace and the Friends of the Earth are
parative effectiveness is needed before one ap- all active in health promotion. Their contribution
proach or another is taken. An example of a to the provision of knowledge to individuals, influ-
population approach was the North Karelia Com- ence on public policy and help in reorientating the
munity trial, which aimed to reduce the incidence health services is increasingly recognized.
of heart disease in a Finnish community by means
of changes in people’s diet, smoking habits and ex-
Health promotion programmes
ercise compared with a control community. Health
promotion campaigns targeted at particular groups There are many different health promotion pro-
have also been used successfully, for example in grammes. Some leading examples of current
the effort to reduce the spread of HIV amongst in- activities are outlined below.
travenous drug users by the introduction of
needle-exchange schemes.
In the UK many different professional groups
and lay organizations are involved in health edu-
cation and health promotion.

PID13 5/21/04 11:39 AM Page 99

Health promotion and education Chapter 13

nicotine replacement therapy. This is another ex-

Health promotion
ample of how the health service can begin to move
Target areas include: from providing a curative approach to one where
• Smoking prevention and education is the goal. It is impor-
• Alcohol
tant to remember that most people start smoking
• Nutrition
when they are teenagers and thus strategies target-
• Exercise
• Sexuality ed at children have also been encouraged, for ex-
ample getting local authorities to enforce the law
on sales of cigarettes to the under 16s.

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-
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

PID13 5/21/04 11:39 AM Page 100

Chapter 13 Health promotion and education

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
• Increase the price of alcohol
through the British Nutrition Foundation
• 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
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
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

PID13 5/21/04 11:39 AM Page 101

Health promotion and education Chapter 13

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
The medicalization of contraception led doctors
Strategies to increase exercise to become involved in a number of other initia-
• Healthy public policy, e.g. cycle tracks
tives such as cervical screening and well women
• 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
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
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

PID13 5/21/04 11:39 AM Page 102

Chapter 13 Health promotion and education

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.

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

PID14 5/21/04 11:41 AM Page 104

Chapter 14 Control of infectious diseases

the community long after they recover from the

Transmission survival
Human immunodeficiency virus (HIV) is of par- Organisms vary in their capacity to survive in the
ticular interest because the reservoir of infection is free state and to withstand adverse environmental
human. All carriers are infectious. Infectivity is at conditions, for example heat, cold, dryness. Spore-
its highest around the time of seroconversion often forming organisms, such as tetanus bacilli which
when HIV infection has yet to be diagnosed and can survive for years in a dormant state, have a
again later when HIV disease (the symptomatic major advantage over an organism like the Gono-
phase) occurs. coccus which survives for only a very short time
outside the human host.

Other biological or environmental

Life cycle
These include:
• animals, for example Escherichia coli, rabies, The life cycle of certain organisms has important
malaria, psittacosis and hydatids; consequences in the spread of disease. Organisms
• foodstuffs, for example Salmonella, Campylobac- such as the malaria parasite which have a complex
ter and Listeria; life cycle requiring a vector are more vulnerable
• water, for example giardiasis, schistosomiasis than those with simpler requirements for transmis-
and cholera; sion. In many infections by such organisms,
• soil and the environment, for example anthrax, humans are an accidental host.
Legionella, tetanus.

Host susceptibility
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

• Direct contact — touching, kissing or sexual inter-

course, e.g. Staphylococcus, Gonococcus and HIV
• Vertical transmission (mother to fetus), e.g. hepatitis Age
B, Listeria, HIV, rubella and cytomegalovirus
The very young and the elderly are more suscepti-
• Inhalation of droplets containing the infectious agent,
ble to infectious diseases than are older children
e.g. tuberculosis, measles, influenza
• Ingestion of food or water that is contaminated, e.g. and younger adults. However, some common dis-
Salmonella, Giardia, Norwalk virus, hepatitis A eases of childhood such as measles, mumps and
• Injection either by human interference or by insects, chickenpox can be more serious when they occur
e.g. hepatitis B and C, tetanus, malaria in adolescents and young adults.

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-

PID14 5/21/04 11:41 AM Page 105

Control of infectious diseases Chapter 14

perience higher age-specific mortality rates than

Population (herd) immunity
females for most diseases.
The resistance of groups of people to the spread of
infection is termed population (or herd) immu-
nity. It depends on the proportion of individuals in
The state of nutrition of the host is very important. the population who are immune. If this is suffi-
For example, in developing countries, measles may ciently high, chains of transmission of the agent
have a mortality of 5% amongst those who are cannot be sustained because susceptible people in
poorly nourished whilst in the UK the case fatality the group are shielded from exposure to infected
rate is 0.02%. It is likely that the improvement in people by the immune people around them. The
nutrition during the 19th century was a major degree of herd immunity that will inhibit spread
reason for the reduction in deaths from communi- varies with different infections but is usually less
cable diseases at that time. than 100%. It depends on:
• the frequency of new introductions of infection;
• the degree of mixing which affects opportunities
for contact between infected and susceptible
Some individuals appear to have an exceptional people; and
susceptibility to infections, which is probably • the transmissibility of the infection and dura-
inherited. This can be seen in the similar suscepti- tion of infectiousness of excreters.
bilities of monozygotic twins and different suscep- Herd immunity affects the periodicity of epi-
tibilities of dizygotic twins to certain infections. In demics. So long as each case leads to more than one
national or ethnic groups, natural selection over new infection, the incidence of the disease in-
many generations may eventually breed a relatively creases and herd immunity rises. When herd im-
resistant stock. A good example of this phenome- munity reaches a level at which each case causes
non is the history of tuberculosis in Europe. During less than one new infection, incidence declines. As
the 19th century, the population experienced a individual immunity wanes or new, susceptible
high incidence of this disease which, by causing people are introduced to the group, herd immuni-
high mortality amongst susceptible young adults, ty again declines and the group is again vulnerable.
tended to favour the survival through reproductive This was well illustrated by the periodic epidemics
life of those with higher innate resistance. By con- of measles, which occurred every 2–3 years before
trast, when an infectious disease is first introduced the introduction of measles vaccination (see Fig.
into a community with no prior experience of it, 3.4). Introduction of vaccination programmes
the result can be disastrous. For example, the intro- lengthens the period between epidemics. The
duction of measles to the Greenland Inuits by the higher the immunization rate, the longer the peri-
American forces during the Second World War od. If the antigenic composition of an infectious
caused devastating epidemics with high mortality. agent changes or if an agent previously absent
Some genetic traits can be an advantage; for exam- from the population is introduced, there is no ben-
ple, carriers of sickle-cell disease have a positive ad- efit from herd immunity against that organism and
vantage when infected with malaria. large-scale epidemics may result. For example,
antigenic changes of the influenza virus from time
to time lead to worldwide pandemics.
The occurrence of disease in humans depends
Characteristics of the organism
upon the individual’s susceptibility to the agents
to which he or she is exposed. Defence mecha- The characteristics of the causal organism are also
nisms are natural and acquired immunity (see pertinent to the spread of infectious diseases.
Chapter 15) and population (herd) immunity. These include the following.

PID14 5/21/04 11:41 AM Page 106

Chapter 14 Control of infectious diseases

Some organisms are antigenically more potent

Organism characteristics
than others. Those that invade the bloodstream,
• Infectivity: capacity to multiply in host for example chickenpox, are more likely to pro-
• Pathogenicity: capacity to cause disease in host duce a good immune response than those organ-
• Virulence: pathogenicity in a specific host
isms that only infect surface membranes, for
• Immunogenicity: capacity to induce specific and last-
example the Gonococcus.
ing immunity in host
• Antigenic stability: can induce lifelong immunity

Antigenic stability
Organisms which are antigenically stable or exist
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.
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
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
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.

PID14 5/21/04 11:41 AM Page 107

Control of infectious diseases Chapter 14

The pattern of an epidemic depends on the bio-

Control of infectious diseases
logical properties of the agent, on whether or not
Some infectious diseases can have serious effects the environment is favourable to its survival and
on the health of a population if they are allowed to transmission, and on the immunity of the host
spread unchecked. They may cause epidemics or population. The course of an epidemic is therefore
the disease may become endemic.* In most west- a reflection of time, place and person interaction.
ern countries, such diseases are notifiable by law to Its investigation is an exercise in descriptive epi-
the public health authorities (see Table 8.1, p. 59, demiology. Epidemics are usually due to microbial
for list of infectious diseases notifiable in the UK). agents although they can arise from other causes,
As many of these diseases are food- or water-borne, such as chemical poisoning or mass psychogenic
the local government authority may be partly or illness.
wholly responsible for instituting environmental
control measures. In other infections, control may
be aided by use of vaccines and effective treatment
of cases. Before describing the different types of epidemics
Because of the numbers of people travelling and outbreaks and their investigation it is
around the world the transmission of diseases necessary to explain some of the terms used
between countries is becoming an increasing (Fig. 14.1).
problem. Severe acute respiratory syndrome (SARS)
and West Nile fever are recent examples. Diseases Primary or index case(s) This is the first case (or
that have originated or been endemic in one part group of cases) arising from the introduction of an
of the world are rapidly transmitted to a virgin agent into a community.
population. New measures are required to prevent
such diseases being carried from one country to Secondary cases People who acquire infection from
another. the primary/index case(s) are called secondary

Epidemics and outbreaks

Incubation period This is the interval between
The essential characteristic of an epidemic is that it infection of an individual and the onset of
involves a temporary increase in the incidence of a symptoms. This is different for each organism
disease, usually circumscribed both in its location and may vary for the same organism according to
and in respect of the groups affected. Rarely, a such factors as the virulence of the particular
worldwide epidemic of an infectious disease may strain, the infecting dose and the susceptibility of
occur (pandemic). The term outbreak is used to the host.
refer to the localized temporary increase in the in-
cidence of a particular disease where the cases are Serial interval/generation time This is the interval
potentially linked to each other. As few as two cases between the onset of primary and secondary cases.
of a disease, associated in time and place, in cir- This interval may be shorter or longer than the in-
cumstances where the disease is not a usual occur- cubation period depending on the duration of in-
rence and/or a particular threat are sufficient to fectivity of the primary case, which may start well
constitute an ‘outbreak’ requiring investigation, before and continue for some time after the onset
for example meningococcal infection. of symptoms. When infection in intermediate
cases is subclinical, the serial interval may be more
*An endemic infection is one that is usually present prolonged than usual.
in a given geographical area or population group at
relatively high prevalence and incidence rates in Derived infection This is an infection arising by
comparison with other areas or populations. direct transmission from an infected contact.

PID14 5/21/04 11:41 AM Page 108

Chapter 14 Control of infectious diseases

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

Case 1 Case 2

Case 1 Case 2

Case 1 (4d) Case 2 (5d) Case 3 (6d)

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

PID14 5/21/04 11:41 AM Page 109

Control of infectious diseases Chapter 14


Number of cases
20 Buffet
Figure 14.2 Number of cases accord-
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.)

