6316.pdf
Media
- extracted text
-
I
MANUAL
OF EPIDEMIOLOGY
FOR DISTRICT HEALTH
MANAGEMENT
Edited by
J.P. Vaughan
London School of Hygiene and Tropical Medicine
London, England
R.H. Morrow
UNDP/World Bank/WHO Special Programme
for Research and Training in Tropical Diseases
World Health Organization
Geneva, Switzerland
World Health Organization
Geneva
1989
(>
l
9^
( X- \ 'l C'Cl
*
06316
The World Health Organization is a specialized agency of the United
Nations with primary responsibility for international health matters and
public health. Through this organization, which was created in 1948, the
health professions of some 165 countries exchange their knowledge and
experience with the aim of making possible the attainment by all citizens of
the world by the year 2000 of a level of health that will permit them to lead
a socially and economically productive life.
By means of direct technical cooperation with its Member Stares, and
by stimulating such cooperation among them, WHO promotes the develop
ment of comprehensive health services, the prevention and control of
diseases, the improvement of environmental conditions, the development
of health manpower, the coordination and development of biomedical
and health services research, and the planning and implementation
of health programmes.
These broad fields of endeavour encompass a wide variety of activities,
such as developing systems of primary health care that reach the whole
population of Member countries; promoting the health of mothers and
children,- combating malnutrition; controlling malaria and other communi
cable diseases including tuberculosis and leprosy,- having achieved the
eradication of smallpox, promoting mass immunization against a number of
other preventable diseases; improving mental health; providing safe water
supplies; and training health personnel of all categories.
Progress towards better health throughout the world also demands
international cooperation in such matters as establishing international
standards for biological substances, pesticides and pharmaceuticals;
formulating environmental health criteria; recommending international
nonpropnetary names for drugs,- administering the International Health
Regulations; revising the International Classification of Diseases,
Imunes, and Causes of Death; and collecting and disseminating health
statistical information.
Further information on many aspects of WHO's work is presented in
the Organization's publications.
Community Health Cell
Library and Documentation Unit
367, "Srinivasa Nilaya"
Jakkasandra 1st Main,
1st Block, Koramangala,
BANGALORE-560 034.
Phone : 5531518
II
Manual of Epidemiology for District Health Management
WHO Library Cataloguing in Publication Data
Manual of epidemiology for district health management.
1. Epidemiologic methods 2. Epidemiology 3. Community health services—
organization &. admimstration
I. Vaughan, J.P. H. Morrow, R.H.
(NLM Classification: WA 950)
ISBN 92 4 154404 X
© World Health Organization 1989
I
I
Publications of the World Health Organization enjoy copyright
protection in accordance with the provisions of Protocol 2 of the Universal
Copyright Convention. For rights of reproduction or translation of WHO
publications, in part or in toto. application should be made to the Office of
Publications, World Health Organization, Geneva, Switzerland. The World
Health Organization welcomes such applications.
The designations employed and the presentation of the material in this
publication do not imply the expression of any opinion whatsoever on the
part of the Secretariat of the World Health Organization concerning the
legal status of any country, territory, city or area or of its authorities, or
concerning the delimitation of its frontiers or boundaries.
The mention of specific companies or of certain manufacturers7 products
does not imply that they are endorsed or recommended by the World Health
Organization in preference to others of a similar nature that are not men
tioned. Errors and omissions excepted, the names of proprietary products are
distinguished by initial capital letters.
The editors alone are responsible for the views expressed in this
publication.
Printed in England
89/7940—J.B. Offset—6000
I
Ill
Contents
Tv
I<
LIBRARY
)c
I
A?’
Foreword—A.O. Lucas i
Preface and acknowledgements
Vll
1
District Health Management
2
Epidemiological Principles
3
District Population 21
4
Epidemiological Health Information 33
5
Reporting and Surveillance Systems 45
6
Controlling an Epidemic 59
7
Epidemiological Surveys 71
8
Organizing Investigations and Surveys
9
Record Forms and Coding 93
10
Data Processing and Analysis
99
11
Presenting Health Information
113
12
Communicating Health Information
13
Epidemiology and District Health Planning
14
A B C of Definitions and Terms
1
9
87
125
131
155
Appendices
1
Ethical guidelines for epidemiological investigations
2
Estimating sample size for a prevalence study
3
Using random numbers 177
4
Organizing an epidemiological survey
5
Screening and diagnostic tests 189
6
Age standardization 193 .
Index 195
179
175
169
V
Foreword
In clinical medicine, diagnosis is the basis for effective manage
ment of the patient. On first contact, the clinician asks about the
patient's symptoms, conducts a physical examination and carries out
relevant laboratory and other special investigations. On the basis of
this initial assessment, treatment is instituted, and the cycle re
peated thereafter to monitor the patient's progress and td guide
future interventions. Diagnosis, as the foundation and the main
pillar of clinical medicine, is a familiar and well accepted notion.
Diagnosis is equally important in public health. Like the clinician, the public health practitioner must establish a diagnosis as the
basis for effective action. As the clinician momtors the course of the
illness in the patient, so the public health worker must continually
assess progress within the community. The most powerful tool at
the disposal of public health workers for diagnosis and monitoring of
community health is epidemiology. Used skilfully and imagina
tively, it can help define the pattern of health and disease within
populations and groups, identify environmental, behavioural and
other social factors that influence the health of the community, and
provide objective assessments of the impact of various interventions.