14 Unvaccinated
Number of cases

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-

PID14 5/21/04 11:41 AM Page 110

Chapter 14 Control of infectious diseases

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

• Descriptive enquiries into the facts of the outbreak

• Investigation of reservoirs and vehicles of infection An epidemic may originate from an individual
• Analysis of the data collected who has had a minor clinical episode or from a car-
• Formulation of a causal hypothesis rier who was ill many years previously. Therefore, a
• Testing its validity in the control of the outbreak careful history should be taken from all contacts of
the patients.

Descriptive enquiries Animal

• Verify the diagnosis by clinical and laboratory Enquire about the contacts patients may have had
investigation of the cases. with sick animals or animal products known to
• Verify the existence of an epidemic by compari- harbour the infection concerned.
son with previous incidence of the disease in the
same population. Environment
• Compile a list of all cases and search for unre- Investigate sources of foods consumed by affected
ported cases by alerting hospitals and general prac- individuals and the circumstances of their produc-
titioners in the district and neighbouring districts. tion, storage, preservation and preparation. Par-
• Investigate patients and others who might be in- ticular attention should be given to looking for
volved in the outbreak. Record the personal char- situations in which cross-contamination or incu-
acteristics of the patients (age, sex, address, etc.) bation of organisms could have occurred. Arrange
and enquire into shared experiences or activities for laboratory examination of food remnants,
that could carry risk of exposure to the suspected milk, and water supplies, and other relevant speci-
agent, for example occupation, school attended, mens from environmental sources, for example
recreational activities, consumption of foods, kitchen utensils, drains, etc., and the typing of any
drugs, etc. organisms that are isolated.
• Identify the total population at risk, i.e. all those
who may have been exposed to the same hazards Analysis of the data collected
as the patients, whether ill or not. • Plot the epidemic curve. This may give some clue
• Ensure that all the clinical and laboratory inves- to the mode of spread and probable time of initial
tigations required to confirm the identity of the in- exposure. For example, an outbreak of Salmonella
fection in patients and to determine the extent of napoli caused by contaminated chocolate bars im-

PID14 5/21/04 11:41 AM Page 111

Control of infectious diseases Chapter 14

ported from Italy is shown in Fig. 14.4. Note the re-

Factors for hypothesis
lationship between the time distribution of cases
and the importation of bars of chocolate. • The properties of the agent, its reservoirs and favoured
• Plot the cases on a map. This will detect cluster- vehicles and also of the nature of the illness it causes
• The probable source and route of transmission. For
ing. The distribution of cases must be examined
this purpose the typing of the organisms may be particu-
with reference to that of the population at risk.
larly helpful
• Analyse the incidence rates in different groups. • Time and duration of exposure of the patients to the
This can be done, for example, for age or occupa- agent in relation to the onset of their illness
tion. A high rate in a particular group suggests that • Attack rates of the different subgroups of the popula-
the cause lies in a common experience of its mem- tion at risk
bers. Attack rates must be calculated both in those
exposed and in those not exposed to the suspected
agent. It should be noted that variations in the
biological response to infection may result in clin- Testing validity in the control of the outbreak
ical attack rates of less than 100% in the exposed Seek support for the causal hypothesis by further
population. investigation of cases, if necessary, to confirm the
• Look for a quantitative relationship. This may proposed explanation of their illness. Carefully de-
exist between the degree of exposure (or dose) and signed case–control studies may be very helpful in
attack rate, for example amount of suspect food this. Implement appropriate control measures on
consumed or closeness to a source of pollution. For the assumption that the hypothesis is correct and
example, in the outbreak of Salmonella typhimu- monitor their success in reducing the incidence of
rium referred to under ‘Common source epidemics’ further cases.
(p. 108), food histories were obtained from 266
delegates at the suspect meal. Of these guests, 196
Control of food-borne infection
reported illness. The food-specific attack rates
showed clearly that chicken was the probable vehi- The most frequently reported notifiable infectious
cle of infection (Table 14.1). diseases are food poisoning and gastrointestinal
infections. They illustrate well some of the biologi-
Formulation of a causal hypothesis cal and environmental factors that are conducive
The hypothesis should take account of the to the occurrence of outbreaks and the approach
following. to their investigation and control outlined above.

Recall of chocolate and health warning
202 Primary household cases
35 43 Secondary cases
Number of cases

Later importations of chocolate



March importation of
10 chocolate

Figure 14.4 Number of cases of in- 5

fection with Salmonella napoli from 0
chocolate during April–August 1982. April 4 May 2 June 6 July 4 August 1 September 5
(From Roberts JA, Sockett PN, Gill ON. Date
Br Med J 1989; 289: 1227.)

PID14 5/21/04 11:41 AM Page 112

Chapter 14 Control of infectious diseases

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.)

Eaten Not eaten

Food Ill Total % Ill Total % RR

Tuna 70 98 72 127 169 76 0.9

Ham 48 63 77 149 204 73 1.1
Beef 29 46 64 168 221 76 0.8
Salmon 38 46 84 159 221 72 1.2
Egg mayonnaise 67 89 76 130 178 73 1.0
Pâté 50 66 76 147 201 74 1.0
Beef sandwiches 10 13 79 187 254 75 1.1
Ham sandwiches 15 20 75 182 247 74 1.0
Chicken 182 213 86 15 54 29 3.0*
Quiche (cheese) 80 108 50 117 159 74 0.7
Quiche (ham) 18 21 86 179 246 73 1.2
Ham and turkey pie 103 137 76 94 130 73 1.0

*c2 = 70.7; P < 0.01.

They also exemplify the complementary roles

Sources of contamination
of the health agencies and local authorities
in the investigation and management of an Food may become polluted or infected at any stage
outbreak. during its manufacture and processing, distribu-
tion or preparation for consumption.

Causes of food poisoning

Food poisoning may be caused by either
microorganisms or chemicals. In the case of Salmonellosis usually owes its origin to the infec-
microbiological food poisoning, the food may tion of livestock through their food or by cross-
be either the vehicle whereby an agent is trans- infection within herds or poultry flocks.
mitted or the growth medium for the organisms.
For example:
Manufacture and processing
• salmonellosis may be caused by the organism
being transmitted from poultry to humans in eggs; In 1964 an outbreak of typhoid in Aberdeen was
• staphylococcal food poisoning may arise if dur- caused by corned beef which had probably become
ing preparation the food becomes infected from a contaminated by use of polluted water to cool cans
septic lesion in the food handler. If the food is then which had defective seals. The Lanarkshire out-
stored for long enough at a temperature which al- break of E. coli O157 noted above was due to con-
lows the organism to multiply, the toxins pro- tamination of cooked meat products prepared in a
duced may result in severe symptoms of food butcher’s shop.
poisoning in those who eat it.
The harmful effects of chemicals may arise
Storage and distribution
from either accidental contamination or the
deliberate addition of chemicals to food as Outbreaks of food poisoning due to a variety of
preservatives or in order to improve its taste or agents have occurred because butchers, dairies
appearance. and ice cream vendors have paid insufficient

PID14 5/21/04 11:41 AM Page 113

Control of infectious diseases Chapter 14

attention to hygiene when storing and selling their premises and equipment, and on facilities for
products. the storage and protection of food from

Preparation for consumption

Education of food handlers
In domestic households and in catering establish-
ments, poor technique, particularly in relation to However strict the law, the avoidance of food
avoiding contact between raw and cooked meats, poisoning depends heavily on those who prepare
inadequate thawing of frozen foods, insufficient it. They should understand the importance of
cooking and subsequent careful control of tem- such matters as personal and kitchen hygiene
perature during storage and serving, together with in the avoidance of contamination or cross-
inadequate attention to cleanliness of premises contamination of foods. They should also appreci-
and equipment, may lead to food poisoning, such ate the need, for example, to store food in
as that due to Clostridium perfringens, staphylococ- protected containers and to adequately defrost
cal toxins or Salmonella spp. frozen meat and poultry before cooking. The
dangers of incubating organisms, especially in
preprepared meat dishes, and the importance of re-
Prevention of food-borne disease
frigeration of foods liable to contamination in order
The prevention of food-borne disease depends on to reduce bacterial growth and of the separation of
correct action by many individuals in the complex raw meat from foods to be consumed without fur-
chain of production, manufacture and distribu- ther cooking must also be constantly stressed.
tion. The main ways in which the safety of food is
maintained and good hygienic practice is encour-
Roles of CCDC and EHO
aged are as follows.
Cases of suspected food poisoning should be noti-
fied to the Consultant in Communicable Disease
Quality of products
Control (CCDC) who are now employed by the
There are strict regulations relating to the quality Health Protection Agency (HPA). Their website is
and composition of some foods. This applies par- http://www.hpa.org.uk. The CCDC with the assis-
ticularly to milk and milk products, meat and meat tance of the EHOs employed by the Local Authori-
products, shellfish and the use of food additives by ty are responsible for the investigation of outbreaks
manufacturers. of food poisoning. Outbreaks and single cases of se-
rious infections, such as typhoid, call for immedi-
ate investigation and control measures. The results
Environmental conditions
may call for amendment of food production, stor-
Environmental health officers (EHOs) of local au- age or preparation practices in the establishments
thorities have extensive powers to inspect all food concerned to avoid the danger of further episodes.
premises and to sample foods. If necessary they can In some cases it may be necessary to invoke legal
prevent their sale. The Food and Drugs Act (1955) powers to require replacement of faulty equip-
and other relevant legislation laid down standards ment, cleaning and refurbishment, or even closure
on the construction and cleanliness of food of offending premises.

PID15 5/21/04 11:43 AM Page 114

Chapter 15

virus vaccine against yellow fever provided protec-

tion for troops serving in the tropics. Today, we
Historically, it was common knowledge that peo- have available a great array of vaccines and new or
ple who recovered from some infectious diseases, improved vaccines are constantly being devel-
such as smallpox, rarely contracted that disease oped. The introduction of comprehensive immu-
again. In 1796 Edward Jenner showed that a person nization programmes utilizing vaccines against
who had been deliberately infected with cowpox important diseases has done much to reduce
was subsequently protected against smallpox. This mortality and morbidity worldwide, particularly
led to the introduction of vaccination, one of the amongst infants and children.
first and most effective of all public health meas-
ures. The success of vaccination in eradicating
Passive immunization
smallpox from the UK and eventually from the
world is well known. Discoveries at the end of the Whilst most vaccines aim to induce lasting active
19th century concerning the pathogenicity of bac- immunity against specific infections, passive im-
teria led to the search for further vaccines. The iso- munization can also be used to give short-term
lation of anthrax by Koch in 1876 was quickly protection against a number of diseases. Passive
followed by Pasteur’s (Fig. 15.1) attempts to de- immunization is the donation to the host of spe-
velop attenuated strains that could be used to im- cific antibodies against a particular agent by the in-
munize animals and so protect them against jection of blood products derived from immune
the disease. Pasteur also developed an attenuated animals or humans. It is used to give a degree of im-
rabies virus that proved to be efficacious as a vac- mediate, though temporary, protection to non-
cine in humans. This was followed by other exper- immune individuals who have recently been ex-
iments, which showed that dead microbes, or their posed to a potentially dangerous infection. In such
suitably modified toxic products (toxoids), could circumstances, active immunization may be of lit-
also provoke an effective immune response. In tle benefit because of the delay between adminis-
1888, a diphtheria toxoid vaccine was developed. tration of vaccine and the production of antibodies
A successful vaccine against tuberculosis was not in protective amounts.
developed until 1921, an attenuated strain known Products used for passive immunization are im-
as the bacille of Calmette and Guérin (BCG). Dur- munoglobulins, which are now usually derived
ing the Second World War, tetanus toxoid vaccine from the blood of human donors. The historical
came into widespread use whilst an attenuated practice of using animal (usually horse) sera for

PID15 5/21/04 11:43 AM Page 115

Immunization Chapter 15

there are a limited number of individuals who can

donate their serum for the preparation of these
Passive immunity to common infections occurs
naturally through the transplacental transfer of
antibodies from mother to baby. Similarly, anti-
bodies are present in breast milk and give babies
some protection against relevant infections while
they are being breast-fed.

Active immunization
Active immunity to a disease is acquired naturally
after recovery from infection with the causal
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

PID15 5/21/04 11:43 AM Page 116

Chapter 15 Immunization

fluenzae type b (Hib) vaccine, which is prepared

Vaccine types
from purified capsular polysaccharide. Also acellu-
• Inactivated or killed vaccines lar pertussis vaccine is now used in preference to
• Live vaccines the killed vaccine.
• Toxoids
Vaccines vary in their antigenic potency, i.e.
• Component vaccines
their capacity to induce the formation of protec-
tive antibody. Much current work on vaccine de-
velopment is focusing on producing vaccines that
Inactivated vaccines
will produce a better immune response in a shorter
These are made from whole organisms, which are time. One way of doing this has been particularly
killed during manufacture. Examples include in- effective when producing vaccines for bacteria that
jected polio vaccine (IPV), typhoid, cholera and have a protective polysaccharide capsule. Tradi-
some pertussis vaccines. tional vaccines have used simple capsular poly-
saccharides, but these vaccines have not been
effective in infants, and have not provided long-
Live vaccines
term immunity. Attaching these polysaccharides
These are made from living organisms, which are to larger, more antigenic molecules to produce
either the organisms that cause the disease whose ‘conjugate’ vaccines may overcome these prob-
virulence has been reduced by attenuation (e.g. lems. Antigenic potency can sometimes also be en-
oral polio, measles, mumps and rubella vaccines) hanced by the use of adjuvants such as aluminium
or organisms of a species antigenically related to phosphate or aluminium hydroxide which are in-
the causal agent but which are naturally less viru- cluded in the pentavalent diphtheria, tetanus,
lent (e.g. smallpox (vaccinia) and tuberculosis acellular pertussis, Hib, IPV vaccine.
(BCG) vaccines). In susceptible (non-immune) in-
dividuals these attenuated organisms multiply in
Site of vaccinations
the body to many times the quantity given in the
original dose, but in an immune individual the The route of administration varies between vac-
virus is killed before it has a chance to replicate, so cines. Most are injected, whilst some are given
having little if any effect. This explains why it is orally. The site of the injection is important for
believed live virus vaccines — including measles, two reasons. Firstly, the antibody response varies
mumps, rubella and polio — can safely be repeated depending on whether the injection is given in-
in people who have been vaccinated previously. tramuscularly, subcutaneously or intradermally.
Secondly, the frequency of adverse effects varies
from site to site. Some vaccines, if given too deeply,
can cause severe reactions. For example, BCG
These are produced from bacterial toxins artifi- vaccine must always be given intradermally and
cially rendered harmless (e.g. diphtheria and should only be given by trained vaccinators. Live
tetanus toxoids). polio vaccine is given orally which has the advan-
tage of stimulating local immunity in the intestine
and inhibits later colonization (and transmission)
Component vaccines
of wild poliovirus. Most other vaccines are normal-
These contain one or more of the component anti- ly given by intramuscular or deep subcutaneous in-
gens of the target organism that are necessary jection. In infants, the recommended sites are the
to provoke an appropriate protective antibody re- anterolateral aspect of the thigh or upper arm. If
sponse. Examples of component vaccines, some- the buttock is used, the injection should be into
times called subunit vaccines, include influenza the upper outer quadrant to avoid the risk of
and hepatitis B virus vaccines and Haemophilus in- sciatic nerve damage.

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
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

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
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

Table 15.1 Schedule of routine childhood immunization in the UK.

Vaccine Dose Age

DTaP/Hib/IPV 1st 2 months

2nd 3 months
3rd 4 months

MMR 1st 12–24 months

2nd 4 years*

DTaP/APV Booster 5 years

BCG 1st 10–14 years (or may be given at birth)

Tetanus/IPV Booster 15–18 years (school leaving)

* 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.

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-
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
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

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
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
500 Notificati
Haemophilus influenzae type b (Hib)

Haemophilus influenzae is a common bacterium, 300 1– 4 years

which has a number of antigenic types. It is the H.
influenzae type b (Hib) which is the cause of nearly
100 < 1 year
all invasive and life-threatening infections, par-
ticularly in children under the age of 5 years. It is a 0
1989 1990 1991 1992 1993 1994
major cause of meningitis, with a case fatality rate Year (1989–94)
of around 5%, and also causes life-threatening
epiglottitis in young children. The Hib vaccine,
Figure 15.3 Notifications of Haemophilus influenzae type
first produced in the 1970s, contains purified cap- b vaccine (Hib), 1989–94. (Reproduced with permission of
sular polysaccharide conjugated to a protein. It was the OPCS (Crown copyright).)

Immunization 92%
introduced 81%
200 000
150 000 2500


100 000
30% 1500

50 000
500 Figure 15.2 Whooping cough notifi-
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

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

6 000 danger of accumulation of sufficient numbers of

4 000 susceptible older children to sustain an epidemic.
To reduce this risk a second dose is given with the
2 000
‘preschool booster’ vaccinations.
1931 ‘36 ‘41 ‘46 ‘51 ‘56 ‘61 ‘66 ‘71 ‘76 ‘81 ‘86 ‘91
Year (1931–91)
Figure 15.4 Polio notification in the UK showing the intro-
Mumps is a common but not normally serious
duction of injected polio vaccine and oral polio vaccine,
England and Wales, 1931–92. (Reproduced with permis- illness. However, complications including pan-
sion of the OPCS (Crown copyright).) creatitis, oophoritis or orchitis (leading on occa-

PID15 5/21/04 11:43 AM Page 122

Chapter 15 Immunization

600 000 Measles vaccine introduced 140

500 000 120

400 000 Notifications

300 000 MMR
200 000
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


40 Figure 15.6 Numbers of termina-
400 tions of pregnancies and births with
congenital rubella syndrome (CRS)
20 following the introduction of vaccine
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
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

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.

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
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,

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-
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%
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
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.

Vaccination for the traveller Hepatitis A

Overseas travellers are often exposed to infections This is probably the most common vaccine-
that they are unlikely to encounter at home. The preventable disease contracted by overseas tra-

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
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
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
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.

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.

PID16 5/21/04 11:44 AM Page 127

Chapter 16
Environmental health

tential as well as known risks. The design of a ra-

tional and effective programme to protect against
There has been increasing public concern about the infectious diseases or to reduce the harmful effects
effect that environmental changes might have on of environmental pollutants requires a clear under-
the health of the public. This has led to a renewed standing of the relationship between the agent, the
interest in the real and potential threats from both environment and humans in each particular in-
industrial processes and the pressures arising from stance. Account must be taken of the properties of
urbanization and population growth. Strategies for the agent that affect its ability to cause disease, the
the control and prevention of diseases caused by ways in which individuals and populations react to
noxious physical or biological agents are tradition- it and how the environment can affect the balance
ally based on action directed at containing or elimi- between the two directly and indirectly. It is often
nating the agent. In the UK this is the responsibility difficult to communicate the level of risk to the
of the Health Protection Agency (HPA). Their web- general public. The public wants to know not only
site is http://www.hpa.org.uk/. In some circum- the relative risk, but also their absolute risk of dis-
stances a change in behaviour of the general ease in order that they may make value judgements
population is required, for example by encouraging about various pollutants and other hazards.
people to use public transport to reduce pollution Pollution of the environment is increasingly
from traffic. This requires action by other agencies seen as not only producing physical disease, but
such as local authorities, or the Department of also having social and psychological conse-
Transport or Education. quences. Thus, although doctors are still con-
Adverse effects on health due to environmental cerned with agents such as microbes, chemicals
conditions can be acute or chronic. An example of and ionizing radiation, which cause physical dis-
an acute effect is an epidemic of respiratory disease ease, noise, for example, causing social disruption
brought about by a sudden increase in air pollu- and psychological stress, is of increasing impor-
tion, or poisoning due to a chemical spill. Exposure tance. Global issues such as the destruction of the
to radiation can have long-term effects, for exam- ozone layer and global warming are also attracting
ple cancer or fetal abnormalities. The long-term ef- increasing public concern and demand attention.
fects of adverse environmental influences are often
unknown at the time of the exposure. These are
The social environment
thus considered as potential or unproved risks.
Public health doctors have a duty to warn of po- In many respects, highly developed societies pro-

PID16 5/21/04 11:44 AM Page 128

Chapter 16 Environmental health

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-

PID16 5/21/04 11:44 AM Page 129

Environmental health Chapter 16

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
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

Deaths per day

of respiratory disease.
500 0.5 2

SO2 ppm
Acute health effects

Smoke (mg/m3)

A dramatic example of the acute effects of air pol- 250 0.25 1

lution was the infamous ‘smog’ in London in
December 1952 (Fig. 16.1) when it was estimated Smoke
that the fog was responsible for the deaths of 0 0 0
1st 8th 15th
3500–4000 people. This led directly to the passing December 1952
of the Clean Air Act (1956). This empowered local
authorities to establish smoke-control areas. As a Figure 16.1 Death and pollution levels in the London fog
result, air pollution by smoke declined rapidly in of December 1952. (From Reports of Public Health Medi-
the UK (Fig. 16.2). The benefit was seen when, in cine Subject 95. HMSO, London, 1954.)