I
I
However, in some respects, epidemiology is the victim of its own
success. There is a tendency to assume that epidemiological studies
are so complex that they can be carried out effectively only by highly
trained specialists supported by skilled statisticians. This image has
intimidated many health workers and discouraged the use of this
powerful tool in routine public health practice. In many developing
countries, there are relatively few professionally trained epidemiologists, and'they work mainly in the central offices of ministries of
health in academic institutions and in research institutes. Most of
the epidemiological data available in these countries are derived
from a few special studies conducted by experts. Little use is made of
epidemiological methods in defining and analysing health problems
at community level on a routine basis.
Since the historic conference at Alma-Ata at which the represen
tatives of governments identified primary health care as the key to
the achievement of health for all by the year 2000, ministnes of
health have strived to strengthen health services at the community
level. Major programmes have been launched and national strategies
have been developed to meet this ambitious goal. In order to trans
late these national plans and strategies into effective action at the
community level, health workers need relevant, up-to-date knowl
edge of the pattern of health and disease, and of their detennmants
I
I
■<
Manual of Epidemiology for District Health Management
ineffective.
The rational management of health services at the community
demands .he .maginanve: use <of
3 *^Sk^bSs on epidemiology w« »Mn
■ ■ Ma
Me material availaMe for the
°‘Z!XX T^l^ai Mln
an. gap hy bring1‘Mi* worker. This manual fills
ing simple but efiecnve ep.dejnriogrcal_me.hods
th7toTe;el. It shows how they can use
13= Slue's mrmkhMh pnorrties wi*m the eommurn^,
"a’nd identifying risk factors. It describes
defining high-risk groups,
be used to design health services that are
how this information can
m the needs of the different groups
specifically targeted in Bnaiiuu
relation to
It also illustrates how epidemiological indicaI to evaluwide range of applications, this
“ °f h“lth
care providers at the community and district levels.
Adetokunbo O. Lucas, M.D.
Carnegie Corporation of New York
United States of America
VII
I
Preface and acknowledgements
There are a number of good books that set out the essentials of
epidemiology and a number of courses—mainly in technically ad
vanced countries—that provide training in epidemiology. However,
at present there is little epidemiological advice available that is
directly useful to those responsible for health care at the district
level—the managers who are responsible for implementing primary
health care. This manual attempts to be a practical guide to epidemi
ology and its relationship to planning, management and evaluation.
It emphasizes the use of epidemiological health information in
district health planning and shows how to obtain, analyse and make
use of this information.
The production of this manual, supported throughout its evolu
tion by the Scientific Working Group on Epidemiology of the
UNDP/World Bank/WHO Special Programme for Research and
Training in Tropical Diseases, has a long history' and several experi
mental versions have been produced. Many different authors contrib
uted sections in previous versions and made helpful comments and
suggestions. Those who helped in the writing and revision were.
E. H. Goh, K. Hughes, H.P. Lee, J. Losos, K.C. Lun, S. Lwanga,
W.O. Phoon, C.Y. Tye and F.K. Wurapa.
The first version was largely developed by Dr F.K. Wurapa and his
colleagues at the Tropical Diseases Research Centre in Ndola,
Zambia. That version was revised by participants at a workshop held
in Singapore in 1983, which was hosted by Professor W.O. Phoon
and staff at the Department of Social and Preventive Medicine,
University of Singapore. The participants at the workshop were.
A.A. Buck, H.M. Gilles, K. Hughes, H.P. Lee, K.C. Lun,
R.H. Morrow, W.O. Phoon, J. Storey, J. Teoh, C.Y. Tye and
F. K. Wurapa. In addition, many other people with epidemiological
experience in developing countries have made very useful comments
and provided many ideas for improvements to previous versions.
Professor Phoon and colleagues then undertook the difficult task
of producing the first experimental draft for field testing in 1985.
During 1985-86 Dr E. Lo, of the Ministry of Health, Malaysia, and a
number of district health officers used the manual for six months
and provided a formal evaluation of its usefulness. Further assess
ments were provided by many individuals from countries in Africa,
Asia and Latin America. The main conclusion was that the manual
needed to be more practical and that it should focus on the planning
and management activities that health workers were responsible for
within the context of the district health system.
vlfl
Manual of Epidemiology for District Health Management
1
The present publication draws upon the previous experimental
version, but a substantial amount of new material has been added
and the previous manuscript has been extensively rewntten.
In some respects, this version can still be considered experimen
tal The editors would welcome comments and suggestions for
changes and would particularly like to hear from health workers who
Programme for Research and Training in Tropical Diseases, World
Health Organization, 1211 Geneva 27, Switzerlan .
I
s.
1
CHAPTER 1
District Health Management
I
1.1
1.1
What is a district?
1
1.2
The district health management team
2
1.3
Sources of health information
1.4
Making a community diagnosis
4
5
1.5
Summary of epidemiological and planning
responsibilities of the DHMT
7
What is a district!
The district is the most peripheral unit of local government and
administration that has comprehensive powers and responsibilities.
It may be called by various names: the awraja in Ethiopia, the block
in India, the county in China, the district in Kenya and Malaysia,
the gun in the Republic of Korea, the kabupatan in Indonesia, the
municipality in Brazil, the sharestan in the Islamic Republic of Iran
and the upazilla in Bangladesh.
I
A typical district has a population of between 100 000 and
300 000 people and covers an area of from 5000 to 50 000 square
kilometres. The district headquarters is usually in the main town
where there are the offices of all the principal ministries that are
concerned with district and local affairs, such as health, agriculture,
education, social welfare and community development. The district
is the natural meeting point for "bottom-up" community7 planning
and organization, and for "top-down" central government planning
and development. It is, therefore, a natural place for the local com
munity needs to be reconciled with national priorities.