Emission (1000 tons)

(mg m3)


100 (mg m3)

Emission (1000 tons)
Figure 16.2 Changes in the emission
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)

PID16 5/21/04 11:44 AM Page 130

Chapter 16 Environmental health

health in cities such as Shanghai, Los Angeles and

Water pollution
Mexico City.
The prevention of water-borne disease rests on the
purification and protection of supplies. Adequate
Long-term health effects
and safe water supplies are essential to health. To
The long-term damage to health created by air pol- be safe, drinking water must be free from contami-
lution is difficult to separate from the harmful ef- nation with both pathogenic microorganisms and
fects of other factors such as tobacco smoking, but harmful chemicals. The most serious infections
acute and chronic chest illnesses are more com- spread particularly by water are cholera, typhoid
mon in children and in older people living in areas and dysentery. These are due to the contamination
with persistently high levels of pollution. More re- of water supplies by human excreta. In countries
cently, the contribution of the burning of fossil with modern systems of sewage disposal and do-
fuels, especially in power stations, to the phenom- mestic water supply, spread by this route is ex-
enon of ‘acid rain’ with its destructive effects on tremely rare. Storage assists the purification of
the forests of central and northern Europe, has water by sedimentation of suspended matter
been highlighted. This and the damage to the and by biological action. It is further purified by
earth’s ozone layer caused by the use of chlorofluo- filtration through sand or chemical filters. Finally,
rocarbons as propellants in aerosols and as it is sterilized by chlorination, which oxidizes
coolants in refrigerators and freezers have become organic matter and kills any remaining micro-
matters of grave concern to ecologists. organisms. The dose of chlorine is controlled in
order to maintain a small residual amount of free
chlorine in the public supply. The water is then
Strategies for control
distributed through a closed system of pipes and
The Clean Air Act of 1956 has had a major impact service reservoirs. Its purity is monitored by regular
in the UK in reducing air pollution from the burn- sampling at various points in the distribution
ing of fossil fuels. Monitoring of the emissions system.
from power stations and industrial factories to en-
sure they comply with the law is the responsibility
Chemical pollution
of environmental health officers employed by
local authorities. The strategy to reduce lead in ex- Chemical pollution of water may arise from the
haust emissions from vehicles was initially encour- discharge of effluents from factories into rivers and
aged in the UK by the differential application of streams and also from the use of pesticides and
duty so that unleaded petrol was less heavily taxed fertilizers by farmers in water catchment areas. A
and therefore more attractive to car owners. The al- classic example of industrial pollution of water
ternative strategy to ban leaded fuel is now in occurred in Minimata Bay in Japan in the 1950s. In
place. Despite this, exhaust emissions continue to this instance, pollution with mercury led to con-
be a cause for concern. This has led the European tamination of sea water which entered the food
Union to require the fitting of catalytic converters chain through fish. The result was over 100 deaths
to all new cars, and vehicles with unacceptable ex- in humans, paralysis of many hundreds of others
haust emissions cannot be licensed. The removal and the deaths of thousands of domestic animals.
of chlorofluorocarbons from the atmosphere is Generally in the UK, monitoring by the water au-
being achieved by a number of voluntary agree- thorities prevents chemical pollutants reaching a
ments backed by the influence of powerful envi- level that is harmful. The protection of water sup-
ronmental groups such as Greenpeace and Friends plies is effected through legislation that prevents
of the Earth whose activities have encouraged indi- individuals and companies from polluting water
viduals to shun the use of aerosols and refrigerators sources through the discharge of industrial wastes.
which contain chlorofluorocarbons. This has been strengthened by European Union

PID16 5/21/04 11:44 AM Page 131

Environmental health Chapter 16

legislation. The prevention of run-off of nitrates,

Sewage treatment
fertilizers and pesticides from farmland is a prob-
lem which may require action. Problems have also In modern sewage treatment plants, after separa-
arisen in some special circumstances. For instance, tion of solids by filtering and sedimentation, the
the addition of alum to water supplies in order to liquid sewage is purified by biological oxidation.
make the water clearer can lead to problems for The final effluent, which is both clean and safe, is
people on renal dialysis. This is because the alu- usually discharged into rivers (often to be with-
minium salts become concentrated and can cause drawn further downstream for water supplies!).
encephalopathy in such patients. Unfortunately, some seaside towns still discharge
raw sewage into the sea, sometimes even above
low-tide level. This practice leads to offensive pol-
lution of beaches and under certain circumstances
Where the natural fluoride content of water is high may cause a hazard to bathers. Where there is no
the prevalence of dental caries is substantially less public sewage disposal system, for example in re-
than in low fluoride areas. Controlled experiments mote rural areas and on campsites, excreta are dis-
have shown that this natural benefit can be ob- posed of by using chemical toilets or septic tanks.
tained by artificial fluoridation of water supplies to
a level of 1 p.p.m. Maximum protection is achieved
Ionizing radiation
when fluoridated water is consumed throughout
the years of tooth development, and this benefit is Humans have evolved in an environment bathed
maintained into adult life. Objections have been in ionizing radiation. Today, most of the ionizing
raised to the practice of fluoridation of public water radiation to which a population is exposed still
supplies on the grounds that it is an invasion of in- comes from natural sources. Consequently, we are
dividual liberty and that it has potential dangers. unable to calculate the attributable risk associated
However, trials have failed to demonstrate any with exposure to low levels of ionizing radiation
harm when fluoride is added at the recommended from other sources. However, the ill effects of high
levels. Relatively few water authorities fluoridate doses of exposure are well known. This has led to
their supplies but the practice is now actively en- concerns about the safe levels for both individuals
couraged by the health departments in the UK. and populations. In addition, the potential risk to
Probably the most significant benefit to the popu- the public from nuclear war and industrial and mil-
lation from fluoride has been through the use of itary accidents has led to warnings from concerned
fluoride toothpaste. However for those underprivi- physicians. The nuclear accidents at Three Mile
leged children who are not encouraged to clean Island (USA) in 1963 and Chernobyl (Russia) in
their teeth, or whose mothers do not ingest extra 1987, as well as a number of accidents in nuclear
fluoride during pregnancy, the benefit is lost, and powered warships, clearly demonstrated that these
without fluoridated water supplies there is further fears are well founded.
disadvantage. Ionizing radiation can be in the form of X-rays,
gamma rays (electromagnetic radiation) or alpha
rays and beta rays (particle radiation). Over 85% of
Sewage and waste disposal
the radiation to which people are exposed in the
The provision of an efficient sewage and waste dis- UK comes from natural sources. Around 12%
posal system was probably the single most impor- comes from medical sources and around 1% from
tant public health measure taken in the 19th nuclear fallout and industrial processes. Individu-
century. Although this is now taken for granted, it als can be exposed to very different levels of radia-
remains central to the protection of food and water tion. Some occupational groups such as miners,
supplies, as well as to the maintenance of a clean nuclear industry workers and radiographers/
and safe environment. radiologists may be exposed to much higher

PID16 5/21/04 11:44 AM Page 132

Chapter 16 Environmental health

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.

PID17 5/21/04 11:45 AM Page 133

Chapter 17

In other diseases it may be possible to intervene

at an even earlier stage in their natural history by
Screening is the practice of investigating appar- treating precursor conditions, thereby reducing
ently healthy individuals with the object of detect- the risk that pathology will develop. For example,
ing unrecognized disease or its precursors so that there is evidence that the risk of stroke can be re-
measures can be taken that will prevent or delay duced by controlling blood pressure, and that the
the development of disease or improve the prog- risk of a woman developing invasive carcinoma of
nosis. The rationale behind use of the screening to the uterine cervix is reduced by the detection and
reduce morbidity and mortality is discussed below. treatment of carcinoma in situ. In some circum-
In many diseases, the pathological process is estab- stances it may be possible to identify individuals
lished long before the appearance of the symptoms who are particularly vulnerable to disease, even
and signs which alert people to the need to seek though as yet no abnormality exists. Active inter-
medical advice. By this time, the disease process vention at this stage may reduce subsequent risk.
and the consequent damage may be irreversible or For example, haemolytic disease of the newborn
difficult to treat. For example, in phenylketonuria can be prevented by the administration of anti-D
(an inborn error of metabolism) the abnormality antiserum to the rhesus-negative mother of a
does not usually declare itself before irreversible rhesus-positive fetus.
brain damage has occurred. This can be averted if Screening for genetic abnormalities is an impor-
the condition is detected in the neonatal period tant recent development. The purpose of this
and the affected infant is given a diet low in pheny- screening is to identify people who are apparently
lalanine. In other diseases, patients with signs of normal but at risk for having affected children, i.e.
disease, for example a woman with a lump in the gene carriers. The carrier individuals are then able
breast or a person with impaired vision, may fail to to make informed reproductive choices. The inci-
consult a doctor because the symptoms are not suf- dence of diseases such as Huntington’s chorea,
ficiently troublesome or because of fear or stoicism fragile X syndrome and cystic fibrosis may be con-
or for other reasons. It seems logical to believe that trolled in this way.
if potentially serious diseases are diagnosed and Another application of screening is to protect
treated at an early stage many personal disasters the public health. Some individuals may be in-
may be averted. If so, a programme aimed at their fected with an organism and, although they have
early detection would be a valuable preventive no symptoms, are capable of transmitting it to oth-
service. ers. Such individuals are called healthy carriers.

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’.

PID17 5/21/04 11:45 AM Page 135

Screening Chapter 17

Effectiveness In order to achieve their aim of reduc-

Selective screening
ing levels of morbidity and/or mortality from a
particular disease, screening programmes require a Tests are used to detect a specific disease, or
high uptake rate, especially amongst particularly predisposing condition, in people who are known
vulnerable groups. This is not always easy to to be at high risk of having, or of developing, the
achieve as has been found in cervical cytology condition.
screening where the most vulnerable groups —
social classes IV and V — have the poorest uptake.
Single disease screening
Examples Chest X-rays for evidence of pneumoco-
Screening programmes
niosis in coal miners; amniocentesis for detection
There are two approaches to population screening of chromosomal abnormalities in the fetus in older
programmes. One is to restrict screening to mem- women; retinopathy in people with diabetes.
bers of identifiable ‘high-risk’ groups in a popula-
tion (selective screening) and the other is to
Multiphasic screening
attempt to include everyone regardless of the de-
gree of risk (mass screening). Clearly, it is more Examples Antenatal examinations; pre-
economical to focus screening programmes on employment medical examinations in high-risk
high-risk groups. Efforts can then be concentrated occupations.
on securing high participation rates in order to
maximize the yield of cases in relation to the effort
Mass screening
and expense invested. Whole-population screen-
ing is indicated only where it is impossible to de- Large numbers of people are tested for the presence
fine high-risk groups with sufficient precision to of disease or a predisposing condition without spe-
ensure that they include a high proportion of those cific reference to their individual risk of having or
likely to develop the disease (sensitivity) and the developing the condition.
majority not likely to develop the disease is ex-
cluded (specificity). Even with so-called ‘mass
Single disease screening
screening’, the programme will normally be re-
stricted to certain broad categories determined, for Examples Tests for phenylketonuria and congeni-
example, by age, sex, occupation or area of resi- tal dislocation of hip in infancy; cervical cytology
dence. In both selective and mass screening, the for carcinoma in situ; mammography for breast
programme may be directed to the detection of a cancer.
specific disease, ‘single disease screening’, or in-
clude a range of tests for a number of different
Multiphasic screening
conditions, ‘multiphasic screening’.
Examples Biochemical profiles on hospital pa-
Types of screening tients; routine health ‘check-ups’ (well-woman
clinics, over 75 year olds in general practice, pre-
Selective screening — test for disease in high-risk group:
• single disease screening, e.g. chest X-rays for retirement groups, etc.).
• multiphasic screening, e.g. antenatal examinations
Mass screening — with no reference to risk:
Opportunistic screening
• single disease screening, e.g. cervical screening Some screening only occurs when the opportunity
• multiphasic screening, e.g. biochemical profiles on
arises, for example blood pressure screening for hy-
hospital patients
pertension in general practice, or cervical smears
Opportunistic screening — in general practice
on women using an oral contraceptive. This is of

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
(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-

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

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.

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
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.

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

and safety of air, water and food, the control of oc-

cupational and industrial hazards and environ-
Health services fall into two broad categories: mental pollution. In a complex industrial society,
• personal health services health may be affected by public policy in many
• public or environmental health services. fields that are not normally thought of as specifi-
cally ‘health’ services. For example, education,
transport, housing, industrial, commercial and
Personal health services
economic policies all directly or indirectly influ-
These include the whole range of preventive, treat- ence the health and welfare of society. One of the
ment and rehabilitative services provided for indi- functions of a public health service is to monitor
viduals. In the UK general practice is the main these factors and to provide scientific evidence of
source of primary, domiciliary and ambulatory their health implications. Public health doctors
care. Specialist services tend to be concentrated in have traditionally been involved in the develop-
general hospitals managed by NHS acute trusts. ment and provision of health promotion strategies
Long-term care is organized through primary care with the aim of altering these influences on health.
trusts, local authorities and private and voluntary Most influence at a national level is exerted by pub-
organizations. the preventive services are mainly lic health specialists in the Department of Health.
provided through primary care trusts and general More recently directors of public health have been
practice but additional facilities are available in appointed to the Government Office in each
some areas through a variety of agencies. English region. Public health doctors are also in-
volved in health care needs assessment and in
providing advice on the provision of health ser-
Public health services
vices as well as evaluating their effectiveness and
These are concerned with the control and preven- efficiency. This role is being developed particularly
tion of disease in the community, advice on public at the local level within strategic health authorities
policies for health promotion, assessment of the and primary care trusts.
health care needs of the population, and planning
and evaluation of health services. The control and
preventive function includes the monitoring of
disease and the control of factors in the environ-
ment that may affect health such as the quality

PID18 5/21/04 11:46 AM Page 144

Chapter 18 History and principles

volving people outside the family have had to

Provision of public health medical advice
develop. Such involvement made it necessary to
• Department of Health create a system of payment for care services
• Government Offices in each region and generated the need for professional carers. In
• Strategic health authorities
modern societies, this is organized and funded by
• Primary care trusts
the State.
The process whereby personal care evolved from
being solely a family obligation to being a profes-
The history and evolution of both personal and sional activity with state involvement in its financ-
public health services is described in this chapter. ing and supervision was complex. It was
In Chapter 19 the present arrangements for the de- influenced by the structure of societies, changes in
livery of health services in England and Wales are the expectations of individuals and developments
described. in medical science and technology. In England, the
earliest legislation for the public provision of ser-
vices for the sick was the Act for the Relief of the
History of personal health services
Poor (1598), usually referred to as the Poor Law,
A characteristic of human societies is that they ac- which required parishes to appoint an ‘overseer of
cept responsibility for the care of individuals who, the poor . . . to raise money by local taxation and
through no fault of their own, are unable to care to provide. . . the necessary relief for the lame, im-
and provide for themselves. In general, these are potent, old, blind and other such being poor and
the elderly, the poor and the disabled. The most not able to work.’ This legislation implicitly recog-
basic expression of this obligation used to be care nized the relationship between disablement and
through the extended family, i.e. parents, siblings, poverty and it restricted help to those who had no
children, uncles and aunts, together with others other source of support. In effect, it was a last-resort
who identify with the family. In most rural soci- provision. It was not fully repealed until the pas-
eties the main social unit is still the family. Here, sage of the NHS Act in 1946.
families normally live close together and share the Early arrangements for the care of the sick were
same type of work. In these circumstances the per- rudimentary. They were provided with shelter,
sonal caring aspect of the family’s life is absorbed food and basic care. The roles of the doctors and of
into its normal activities. In modern industrial so- medicine were limited principally to the care of the
cieties, the family is no longer always the main so- wealthy, except through the major charitable hos-
cial unit. The number of single-person households pitals. During the 20th century there was dramatic
is increasing. Couples are increasingly choosing progress in the development of medical skills and
not to have children or to delay having a family. of medical technology. These have affected the
People are more mobile, both geographically and shape of medical services in many ways. The new
socially, and work is normally a separate activity special skills and technologies had to be concen-
from day-to-day life. For this reason, even though trated in institutions (hospitals) in order that they
people may appreciate that they have a responsi- could develop. This brought about a change in the
bility to those of its members who are unable to nature of the hospital from an institution con-
care for themselves, they are often not in a position cerned with the general care of the poor to one that
to assist them. For example, different generations was clearly medically orientated. Many people
may live in different towns, or daytime jobs may who in previous generations had been looked after
preclude people from devoting sufficient time to at home, turned to hospitals for investigation, care
the care of an aged relative. Moreover women are and treatment. Consequently, the social mix of
increasingly becoming full-time workers and so do hospital patients changed and they ceased to be
not have the time to be unpaid carers as was the last refuge of the neglected, the destitute and
often the case in the past. Therefore, services in- those whose families had dispersed. This changed

PID18 5/21/04 11:46 AM Page 145

History and principles Chapter 18

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
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
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
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-

PID18 5/21/04 11:46 AM Page 146

Chapter 18 History and principles

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

PID18 5/21/04 11:46 AM Page 147

History and principles Chapter 18

many such hospitals, including nine, with accom-

General practice
modation for several thousand patients, around
Epsom in Surrey. The distance from London did Specialties developed in the hospitals because of
not deter the planners as they took it for granted the facilities offered there, while the doctors who
that once patients were admitted there was little worked mainly in the community became known
chance that they would ever be discharged. as general practitioners. The majority of the popu-
lation paid their general practitioner a fee for con-
sultation. In the growing conurbations, the fact
Hospitals that people could be seen free of charge in the out-
patient departments of the voluntary hospitals
• Voluntary
• Municipal caused some resentment among general practi-
• Cottage tioners. In order to overcome this, a system was de-
• Fever veloped whereby patients would only be seen in
• Lunatic asylums outpatients if referred by their regular doctor.

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.

PID18 5/21/04 11:46 AM Page 148

Chapter 18 History and principles

should enable individuals who become ill, or think

The NHS Act (1946)
they have a medical problem, to obtain advice,
This Act had three major effects: treatment or sometimes referral to a specialist ser-
• It ensured everyone had free access to a general vice. Primary care services are provided in general
practice, occupational health services, accident
• It brought the municipal and voluntary hospitals under
and emergency departments, first aid rooms and
the control of the (then) Ministry of Health
• It gave added responsibility to the Medical Officer of many other places. The precise location of primary
Health and local authorities in the running of commu- care facilities varies from country to country: in
nity services most societies there are many alternative sources of
such care. In England the primary care trusts have
been formed to improve the coordination of the
activities of general practitioners, district nurses,
The present tasks of personal
health visitors and other community staff.
medical services
The changes in medical practice during the past 50
Secondary care
years have been revolutionary. Today, access to
complex technology, skilled personnel and power- Secondary care is concerned with the provision of
ful therapies is taken for granted. Some illnesses specialist services, which are usually provided
that were inaccessible to medical intervention a within hospitals. Medical care that is dependent
generation ago, can be treated by methods that are upon expensive diagnostic and treatment technol-
now commonplace. In all branches of medicine, ogy is concentrated in hospitals in order to maxi-
however, there remains a need for the traditional mize the use of costly equipment and skilled
role of the doctor, that of an informed professional personnel. The task of specialist services is to diag-
carer. In some cases, medicine still has little to offer nose, to initiate treatment and, when the equip-
other than palliation and understanding; in oth- ment to treat is only available at the hospital, to
ers, once the correct diagnosis has been made the complete the course of treatment.
doctor’s role is simply one of supervising long-term
management. Fortunately, most of the population
Intermediate care
is fit and well for most of the time; they only re-
quire access to medicine when they become sick. Intermediate care has been developed to help re-
Broadly, the sick can be divided into those who re- duce the pressure on specialist services by provid-
quire access to the modern technology of medicine ing more specialized care in the community, to
for both the investigation and treatment of their help keep patients from being admitted to hospital
illnesses, and those for whom such facilities are less and to facilitate early discharge.
important than access to carers who have a thor-
ough understanding of them as people and the ef-
Tertiary care
fects the illness is having upon them. A modern
health service must provide facilities, sensitive to Increasingly, with further sophistication of medi-
individual need, that are accessible to everyone cal technology, a third or tertiary level of care has
who becomes sick and appropriate caring services evolved. This provides specialty services on a re-
for the chronically sick and disabled. gional or sometimes national basis and usually
only accepts referrals from another specialist.
Services that are considered tertiary specialties
Primary care
include neonatal intensive care, cardiac surgery,
Primary care services are required for the whole neurosurgery, renal medicine and oncology. These
population, and for most people they are the first are high-cost services that need to be used effi-
contact with the organized health services. They ciently and with discrimination.