I
r
I
The district is the key level for the management of primary health
care (PHC). Ideally, all health-related activities taking place in the
district should be coordinated into a district health system. The mix
of manpower and facilities providing health care in districts varies
greatly from country to country. In the main communities, rural and
urban, there may be community health workers, clinics and health
centres, together with traditional and private medical practitioners.
A government district hospital and the headquarters staff for all the
district health programmes are often located in the main town. The
district may also have other services run by religious and other
nongovernmental organizations.
Manual of Epidemiology for District Health Management
2
Ss^i^SSFefforts to implement PHC.
Figure 1.1. The central role of the district health office within the district
health system
{Fomilies I
kHwl
---
= am:.:es |
= -.>es |
CT11; ICS
CHW
incentre
Ch-
Fem11 les I
CHW
_
jjjce'ure
neaiif
FgmiI les j
centre
Gove't
jnc V:0
<oliacorction
f-nt l tes
.•rditn worker 1/C |
iamjnr.v
acrtic'.xt'.an
Distf.n
ho',oita;
interectorci
i .e^tre
•*wiitn
Nur r>e
nuiwite
-ch . etc.
Healm auxiIiarv
Environmental nealt"
vector-borne diseases
water. IR/leprosv
:overrren’
• uCi5ito; ona/or
-asoitcl
:ri"tstrctor
P.311:
-“alt" "-rse
-CH.E?!
.jtriv.on
p-j31:c
-ealtn :":cer
E"v;'onmental ■ealt’*
vo<»cr-oorre diseases
-ater. sanitation
'R/Lwrcs*
KEY
<1) CHW . Commit/ neoitn worker
(2) EPI * Expanded □roqrcnrne on iTxnuatiun
(5) 1/C ■ Incnarae
(Ul MCH ’ maternal and cnild nealt"
<5) 10 ■: •edical off leer
- Non-goverrvnentol organize’‘.on
(6) NGO
i
i.
1 2. The district health management team
The district
health office is usually managed• ‘byr a team of health
S2E2SESL
officer.
3
District Health Management
The health services extend from the community health workers
to the hospitals. The district hospital may be the main centre for
curative health care and is commonly referred to as the first referral
level. As well as organizing the health services, the DHMT also has
to collaborate with local government and nongovernmental organi
zations, to Haise with community representatives and organizations
and to practise intersectoral coordination.
It is useful to divide the team's work into four main areas of
responsibility:
• District health planning, including community participation,
local government and intersectoral coordination and collaboration
in health.
• District health administration and the management of all com
munity health programmes.
• Training and supervision of all health staff.
• District hospital and outpatient services.
Figure 12. The main responsibilities of the district health management team
DISTRICT ADMINISTRATION
HOSPITAL WORK
DHMT
TRAINING
J
DISTRICT PLANNING
If the DHMT is to carry out all these responsibihties effectively,
one of its main priorities will be to gather and use a whole range of
health information. An understanding of epidemiology is essential
for all members of the team, since it will enable them to use health
information in health planning, management and evaluation.
4
Manual of Epidemiology for District Health Management
1.3 Sources of health information
The starting point for all health information is a good knowledge
of the district's population and the total number of people who are
"at risk" of needing a service. For example, how many pregnancies
occur each year in the district and what percentage of deliveries are
supervised by a trained health worker? What proportion of yotmg
children are fully vaccinated against measles and tuberculosis What
percentage of households have a reasonable water supply or a toilet.
To answer such questions the DHMT needs epidemiological
health information on:
• the population of the district, its age-sex structure, migration and
vital statistics.
• the main causes of morbidity and mortality.
• the organization of the district health services, particularly in
regard to access, coverage and effectiveness.
Health information is available from a variety of sources. In a few
countries the ministry of health may have the mforniation- ^ady
available for each district; in other countries the information exists,
but is scattered in the reports of various ministries or agencies,- and
in other countries much of the information, particularly maps,
census data, demographic and vital
ve
form that is usable at the district level, so that the DHMT may have
to obtain its own information.
How health information is collected within a district vanes con
siderably from country to country. All countries have a system for
collecting data recorded by PHC facilities, which is then collated at
the district level and reported to the ministry of health. This routine
health information system will be more developed and more reliable
in some countries than in others.
Health information may also be collected through a surveillance
system organized by the DHMT for a specific health problem or
disease, from the reports and analyses of epidemics and from special
investigations or surveys.
Other useful sources of district health information include health
surveys made by the ministry of health, census and demographic
surveys made by the bureau of statistics, and data collected by r
lated sectors, such as agriculture and education, and by nongovern
mental organizations, such as community and religious groups.
The following is a summary of the information that the district
health management team may need:
District Health Management
General information
• The district's history, physical and climatic characteristics, com
munity organization, economic development, people's occupa
tions, and organization of local government.
• Geographical distribution of villages and towns, major roads and
important features such as rivers and mountains.
Population
• The district's population size, age and sex structure, geographical
distribution, migration patterns and growth rate.
Health status, morbidity and mortality patterns
• Demographic indices for birth and fertility rates and for maternal,
infant, child and overall mortality rates.
• Common causes of morbidity, mortality and epidemic diseases.
• Important underlying health problems such as food availability,
housing, water supply and excreta disposal.
Health services
• Number and distribution of governmental and nongovernmental
facilities, personnel and programmes.
• Adequacy of management support, logistics and supplies.
District health programmes
• Pregnancy: antenatal, delivery and postnatal care.
• Nutrition: growth monitoring and malnutrition.
• Immunization: expanded programme on immunization.
• Environmental health: water supplies, excreta disposal
and hygiene.
• Communicable diseases control: cases diagnosed and
control activities.
1.4 Making a community diagnosis
There is a similarity in approach between clinical medicine and
community health. The clinician examines the individual patient
and has to recognize and identify the pathological significance of the
clinical symptoms and signs in order to make a specific diagnosis
and to prescribe the appropriate treatment.
5
1
1
8
Manual of Epidzmiology for District Health Management
■
In community health epidemiological skills are needed to exam
ine the whole population and to select the most suitable diagnostic
indicators that describe and explain the health problems in the
district. It is then necessary to make a community diagnosis and
decide which programmes would be most effective in raising the
health status of the population.
A clinician may order a variety of laboratory or other special tests
after making a prehminary assessment of a patient, based on the case
history and physical examination. In the same way, the DHMT may
need to organize special surveys in order to obtain more epidemio
logical information than is provided by the routine health mformation system.
However, there is a fundamental difference in the approach. The
clinician usually sees a patient after the disease has started and so
treating patients usually does little to reduce the number of new
cases of that disease or to remove the underlying health problems. By
contrast, the epidemiologist attempts to understand why the disease
exists in the first place and how it can be prevented. Ability to apply
the epidemiological approach is thus a fundamental skill tor all
health workers working in community health programmes that aim
to reduce disease and improve the community's health status. This
comparison is summarized in Table 1.1 and Figure 1.3.
Table 1.1. Comparison of clinical medicine and community health programmes
Clinical medicine
Community health programmes
1. Objective
Cure carient of disease
Improve health status of community
2. Information
Clinical history, physical
examination and
Population data, health problems, disease
patterns, availability of health services
laboratory investigations
3. Diagnosis
Differential diagnosis
and probable diagnosis
Community diagnosis and priorities for action
4. Action plan
Treatment and rehabilitation
Community health programmes
5. Evaluation
Follow-up and assessment
Evaluation of changes in health status
required
7
District Health Management
Figure 1.3. Clinical diagnosis and community diagnosis compared
HISTORY
TALK WITH COMMUNITY
EXAMINATION
RECORDS
TESTS
SURVEYS
PATIENT
DIAGNOSIS
COMMUNITY
DIAGNOSIS
1.5 Summary of epidemiological and planning
responsibilities of the DHMT
Although the responsibilities of the DHMT will vary from coun
try to country, its duties will involve planning for PHC, including all
the promotive, preventive and curative health services. To do this
the DHMT will need health information. It is the responsibility of
the team, therefore, to obtain this information and then to use it in
its work.
In order to perform the health planning and management cycle,
starting with the district population, the DHMT will need epidemio
logical skills (outhned in Figure 1.4) for the following tasks:
Storting from the district population:
• Define population groups by age, sex and location.
• Assess health and disease problems, particularly important causes
of morbidity and mortality.
• Collect health data through routine services, surveillance, epi
demics and surveys.
• Produce health information by analysis of data.
8
Manual of Epidemiology for District Health Management
I
• Interpret and communicate the health information.
• Assess the health status of the general population and
high-risk groups.
• Decide on the priority health problems.
• Use health information to choose between alternative
interventions.
• Implement improvements to health programmes.
• Estimate increase in access and coverage.
• Evaluate effectiveness of health programmes in reducing health
problems, morbidity and mortality.
• Determine changes in health status of the district population.
Figure 1.4. Tasks requiring epidemiological and
planning skills
District
Population
Improvements in
Health Status
t
I
Define population
groups
Evaluate
effectiveness
t
I
Assess health and
disease problems
I
I
Collect health
data
Produce health
information
I
Assess district
health status
l
Increase access
and coverage
t
Implement
programmes
t
Choose alternative
interventions
t
Decide on priority
health problems
I
I
I
1
Manual of Epidemiology for District Health Management^
10
WHO? WHERE? AND
WHEN? ARE KEY
QUESTIONS FOR SOLVING
HEALTH PROBLEMS
E“=Ha=s^~‘
approached through a senes of questions.
is the health problem, disease or condition, and what are its
What
manifestations and characteristics?
is affected, with reference to age, sex, soc^as?s' ethmC
Who
group, occupation, heredity and personal habits.
Where
When
How
Why
does the problem occur, in relation to place of residence,
geographical distnbution and place of exposure.
does it happen, in terms of days, months, seasons or years?
does the health problem, disease or condition occur, and
what is its association with specific conditions, agents'
vectors, sources of irdection, susceptible groups and other
contributing factors?
does it occur, in terms of the reasons for its persistence or
occurrence?
have been implemented as a result of the
So what interventions
" ■" Nation gained and what was their effectiveness? Have
there been any improvements in health status.
.
,
WHICH POPULATION
GROUPS ARE AT HIGH
health services but also on those who do not atten .
A most important concept in epidemiology is that; ofj
; RISK?
-----------------------it^co^cepro'f SeX^mt^^^
patients attended for leprosy treatment, or how many houses
9
CHAPTER 2
Epidemiological Principles
2.1
Definition and approach
9
2.2
Descriptive epidemiology
11
2.3
Measuring frequency
12
2.4
Numbers and rates
13
2.5
People, episodes or attendances?