PID18 5/21/04 11:46 AM Page 149

History and principles Chapter 18

general practice. In the absence of such resources,

Continuing care
many avoidable illnesses will occur to the disad-
Another type of care is required for the long-term vantage of the individuals and society as a whole.
sick and those who do not require the facilities of a The role of the general practitioner is very broad
high-technology hospital. Ideally, this should be but few of the activities are exclusive to him or her.
provided as close to the residence of the patient as The most expensive areas in the provision of
possible. Sometimes it is feasible to provide it at medical care are the acute hospitals. It is hard for
home. Much of the work of primary care trusts falls any society to achieve an ideal balance in its provi-
into this category and most of this type of care is sion of services and there will always be a need to
provided by general practitioners, supported by modify provision in the light of the circumstances
district nurses, community physiotherapists, of each community. In general, rich countries can
health visitors, etc. About 80% of the consultations afford the luxury of expensive technology but
(whether in the surgery or at the patient’s home) poorer and developing countries need to concen-
with general practitioners are generated by about trate their sparse resources on personal preventive
20% of the population. A large proportion of that services, primary care and secondary care that is
20% are the chronic sick who depend entirely on not dependent on expensive medical technology.
the general practitioner and his or her primary care
team for their medical care. There are other sources
Public health services
of continuing care. Long-term mental health ser-
vices can be provided in psychiatric units but are Until the early 19th century, there was little public
increasingly provided by community mental demand for state intervention in matters of health
health teams outside the hospital. Similarly insti- and welfare. During the 1820s and 1830s the so-
tutions for the care of people with learning disabil- called ‘sanitary reform movement’ began to gain
ities have been replaced by community residential momentum. It was particularly promoted by the
accommodation. Other facilities include nursing lawyer and philosopher Jeremy Bentham. He led
homes and hospitals for the care of older people, the push for reform and encouraged the notion
hospices for the dying, homes for the young that the State should bear some responsibility for
chronic sick and centres for those disabled by seri- the health of its people. His ideas were carried on
ous permanent injury or disease. In the past, the by his followers of whom the most notable was
decision to provide long-term care in a specialist Edwin Chadwick (Fig. 18.1). Chadwick produced a
institution rather than in the patient’s own home Report on the Sanitary Condition of the Labouring Pop-
was affected more by the social circumstances and ulation of Great Britain in 1842 which highlighted
the availability of the family and friends to provide the economic costs of an unhealthy workforce.
basic support than by the patient’s medical condi- This approach gained some support in Parliament
tion. Today, cost is also a consideration and if and led to the Nuisances Removal Act (1846)
people can be supported cost effectively in the which gave local authorities the power to clean up
community by professional carers then this option the towns though this was not a requirement on
is increasingly being pursued. the authority. In 1847, Liverpool appointed the
country’s first Medical Officer of Health, Dr W. H.
Duncan. The next year, the first Public Health Act
Preventive medicine
(1848) was passed in the wake of a disastrous out-
Personal health care services must include easy break of cholera. This Act which encouraged, but
access to preventive medicine (immunization, did not compel, local authorities to employ med-
screening, health education, family planning, ical officers of health also appointed the first na-
etc.). This is provided in a variety of ways including tional authority with a responsibility for health in
mother and child clinics, school clinics, well- England; the General Board of Health. Opposition
women clinics, occupational health centres and led to the disbanding of the Board in 1854. Despite

PID18 5/21/04 11:46 AM Page 150

Chapter 18 History and principles

then became evident, but the importance of isola-

tion and quarantine, as well as personal and
communal hygiene measures, in preventing the
transmission of infectious diseases was recognized.
The provision of vaccination against smallpox had
been a state responsibility since the Vaccination
Act (1853) but vaccination against other infectious
diseases made little impact until the mid-20th
When local authorities were established in their
modern form in the 19th century, one of their
principal roles was to administer environmental
health services. Over the years, they have acquired
a range of other functions, but their environmen-
tal health departments, staffed by environmental
health officers, continue to be the principal local
agencies responsible for monitoring and enforcing
many aspects of environmental standards, for ex-
ample food, water supplies and sewage disposal, air
quality, housing and working conditions other
than in factories (which are the responsibility of
the Health and Safety Executive). They also carry
Figure 18.1 Edwin Chadwick (1800–90), author of the Re- statutory responsibility for the investigation and
port on the Sanitary Condition of the Labouring Population control of communicable disease in the commu-
of Great Britain published in 1842. nity, obtaining medical advice for this and other
purposes from doctors (consultants in public
health medicine) employed by the corresponding
its demise, further public health legislation contin- health agencies. The Ministry of Health was creat-
ued to be passed including the 1871 Act (during a ed in 1919 to exert more effective control over
major smallpox epidemic), and the Public Health local bodies in the field of public health. The last
Act (1875) which obliged local authorities to im- important Public Health Act before the National
prove provisions for the disposal of sewage, to pro- Health Service Act was passed in 1936. It codified
vide pure water supplies and street cleaning, and to and simplified practice relating to environmental
improve housing standards and many other as- and personal hygiene. Thus, by 1946, public envi-
pects of urban life. The authorities were also then ronmental health practice and its administration
obliged to appoint medical officers of health to had evolved a structure close to its modern pattern
advise them on matters relating to the health of but it remained separate from provision of the
the community. Interestingly, occupational health treatment of the sick.
services were not included in this legislation and Recently, the term ‘The New Public Health’ has
remain outside the NHS to the present day. come into use (see Chapter 13). Two books were
In the 1870s the discovery of the nature of many published in the mid-1970s — The Medical Nemesis
infectious diseases began and public health en- by I. Illich and The Role of Medicine — Dream, Mirage
tered the so-called ‘Germ Theory Era’ in which mi- or Nemesis by T. McKeown — which challenged the
crobes were recognized as the causes of many of the importance of high-technology medicine in im-
most significant diseases of the time. This led to a proving the population’s health status. They pro-
more scientific approach to the control of infec- moted the view that improvements in nutrition,
tious diseases. The value of the sanitary reforms housing and wealth of the people were the most

PID18 5/21/04 11:46 AM Page 151

History and principles Chapter 18

significant factors in improving life expectancy intersectoral collaboration were instigated in a

and reducing morbidity over the last century. The number of European cities. In the UK, Liverpool
economic recession caused by the oil crisis of the became one of the first to embrace this concept
early 1970s led to a curtailing of spending on and soon over 120 cities worldwide were involved
health care. This and the escalating costs of mod- in implementing the Healthy City strategy. In the
ern medicine encouraged a search for alternative UK, in pursuit of the ‘Health for All’ strategy, the
ways to improve people’s health. The World Health DoH in 1991 published a document entitled The
Organization (WHO) began to expound the con- Health of the Nation which set out a number of goals
cepts of ‘Health for All’ based on preventive strate- for improving public health. This again empha-
gies and universal access to basic health services. A sized a commitment to the pursuit of health, as
WHO meeting held in Southern Russia in 1978 for- well as the provision of health care. The strategy
mulated the Alma Ata Declaration on Primary involved prioritizing objectives, setting targets and
Care. This Declaration proposed a number of monitoring and reviewing progress. This commit-
strategies for improving health, and emphasized ment to public health was confirmed by the in-
that primary care should be the main focus of coming Labour Government with its white paper
national health services in all countries. Our Healthier Nation published in 1998.
It was in the context of financial stringency and
growing appreciation of the influences of environ-
mental and economic and social factors on health Milestones in public health in the UK
that the UK Government commissioned a study 1842 Edwin Chadwick’s Report on the Sanitary Con-
led by Sir Douglas Black to examine inequalities in dition of the Labouring Population of Great
health. The Black Report entitled Social Inequality Britain
and Health was published in 1980 and despite ini- 1848 First Public Health Act and formation of the
tial political resistance became a major influence Board of Health
on public health doctors’ thinking about ways to 1853 Vaccination Act
1870s ‘Germ theory’ of disease gains momentum
improve health. At the same time, the WHO offi-
1875 Disraeli’s Public Health Act
cially adopted ‘Health for All by the Year 2000’ as 1919 Ministry of Health formed
policy: this included a commitment to the idea of 1946 NHS Act
equity in health (both within and between coun- 1980 Black Report on Social Inequality and Health
tries), a commitment to community consultation 1990 Health of the Nation published
and a greater emphasis on prevention and health 1998 Our Healthier Nation
promotion as strategies to improve health. The
‘Health for All’ strategy was adopted by the Euro-
pean Region of the WHO which modified and de-
Public health doctors
veloped appropriate health targets for Europe.
These targets included proposed changes to the In 1974, the medical officers of health and their
structure and process of health care as well as de- personal health service responsibilities were
tailing targets on specific health outcomes. In the brought into the NHS. At the same time they, to-
early 1980s, the city of Toronto took the WHO con- gether with doctors working in medical adminis-
cepts of ‘Health for All’ and, using community con- tration and in relevant university departments,
sultation, came up with a plan for a ‘Healthy City’. joined forces to form the new specialty of ‘commu-
This involved collaboration between the health nity medicine’. An erosion of the standards in
service agencies and the city authorities; it gen- some areas of traditional public health became ap-
erated a number of projects that aimed to improve parent in the early 1980s, illustrated by a number
the city environment and people’s health. The of serious outbreaks of infectious diseases. This led
European Region of the WHO quickly adopted the the Government to set up an enquiry into the pub-
idea of Healthy Cities and trials of the strategy of lic health function. The subsequent report, Public

PID18 5/21/04 11:46 AM Page 152

Chapter 18 History and principles

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
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
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.

PID19 5/21/04 11:47 AM Page 153

Chapter 19
The National Health Service

nation of the long-standing pool of untreated

chronic illness, the cost of the proposed National
The important report by Sir William Beveridge (Fig. Health Service would fall. They failed to anticipate
19.1) on social and allied services, published in the possibility of changes in the expectations
1942, was the culmination of many years of pres- of the public, the consequences of an ageing
sure for social reform. It recommended that the population and increases in the costs of medical
State should finance and provide a comprehensive technology.
social security system and that it should be under- The wartime coalition government accepted
pinned by a comprehensive national health ser- Beveridge’s proposals for comprehensive national
vice. It specifically proposed that a national health social security and health care systems but was un-
service should provide the facilities which would: able to implement them immediately. It charged
‘. . . ensure that for every citizen there is available the Minister of Health for England and Wales and
whatever medical treatment he requires in whate- the Secretary of State for Scotland with the respon-
ver form he requires it, domiciliary or institutional, sibility of initiating consultations with representa-
general, specialist or consultant, and will ensure tives of the medical profession, the voluntary
also the provision of dental, ophthalmic and surgi- hospitals and the local authorities. Discussions
cal appliances, nursing and midwifery, and reha- began early in 1943 and on 8 February 1944 a
bilitation after accidents’. White Paper on The National Health Service was
Apart from humanitarian considerations, which published. Its stated objective was: ‘to show what
were the principal motivations for the proposals, is meant by a comprehensive service and how it fits
Beveridge made the apparently logical assertion with what has been done in the past, or is being
that such a health service would reduce the costs of done in the present, and so help people to look at
social security payments by decreasing the amount the matter for themselves’. The publication of the
of illness in the population. It was thought that White Paper served to crystallize ideas and to stim-
this would increase the general appeal of the pro- ulate criticism. By the end of 1944 the Minister
posals. When estimating the possible costs of a of Health submitted the suggestions that he had
health service, Beveridge and his colleagues made a received from all interested parties to the
further, and, as it turned out, disastrously naive as- Government. Basically, the proposals involved the
sumption. They assumed that, as the health of the Government taking financial and other responsi-
population improved because of the abolition of bilities for the municipal, voluntary and other hos-
poverty, better preventive medicine and the elimi- pitals, for the general practitioner services as set up

PID19 5/21/04 11:47 AM Page 154

Chapter 19 The National Health Service

administered by a hospital management commit-

tee. The teaching hospitals (both undergraduate
and postgraduate) were autonomous from the re-
gional hospital boards. Each had their own board
of governors, which worked in close cooperation
with the governing body of the associated univer-
sity institution.