15
i
2.6
Defining a case
15
i
2.7
Making use of rates
16
2.8
2.9
Health indicators
17
Types of indicators
18
2.10
Health status indicators
19
I
i
I
2.1
Definition and approach
A useful definition is that epidemiology is the study of the distribution, frequency and determinants of health problems and disease
in human populations. The purpose of epidemiology is to obtain,
interpret and use health information to promote health and reduce
disease. The basic epidemiological concepts are highly practical and
are relevant not only for the members of the district health manage
ment team but for all health workers.
It is useful for the DHMT to consider four phases in the use of the
epidemiological approach. The first is descriptive epidemiology. It
asks: what is the problem and its frequency, who is involved, where
and when? In the above definition it is the part concerned with
disease distribution and frequency.
The second phase is often called analytical epidemiology because
it attempts to analyse the causes, or determinants, of diseases by
testing hypotheses to answer such questions as: how is the disease
caused and why is it continuing?
The third phase is intervention or experimental epidemiology in
which clinical and community trials are used to answer questions
about the effectiveness of new methods for controlling diseases or for
improving underlying conditions.
The fourth phase may be called evaluation epidemiology because
it attempts to measure the effectiveness of different health services
i
Epidemiological Principles
WHAT COVERAGE HAS
BEEN ACHIEVED?
sprayed with residual insecticide; they also need to know the total
number of children, leprosy patients or households that should have
received the immunization, treatment or insecticide. A comparison
of those who actually received the service with those who should
have received it is called coverage. Achievement of high coverage by
the main programmes is the single most important managerial task
of the DHMT. An assessment of the coverage being achieved is the
starting point for improvements in district health planning.
I
2.2 Descriptive epidemiology
The first stage in understanding a health or disease problem from
an epidemiological perspective is to describe it by the characteristics
or variables of who? where? and when? After all the information has
been assembled, the second stage is an attempt to explain all the
facts.
A
Who!
The most important variables are age, sex, education, occupation,
income, cultural and religious group, family size, nutritional state
and immune status.
I
Other groupings might be by such characteristics as clinic attend
ance and non-attendance, those with latrines and those without, or
normal and low-birth-weight infants.
Any relevant variable may be used, provided that subjects can be
clearly placed in one category or another.
Where}
The place where people live or work may partly determine which
health and disease problems they suffer from and what use they
make of the available health services. For example, the variables
might be:
• Town, village or isolated dwelling.
i
• High or low altitude.
• Proximity to rivers, forests, wild animals or sources of toxic
substances.
• Distance from dispensary, health centre or hospital.
When}
1
I
It is important to know when health problems are most severe, or
when the incidence of new cases is greatest. To show this, cases,
episodes or events can be grouped according to new cases per day,
week, month or year. The time period depends on what is being
analysed, for instance:
11
12
Manual of Epidemiology for District Health Management
• New cases of cholera per day.
• New cases of measles per week.
• New pregnant mothers registering per month.
• New cases of kala-azar in one year.
Figure 2.1. Descriptive epidemiology - who? where? and when?
WHO
WHERE
WHEN
2.3 Measuring frequency
The two main measures of the frequency of diseases, health prob
lems and the use made of health services are incidence and preva
lence. It is most important to be clear about which is being used.
——---------------------INCIDENCE MEASURES ALL
NEW CASES DURING A
PERIOD OF TIME
Incidence measures the number of new cases, episodes or events
occurring over a defined period of time, commonly one year. Incidence is the most basic measure of frequency and is the best indica
tor of whether a condition is decreasing, increasing or remaining
static. It is, therefore, the best measure to use in evaluating the
effectiveness of health programmes. It is also the measure used in
surveillance systems and for analysing how people are using the
health services.
Examples include the births and deaths occurring in a district in
one year, cases of neonatal tetanus diagnosed per year, number of
women attending antenatal clinic for the first time per month, and
number of trypanosomiasis cases diagnosed per year.
Prevalence measures the total number of existing cases, episodes
or events occurring at one point in time, commonly on a particular
day. Prevalence may be more complicated to interpret than incidence
because it depends upon the number of people who have developed
their illness in the past and have continued to be ill to the present
time. It is a combination of the previous incidence of a condition and
its duration. Examples of frequency measured by prevalence are the
Epidemiological Principles
PREVALENCE MEASURES
ALL CASES AT ONE POINT
IN TIME
total number of leprosy patients on a register at the beginning of
each month or the number of hospital beds occupied per day.
Whereas prevalence is also very useful for chronic conditions, inci
dence is more useful for those diseases with a short average duration
like measles, diarrhoea, and pneumonia. Cross-sectional surveys
generally provide information about prevalence and are particularly
useful in establishing information about chronic diseases such as
leprosy or schistosomiasis.
Under stable conditions, incidence and prevalence are related by
the following formula:
prevalence = incidence x average duration of the condition.
Thus for those conditions with a long average duration, such as
leprosy and tuberculosis, the incidence per year is much lower than
the prevalence. For example, the prevalence rate of pulmonary tuber
culosis is commonly between 0.5% and 1.0% (or 5-10 per 1000
people) and the average duration of untreated illness is estunated to
be about 4 to 5 years. This means that the incidence of new cases of
pulmonary tuberculosis is 0.1% to 0.2% or 1-2 cases per 1000 people
per year. In countries with good diagnostic and reporting systems the
incidence of new cases of tuberculosis may be used. However, in
many developing countries without such systems, reliable informa
tion can usually be obtained from cross-sectional surveys which
provide prevalence data.
2.4 Numbers and rates
Incidence and prevalence are used in reporting health information
and may be given as a whole number or a calculated rate.
The most readily available data will be in absolute numbers.
These are often used in monitoring the occurrence of important
infectious diseases, especially in outbreaks, when the populations
involved are restricted in time and locality and the population struc
ture can be assumed to be stable.
When we have to look at trends over a period of time, or compare
rhe frequency of diseases between subgroups or communities, using
the total number of cases can lead to invalid conclusions. The popu
lation size and age-sex structure of each group must, also be consid
ered before the groups can be compared and the information should
then be expressed in terms of incidence or prevalence rates.
When calculating rates, the events or cases are related to the
population which has given rise to them. This is the population at
risk and refers to the group of people who have the potential to get
rhe disease and thus may contribute to the total number of cases. For
example, such a "population" may refer to the whole population of a
13
14
Manual of Epidemiology for District Health Management
district or a part of the district or all the people in a particular age
sex group.
The number of cases (made up by counting people, episodes or
attendances) is called the numerator and the total population at risk
the denominator. All people in the denominator must be, by defini
tion, at risk of becoming a part of the numerator population. Each
rate must have a time period or a date attached to it and this should
always be stated.
Incidence and prevalence rates are defined as follows:
incidence rate
new cases in specified period of time
x factor
total population at risk
existing cases at specified point of time
Prevalence rate
x factor
total population at risk
The rate is multiplied by a factor, the size of which is chosen so as
to enable the rate to be expressed as a suitable whole number; the
factor used is commonly 100, 1000, or 10 000.
Example 1: In a district with an estimated mid-year population of
200 000 people, there were 40 cases of kala-azar reported during
1987.
Incidence rate
40___
200 000
0.0002 cases per person per year.
Depending on which factor is used, this can be expressed as:
0.02 cases per 100 people per year, or
0.2 cases per 1 000 people per year, or
2.0 cases per 10 000 oeople per year.
Example 2: On 1 June 1987 there were 120 registered leprosy cases
in a district with an estimated population of 200 000 people.
Prevalence rate
120
x
200 000
0.06 cases per
factor
100 people on 1 June , or
0.6 cases per 1 000 people on 1 June.
The importance of using the appropriate denominator in the
calculation of various rates needs to be emphasized. If we are estimating the prevalence rate for a particular disease, then the denomi
nator used should be the total number of individuals who may be at
risk of contracting the disease in question. In the case of a sample
survey, this denominator may comprise all individuals in the
Epidemiological Principles
sample. If, on the other hand, we wish to estimate the "positive"
rate for Plasmodium vivax from the blood samples that have been
collected for the same survey, then the denominator used generally
should be the total number of individuals from whom blood samples
have been taken and the slides read—not the total sample size.
For age- and sex-specific rates the denominator includes only the
people in the relevant age or sex groups. For example, in the age
specific fertility rate for women aged 20-24, the denominator should
comprise only females in the study sample who are between the ages
of 20 and 24 years.
■ I
2.5 People, episodes or attendances^
■ r
I
I
ARE YOU COUNTING
PEOPLE EPISODES OR
ATTENDANCES?
It is extremely important to be quite clear whether the number of
cases is made up by counting people or episodes or attendances.
With diseases such as malaria or diarrhoea a person may have several
separate attacks in one year and may attend a clinic two or three
times for each attack. In this situation, only one person has been ill
but he or she has suffered several separate episodes in one year and
attended a health service several times for each episode. On the
other hand, a tuberculosis patient may only count as one episode and
be registered as one case but may have attended 12 times in the past
year.
What do we count—people, episodes or attendances? If we need to
estimate the proportion of the population sick with a chronic disease
(prevalence), we should count the total number of sick people. To
evaluate the effectiveness of a malaria control programme, we need
data on the number of new episodes (incidence) detected, commonly
in one year. If we are studying the use of health services, information
on new and repeat attendances is usually required.
1
2.6 Defining a case
Before starting to use incidence or prevalence measures, it is most
important to decide quite clearly how a case is to be defined. Failure
to do this can easily lead to confusion and misunderstandings. For
instance, people living in a malaria endemic area with fever, head
ache and body aches may be called malaria cases and treated as such.
However, the ministry of health would probably only accept these as
definite cases if they were confirmed by a positive blood slide.
To overcome these problems it is a good practice for some dis
eases to establish diagnostic criteria and classify cases into several
groups, such as possible, probable or definite, accordingly. For ex
ample, a patient with a fever, headache and body aches could be said
?
15
Manual of Epidemiology for District Health Management
18
to be a possible case of malaria; a probable case might be someone
who also responded well to antimalarial treatment. Only if there was
a positive blood slide for malaria parasites might the case be called a
definite one. However, it still has to be recognized that the symp
toms might be due to some other illness, particularly in children
living in malaria endemic areas.
Figure 2.2. The use of diagnostic criteria to define possible, probable and
definite cases of malaria
POSSIBLE
PROBABLE
DEFINITE
Cases with fever,headache
and body aches
Possible cases who
also respond to
treatment
Probable cases who
have positive blood
slide as well
I
A poor definition of what constitutes a case, episode or attendance
is a frequent problem and one which may lead to false estimates of
frequency and to false conclusions about changes in frequency.
I
i
2.7 Making use of rates
WHAT ARE THE CRITERIA
FORA POSSIBLE,
PROBABLE AND DEFINITE
CASE?
There are two main reasons for using rates as opposed to whole
numbers.
• To make comparisons between two different populations that
may have different numbers of people at risk, by standardizing for
population size. For example, it may be important to make com
parisons between several districts or between what is happening
now in the district compared with 10 years ago.