General practitioner services

The administration of general medical services was
the responsibility of 134 executive councils. They
• general medical services (family doctors)
• general dental services
• pharmaceutical services
• ophthalmic services.
All of the above services were provided on an inde-
pendent contractual basis. This means that the
general practitioners, dentists, opticians, oph-
Figure 19.1 Sir William Beveridge, author of the report on thalmic medical practitioners and pharmacists are
Social and Allied Services, published in 1942. not employed by the NHS; they are paid by the
NHS for the services they provide. The executive
under the 1911 Act, and for municipal public
councils had limited disciplinary and planning
health services and other aspects of personal and
functions; their main role was that of a paying
preventive medical services. Access to all services
agency. Technically, general practitioners and
was to be without direct charges at the time of use
other independent contractors were directly ac-
for all residents of the country. In essence, the
countable to the Minister. In effect, the contra-
availability of the then existing services was to be
ctual position of this group of practitioners was
extended, but their basic philosophy and adminis-
little different to that under the 1911 legislation
tration changed little. On 19 March 1946, a Bill
except that the services were now available free to
providing for the establishment of a comprehen-
all citizens.
sive health service was presented to Parliament.
Royal Assent was given to the National Health Ser-
vice Act on 6 November 1946 and the Service was Local authority services
launched in July 1948.
The local government authorities were responsible
for the care and aftercare of patients in the com-
The National Health Service, 1948
munity and for the prevention of disease. Specifi-
The administration of the original health service was di- cally, they were responsible for:
vided into:
• antenatal care
• Hospital services
• midwifery
• General practitioner services
• Local authority services • infant and child welfare
• district (domiciliary) nursing
• health visiting
Hospital services
• school health services
Fourteen regional hospital boards were established • immunization
within which there were 290 hospital groups each • ambulance services

PID19 5/21/04 11:47 AM Page 155

The National Health Service Chapter 19

• 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

PID19 5/21/04 11:47 AM Page 156

Chapter 19 The National Health Service

including health and education whilst at the same

Secretary of State for Health
time introducing the principles of the market place
into these services. In the late 1980s, the financial
restrictions placed on institutions by health autho- Department of Health

rities led some hospitals to seek an alternative

funding structure. The idea of independent hospi- Special Health Authorities
Strategic Health Authorities
tal trusts was born. A bill was passed to allow the (e.g. NICE, HPA, HDA)
creation of NHS trusts, which would be funded
directly from the Department of Health (DoH). The NHS Trusts Primary Care Trusts
move to trust status was to be voluntary and ini-
tially most hospitals chose to remain as directly Independent contractors
(GPs, dentists, etc.)
managed units of DHAs. Another key reform en-
abled general practitioners to hold funds on behalf
of their patients including budgets for medicines Figure 19.2 Organization of health services in England,
and some secondary services such as non-urgent
surgical and medical services, for example or- their geographical areas and that, within general
thopaedics, dermatology, etc. These general practi- strategies and financial limits, they have consider-
tioners could then choose whether to use their able autonomy to allow them to respond to local
funds to purchase services from the local hospital, needs. The general strategies and financial alloca-
from a trust or from the private sector. Choices tions are decided by the Government and the NHS
could be made on the basis of quality and avail- Management Board. The StHAs’ role is to monitor
ability of service or on price or a combination of the PCTs and to ensure that patients are receiving
the two. There was an expectation that the intro- an equitable and high-quality service.
duction of competition into the NHS would con- The Secretary of State for Health is responsible to
trol costs and improve quality through the Parliament for the NHS and as a member of the Cab-
pressure of an artificial market. At the same time, inet is able to bring the needs of the Service to the at-
the contract between general practitioners and tention of the Government and to argue the case for
hospitals became more explicit. In many ways funds. The Secretary of State is also responsible for
these ideas brought about the planned changes. the enactment of government policy on health mat-
However not all general practitioners took up the ters, accounting to Parliament for the expenditure
challenge of fundholding and after a time there ap- of the Service and for its performance. The Secretary
peared in some areas a two-tier system where pa- of State is responsible to Parliament for:
tients registered with a fundholding practice • promoting and protecting the health of the
received different services than those in non- nation;
fundholding practices. In 1997 the new Labour • providing a national health service in England;
Government reformed the health services yet • social care including oversight of personal social
again, leading in 2002 to the abolishment of RHAs services provided by local authorities in England.
and DHAs which were replaced by 28 Strategic The DoH in England provides administrative
Health Authorities (StHAs) and 309 Primary Care support to the Secretary of State and the depart-
Trusts (PCTs) (Fig. 19.2). mental ministers. Its main functions are to assist
them by:
• supplying the information they need regarding
The present management
the working of the service;
• advising them on the choices available when
The principles that govern the management of the making policy decisions;
service are that the PCTs have responsibility for • advising on the possible consequences of the
purchasing health services for the population of available options;

PID19 5/21/04 11:47 AM Page 157

The National Health Service Chapter 19

• 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:
• 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

PID19 5/21/04 11:47 AM Page 158

Chapter 19 The National Health Service

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
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
services for the elderly, some mental health
services and services for those with learning
Commission for Health disabilities.
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

PID19 5/21/04 11:47 AM Page 159

The National Health Service Chapter 19

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
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.

PID19 5/21/04 11:47 AM Page 160

Chapter 19 The National Health Service

duced health care systems that are funded either by

Planning health services the State or through local authorities. These either
underwrite high-risk individuals or offer state-
Objectives supervised and subsidized insurance. In all of these
schemes, the State bears all or part of the cost from
The health service has no single and easily defin-
general taxation. In most, the user of services pays
able objective. Various facilities are provided, in-
all or part of the cost and then reclaims a propor-
cluding specialist services for the acute sick,
tion from the insurance fund. This is said to be ad-
preventive services, primary care and care for the
vantageous because it makes people aware of the
chronically sick and disabled. Most of the work
true cost of medical care. This type of system oper-
involving direct intervention in acute sickness is
ates in most western European countries other
purely medical in content, i.e. it is mainly depend-
than the UK.
ent upon the technical skills of doctors, supported
The British system is unusual in being funded
by other highly trained staff. The care of the
from a combination of direct taxation and
chronically sick requires the skill mainly of other
National Insurance contributions to which are
professionals, such as nurses, physiotherapists and
added a range of charges including prescription and
social workers. Many preventive programmes re-
dental charges. There are exemptions from charges
quire action by non-medical professionals, for ex-
for children, the elderly, the unemployed, pregnant
ample teachers and engineers. Health care planning
women and people with certain chronic illnesses.
is necessary in order to match needs, demands and
Medical and other professional practitioners retain
available resources within this complex system.
a large measure of independence, and responsible
use of the service by the public is encouraged by the
various charges made to patients.
Whatever the source of finance there is a limit to
• Financial the amount of money that individuals or govern-
• Staff ments can spend on health. It follows that if a high
• Facilities proportion of the available money is spent on one
type of service, for example acute services, less is
available for other important aspects of care, for
example care of the chronically sick. In a state sys-
Financial resources
tem, decisions about how the available money
The cost of modern medicine is now such that few should be spent are political, and they must re-
people can afford to budget for it out of their in- main so, because the politicians are ultimately ac-
come. In most countries where a state-funded sys- countable for all public expenditure. In privately
tem of care does not exist, many people insure financed systems, the balance is determined by the
themselves against medical expenses. The diffi- amount of money each individual has and is will-
culty about this is that the risks of long-term ill- ing to spend. As the working population has the
nesses are difficult for a commercial company to greatest spending power, this usually results in a
underwrite. Even if this were possible, high premi- growth of acute services to the detriment of servic-
ums would have to be imposed. Chronic ill health es for the elderly, the mentally ill and the disabled.
affects the individual’s earning capacity and its
prevalence increases with age. Thus, the most vul-
Staffing resources
nerable members of the community are the least
able to maintain payment of premiums. In order to The second constraint on health service planning
overcome some of the obvious dangers of making is the numbers and types of trained personnel that
each individual responsible for his or her own are available. The principal groups involved are
medical expenses, many countries have intro- doctors, nurses, therapists and technicians. Ab-

PID19 5/21/04 11:47 AM Page 161

The National Health Service Chapter 19

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
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.

PID20 5/21/04 11:48 AM Page 162

Chapter 20
Health targets

or increasing the number of people screened and

treated for preventable diseases. Structural targets
During the past 25 years there has been a shift in are aimed at improving the health services in areas
emphasis from simply providing access to health where there is good evidence that an intervention
care to one of attempting to improve the health of will reduce disease such as increasing the number
the population. This change has been promoted by of chest pain clinics or breast screening services.
the World Health Organization (WHO), which in
1975 outlined the concept of ‘Health for All by the
Year 2000’. ‘Health for All’ was officially adopted as WHO targets
WHO policy in 1981. It advocated that the pursuit The 38 targets of ‘Health for All’ were divided into the
of health, rather than only the cure of disease, following subsets.
should be the aim of health policy makers. In addi- Outcome targets for improvements in health, for
tion, it was suggested that policy makers should example:
strive for more equitable health status both within • eliminating preventable diseases, e.g. measles
and between countries. The WHO suggested these • reducing mortality from diseases, e.g. heart disease
and stroke
goals could best be achieved through promotion of
Process targets for activities needed to make these im-
healthy lifestyles, the elimination of preventable provements, for example:
diseases and the provision of comprehensive • policies to reduce smoking
health coverage based on primary health care. • introducing population-based disease screening
‘Health for All’ suggested that countries should de- Structural targets designed to improve health services
velop health targets that could be monitored to en- management/organization, quality of care, staff train-
sure that the strategies of improved health status ing, etc.

and equity were being achieved. The European

Regional Office of the WHO suggested 38 targets
to assist member states in setting their own targets. Many countries worldwide have adopted the
These targets were divided into outcome, process idea of setting national targets aimed at improving
and structural targets. Outcome targets are health the health of the people. The number and types of
targets that aim to reduce the incidence of particu- target have varied widely depending on local needs
lar conditions or deaths from certain diseases. and the available resources. In July 1992 the UK
Process targets are aimed at reducing the activities Government published its strategy for improving
that cause disease such as reducing smoking rates health in England as a paper entitled Health of the

PID20 5/21/04 11:48 AM Page 163

Health targets Chapter 20

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

Cancers Known causes

Cancer still affects almost every family in Britain at Ninety per cent of lung cancer deaths are asso-
some time. Around two in five people develop can- ciated with smoking. The differential trends in
cer during their lifetime, and one in four people die lung cancer deaths between men and women cor-
from it. Our Healthier Nation has stated that the aim relate with their changing patterns of smoking. It is
is to reduce the death rate from cancer in people estimated that non-smokers who are regularly ex-
under 75 years by at least a fifth by 2010, poten- posed to tobacco smoke carry an increased risk of
tially saving up to 100 000 lives in total. Cancers 10–30%. The smoking habits of the different social
account for around 25% of the deaths in the UK. classes is changing, resulting in growing inequity
There are many types of cancer and the causes of in health status amongst the lower classes due to
each differ. Health targets concentrate on the most the effects of cigarette smoking (Fig. 20.2). Other
common and on those whose prevention would causes of lung cancer include exposure to radon
add the most years to life. Our Healthier Nation has gas, which occurs naturally in some parts of the


4000 Female
Death rate




Figure 20.1 Age-specific death rates

for men and women due to malignant 25–34 25–34 35–44 45–54 55–64 65–74 75–84 85+
neoplasm of the trachea, bronchus Age (years)
and lung (England and Wales, 1998).