• To calculate the number of expected cases. By using a known rate
(e.g. the national prevalence of leprosy may be 1 per 100 people,
or the infant mortality rate 12.0 per 1000 infants per year), the
approximate number of cases that are expected to occur in the
district in one year (total leprosy patients and infant deaths) can
be calculated.
It is most important for DHMTs to use the best of the known
rates to calculate the expected number of cases, since it is these that
Epidemiological Principles
the district health services have to serve. Known rates are likely to
be more accurate than those calculated from district data. This
problem is explained in more detail in Section 4.7.
2.8 Health indicators
RATES ENABLE US TO
MAKE COMPARISONSAND
CALCULATE THE NUMBER
OF EXPECTED CASES
Indicators are a measure that can be used to help describe a situ
ation that exists and to measure changes or trends over a period of
time. Most health indicators are quantitative in nature but some are
more qualitative.
The DHMT needs to use health indicators to analyse the district's
commitment to pohcies for socioeconomic development and PHC,
to monitor progress in implementing health programmes, and to
evaluate their impact on the health status of the population. Health
indicators are necessary in order to:
• Analyse the present situation.
• Make comparisons.
• Measure changes over time.
HEALTH STATUS IS AN
INDICATOR OF
DEVELOPMENT
Indicators provide a means of comparing different districts in the
country and measuring their progress in raising health status. Indica
tors can bring out the difference in health status between particular
subgroups in the population, such as the privileged and the poor, or
between rural and urban areas. Health and nutritional indicators are
also indirect measures of overall development and direct indicators
of the quality of life. In fact, development planners and economists
are increasingly using social and health status indicators as a guide
to monitor progress with different development strategies.
Health indicators may measure the actual situation directly or
they may be used as indirect measures. For instance, the infant
mortality rate (IMR) is a direct measure of the actual risk of infants
dying in their first year of life, but the IMR is also used as an indirect
measure of overall socioeconomic development.
However useful an indicator may be, there are technical and
financial problems in collecting the necessary data. But how accurate
and valid do the data have to be for the indicator to be useful? This
varies with the indicator and how it is going to be used. For analys
ing the present situation and for making many comparisons, indica
tors for use in policy-making and health programme management do
not need to be highly accurate. For instance, it is usually sufficient
to know that the IMR is between 40 and 60 per 1000 live births or
100 and 120 per 1000, or over 150 per 1000.
However, when measuring changes in health status over rela
tively short periods of time, such as five years, much greater accu
racy is required. For instance, the IMR needs to be very carefully
17
18
Manual of Epidemiology for District Health Management
calculated if it is to be used as a measure of the improvement in the
district's health status. In these situations, a trend over some time is
the best indication that the situation is either improving, deteriora
ting or remaining unchanged.
Information for calculating these indicators comes from:
• Registration of births, deaths and diseases.
• Population censuses.
• Routine health information systems.
• Surveillance.
• Investigation of epidemics.
• Sample surveys.
2.9 Types of indicators
It is useful for the DHMT to classify health indicators into
those for:
• Health policies.
• Social and economic development.
• District population.
• Provision of health care.
• Health status.
The use of health policy indicators may involve a considerable
degree of judgement by the DHMT, as they are difficult to quantify
at the district level. They include such indicators as the level of
political commitment to PHC and the availability of a public policy
statement and written health plans; the allocation of manpower and
financial resources from the total district resources available,- the
degree of equity in the distribution of resources and facilities
throughout the district; the availability of a decentralized organiza
tion in the district for health planning and management; mecha
nisms for community participation; the degree of intersectoral
coordination; and the amount of collaboration between government
and nongovernment health organizations.
Social and economic indicators are useful in analysing the under
lying situations that affect health. This group includes such indica
tors as level and distribution of economic wealth; types and levels of
employment; school enrolment and adult literacy; availability of
reasonable housing and number of people per room; and availability
and distribution of food supplies by household and by season.
Population indicators cover such factors as age-sex structure,
density, distribution and migration. Other indicators are concerned
Epidemiological Principles
with population growth, such as birth and death rates, fertility and
rate of natural increase.
Indicators on the provision of health care are concerned mainly
with access to health programmes and facilities, particularly com
munity health workers, subcentres, health centres, first referral level
hospitals and coverage by the eight essential elements of PHC:
health education; food supplies and proper nutrition,- safe water and
sanitation,- maternal and child health, including family planning,immunization; prevention and control of endemic disease,- appropri
ate treatment,- and provision of essential drugs and supplies. In addi
tion, other indicators are concerned with the available resources,
such as the number of facilities and health workers in the district
and the finances or money available for PHC.
Health status indicators are mainly concerned with nutritional
status, morbidity and mortality. These are considered in more detail
below.
2.10 Health status indicators
The most useful indicators of health status can be grouped in
three categories:
• Nutritional status.
• Morbidity.
• Mortality.
Nutritional status can be estimated in several ways. The percent
age of newborn babies who have a low birth weight (LBW)—less than
2500 grams—is widely used. Anthropometric measurements, such as
weight-for-age, height-for-age, weight-for-height and mid-upper-arm
circumrerence are also commonly used for assessing nutritional
status of infants and young children. Health status is indicated by
the percentage of children who are classified as suffering from mild,
moderate and severe malnutrition.
Morbidity indicators are generally based on the disease-specific
incidence or prevalence rates for the common and severe diseases,
such as malaria, diarrhoea or leprosy. A simple method for assessing
morbidity is to analyse the pattern for all ages together and to derive
the ten commonest causes of ill health. A more accurate method is
to analyse each major age group separately.