PID20 5/21/04 11:48 AM Page 164

Chapter 20 Health targets

Social class

Manual skilled
Partly skilled

All social classes

0 10 20 30 40 50 60 70 80 90
European age-standardized* rate per 100 000 population Figure 20.2 Mortality from lung can-
* Adjusted for differences in age structure of the population cer by social class in men aged 20–64
in England and Wales, 1991–3.


Death rates per 100 000 in 1998




















Figure 20.3 Age-specific death rates













Age per 100 000 for carcinoma of female

breast (England and Wales, 1998).

UK, and exposure to certain chemicals, e.g. as- Breast cancer

bestos, arsenic, mustard gas.

Frequency and trends

Breast cancer is the most common cancer for
Targets to reduce deaths from lung cancer are women in the UK with 41 000 new cases each year.
mainly aimed at reducing smoking. Strategies that It causes more deaths in women than any other
have been suggested include advice from general cancer, about 13 000 per year. About a third are
practitioners for patients to give up, wider promo- in women under the age of 65 years. The death
tion of the dangers of smoking and the use of adju- rate increases with age (Fig. 20.3). Breast cancer
vant therapy such as nicotine replacement rates in the USA are higher than those in the UK
therapy. There has also been an initiative to im- and the rates in Asian and Hispanic countries are
prove access to specialist services to try and reduce lower.
the case fatality rate by diagnosing and treating pa-
tients earlier.

PID20 5/21/04 11:48 AM Page 165

Health targets Chapter 20

and 16 000 deaths each year. More than 90% of

Known causes
cases occur in people over the age of 50 years.
Migrant studies have shown that environmental There is a steady increase in the death rate up to the
and lifestyle factors are important in the aetiology age of 80 years with an excess of deaths in these age
of breast cancer. Many risk factors for breast cancer groups in men compared with women (Fig. 20.4).
have been identified. It is one of the few cancers
where the risk is greater in women from a higher
Known causes
social class. Other risk factors include late age at
first pregnancy, late age at menopause and parity Risk factors for colorectal cancer include a diet
(with nulliparous women being at increased risk). high in red meat and fat and low in vegetables and
Women who have breast fed their babies are at fibre. Exercise seems to be protective. People with a
reduced risk. history of ulcerative colitis and familial polyps are
at increased risk.

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.

Colorectal cancer Testicular cancer

Frequency and trends Frequency and trends

Colorectal cancer is the second most common Each year, there are over 1600 new cases of testicu-
cause of cancer in the UK with 35 000 new cases lar cancer and it is the most common cancer to


Death rates per 100 000





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).

PID20 5/21/04 11:48 AM Page 166

Chapter 20 Health targets

affect men aged 15–49 years of age in the UK. It re-

sponds particularly well to treatment. Over 9 in 10
Cervical cancer
patients will be cured and fewer than 100 men a
year die from testicular cancer. Frequency and trends
The importance of cervical cancer as a health target
Known causes relates to the fact that it is the second most com-
mon cancer in middle-aged women (after breast
Risk factors include cryptorchidism (an unde-
cancer), and potentially the outcome can be modi-
scended testicle), having a close relative who has
fied by comprehensive screening and effective
had testicular cancer, infection with the human
treatment of precancerous conditions. There are
immunodeficiency virus and having Klinefelter’s
around 3200 new cases and 1100 deaths each year.
syndrome. Testicular cancer is most common in af-
The annual number of deaths from cervical cancer
fluent Caucasians and with the exception of New
has fallen steadily since the 1950s and is now
Zealand Maoris, the disease is rare in non-
falling by 7% per annum (Fig. 20.5).
Caucasian populations.

Strategies Known causes

A simple, regular self-check can help detect the Cervical cancer is more common in women who
early signs of the disease. Although most lumps become sexually active at a young age, in those
found by self-examination are benign they should with multiple sexual partners and in those whose
be reported to a doctor as soon as possible. Infor- regular sexual partner has had multiple partners. It
mation leaflets about testicular cancer entitled A has been shown to be associated with the sexual
Whole New Ball Game are available from the rel- transmission of human papillomavirus, particu-
evant charities such as Cancer Research UK (website larly types 16 and 18. There are positive associa-
http://www.cancerresearchuk.org/). tions with child bearing, with an increased risk for


2500 Number



1953 1963 1973 1983 1993 Figure 20.5 Number of deaths from
Year carcinoma of the uterine cervix in
England and Wales, 1953–98.

PID20 5/21/04 11:48 AM Page 167

Health targets Chapter 20

those who start having children at a young age,

Known causes
and the risk increases with increasing parity. There
has also been shown to be an association with Both malignant melanoma and squamous cell car-
smoking. Barrier methods of contraception reduce cinoma are associated with excess exposure to ul-
the risk of cervical cancer. traviolet radiation and are thus potentially
preventable. The increase is thought to be related
to exposure to sunshine. The hypothesis that
malignant melanoma is associated with exposure
The death rate has fallen because of better treat- to sun is supported by the fact that it is more
ment and because of the introduction of the cervi- common in the higher social classes and where
cal screening programme, which offers screening Anglo-Saxon populations are resident in tropical
to all women aged 20–64 years of age. regions such as Queensland (Australia) and parts of
the USA. Young children who have been sunburnt
have an increased risk of malignant melanoma
Skin cancer
later in life.

Frequency and trends

There are three main types of skin cancer: basal cell
The Government has spent 10 years raising
cancers, squamous cell cancers and malignant
awareness about ultraviolet light and the risk
melanomas. Although melanoma is a relatively
of skin cancer. It continues to promote avoidance,
rare skin cancer, it is important because it is the
the use of sun blocks and targeting campaigns
most likely to metastasize and is the most likely to
at high-risk groups. However, a problem with
cause premature death. There are around 2000
a primary prevention programme to reduce
deaths per year in the UK due to skin cancer, of
the incidence of a cancer is that there is likely
which 75% are due to malignant melanoma. Death
to be a long lag time between the initiation of
from malignant melanoma is rising by about 5%
the programme and changes in incidence or
per annum (Fig. 20.6). Non-melanotic skin cancers
affect over 30 000 people a year in the UK and cause
about 500 deaths.


Death rates per million



Figure 20.6 Age-standardized death

rate per million for men and women 1971 1981 1991 2001
from malignant melanoma in England
and Wales, 1974–2002.

PID20 5/21/04 11:48 AM Page 168

Chapter 20 Health targets

data usually exclude deaths in people over the age

Heart disease and stroke of 75 years.
There has been a steady fall in death rates from
Frequency heart disease in adults in England since the 1970s
and similar reductions have been seen in stroke
Coronary heart disease (CHD) is due to atheroma
(Fig. 20.8). Despite these reductions, comparisons
of the coronary arteries and is the largest single
with countries such as France and Japan (which
cause of death amongst men and women in the
have much lower recorded death rates from CHD
UK. CHD and stroke together account for 65 000
(Fig. 20.9)) encourage the belief that further sub-
deaths per year in people under 75 years of age. In
stantial reductions in the incidence of CHD are
those aged under 65 years it accounts for one-third
of deaths in men and one-fifth in women. The
incidence of CHD increases with age, and is greater
Known causes
in men than in women (Fig. 20.7). Often in
older age groups, the certified cause of death can be The risk of CHD is increased significantly in rela-
arbitrary. Consequently, comparative mortality tion to three key risk factors: smoking, hyperten-

25 000

20 000
Death rates per million

15 000 Male
10 000

5 000

0 Figure 20.7 Age-specific death rates

25–34 35–44 45–54 55–64 65–74 75–84 per million for men and women due to
Age coronary heart disease (England and
Wales, 1998).

Death rate per 100 000 population


Rate of decline
slowing down



Figure 20.8 Death rate per 100 000
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.)

PID20 5/21/04 11:48 AM Page 169

Health targets Chapter 20

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
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
3.5 increasing the number of smokers provided with
Relative risk

3.0 free nicotine replacement therapy and other ad-

2.5 junctive therapies. Dietary interventions include
2.0 promoting healthy eating to try and reduce the
1.0 Risk
number of people who are overweight and obese.
076 There are also targets promoting physical activity.
84 91 98 105
Diastolic BP General practitioners and primary care teams
should identify all people with established cardio-
Figure 20.10 Relation between diastolic blood pressure vascular disease and offer them comprehensive ad-
and stroke from meta-analysis of 420 000 patients followed vice and appropriate treatment to reduce their
for 10 years. (MacMahon S. Lancet 1990). risks. Medical interventions include advice and
treatment to maintain blood pressure below
140/85 mmHg, giving low-dose aspirin (75 mg
sion and cholesterol levels. Around 30% of the
daily) to high-risk patients, prescribing statins and
adult population smoke. Studies have shown that
dietary advice to lower serum cholesterol concen-
the relative risk of death from CHD and stroke is in-
trations, and prescribing warfarin or aspirin for
creased in smokers compared to non-smokers and
people over 60 years old who also have atrial
increases with the number of cigarettes smoked.
Similarly, there is a positive correlation between
blood pressure and the risk of CHD and stroke (Fig.
20.10). There is also a direct relationship between Mental health
serum cholesterol and CHD. Related risk factors are
obesity, lack of exercise and stress. Effective inter-
ventions are available for some of these risk factors,
either through appropriate therapeutic interven- Deaths attributed to mental illness are principally
tion or through lifestyle changes. Genetic make- due to dementia and suicide. There are about 4000
up, as indicated by a family history of CHD, and suicides a year, the majority being in people under
gender are two risk factors that cannot be the age of 65 years. The incidence of suicide has
modified. fallen in women since 1980 whilst the rate in men

PID20 5/21/04 11:48 AM Page 170

Chapter 20 Health targets

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
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
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-

160 Male
Death rates per million