Mortality indicators are, mainly, the crude mortality rate for all
ages, infant mortality, 1-4-year old child mortality, maternal mortal
ity, the expectation of life at birth and the disease-specific mortality
rates.
More details on these health status indicators are given in Chap
ter 3 on the district population and in Chapter 4 on epidemiological
18
20
Manual of Epidemiology for District Health Management
health information. The use of health indicators in district health
pl arming, management and evaluation is considered in more detail in
Chapter 13. More details on definitions and the meaning of the
different terms are given in Chapter 14.
A list of basic health status indicators might be the following:
• Fertility rate.
• Nutritional status.
• Infant mortality rate.
• 1-4-year-ord‘mortality rate.
• Maternal mortality rate.
• Life expectancy at birth.
21
CHAPTERS
District Population
3.1
Total population
21
3.2
3.3
3.4
3.5
3.6
3.7
Population density
23
23
26
27
28
3.8
3.1
Demographic rates
Population growth
Sources of population information
Accuracy of data
An example: the Malumfashi Endemic
Diseases Project
31
District population checklist
32
Total population
Most developing countries perform a nationwide population
census about once every 10 years. The information is collected by
enumerators visiting every known household in small enumeration
areas and asking about all people living in the household on a par
ticular day. The figures for each area are added together to give a
total figure for the district, which commonly has a population of
between 100 000 and 300 000 people. The district figures are further
aggregated to give the national figures, which are then published as
census reports. These reports and the district figures are usually
available at government offices at the district headquarters.
A typical population distribution according to age group for a
developing country is shown in Table 3.1; the actual percentage
figures should be available for most countries. Infants under 1 year
old (0-11 months) commonly make up 3-4% of the total population
in many developing countries,- children aged 0-4 years about 18-20%
(one-fifthb and those aged 0-14 years about 40-44% (two-fifths).
Women in the fertile age range (15-44 years) account for about
20-22% (again one-fifth). Fertile women and young children under 5
years, therefore, make up about 40%, or two-fifths, of the total
population. Such national percentage figures can be used to find
the approximate total number of people in the main age groups in
the district.
I
The age-sex structure for the total population can also be shown
by a population pyramid (see Figure 3.1), using the percentage of males
and females in each 5-year age group. The age-sex breakdown may
vary in different areas. For instance, there tend to be more working
22
Manual of Epidemiology for District Health Management
men in towns, near mines and on plantations, and where there is
considerable rural-to-urban migration by men, the villages tend to
have a higher proportion of older people, women and children.
Table 3.1. Typical distribution of population by age
group for a district of a developing country
Age group /years)
Percentage
District population
Less than 1
4
14
26
43
13
100
8 000
28 000
52 000
86 000
26 000
200 000
1-4
5-14
15-44
45+
Total
In districts where there are marked ethnic or tribal differences it
can also be important to know the percentage of each in the different
age groups, particularly as the utilization of health services may be
different for each ethnic or tribal group. For example, the immuniza
tion coverage might be very different between two different ethnic
groups.
Figure 3.1. Population pyramid for Bangladesh
(based on the 1979 national census!
75*
70-74
Females
Males
65-69
60-64
55-59
50-54
S 45'49
0)
40-44
g 35-39
30-34
< 25-29
20-24
15-19
10-14
I
J
03
5-9
0-4
Q
86
4202468
Percentage of total population in each age group
!
ao
23
District Population
■
The information contained in a population pyramid is useful for
providing an estimate of the denominators (the at-risk population)
necessary for calculating certain age- and sex-specific rates. For
example, Figure 3.1 shows that the combined percentage of male and
female children aged 0-4 years is about 17% and the percentage of
women in the fertile age range is about 15.5%. This latter percentage
is slightly lower than might be expected, but the shape of the female
half of the pyramid suggests that there was some under-enumeration
of young females in the census.
IMPORTANT GROUPS IN DISTRICT POPULA TION
0-11 months
0-4 years
1:25
All children
0-14 years
2:5 to 1:2
Women
15-44 years
1:5
Infants
Young children
Women and young children
1:5
2:5
3.2 Population density
This is commonly expressed as the average number of persons per
square kilometre (km2!. The density can vary markedly between
different districts and even within districts. Density tends to be
higher in areas with large towns, fertile soil, and more advanced
development. Migration can be an important factor in areas with
rapidly increasing or decreasing population density. Some districts
will have densities of over 1000 people per km2, particularly in Asia,
whereas in Africa many districts have fewer than 50 people per km2.
I
A knowledge of the district's population density and distribution
is obviously important when planning health services, particularly
for new subcentres and health centres, and in evaluating the access
to and coverage of different health programmes.
3.3 Demographic rates
1
The crude birth rate tCBR) is usually estimated from the census or
special demographic surveys and is given by this formula:
I
:l
CBR
total birtns in one year
total midyear population (all ages, same year)
x 1000
The CBR in high-fertility countries may be around 45 births per
1000 people per year and in areas of lower fertility it may be about
20 births per 1000 per year. The rates are usually available for each
district, and by applying them to the district population we can
24
Manual of Epidemiology for District Health Management
estimate the total number of births expected per year. For example,
in a district of 200 000 people with a CBR of 45 births per 1000,
there would be about 9000 births per year, or about 170 per week.
-Total ••births
•
=
— x copulation
population = —
—- x 200 000 =
1000
woo
9000 per year
If the health information system reports that about 80 births per
week are attended by trained health workers, the coverage can be
estimated to be about 50%. How well is the district doing?
The fertility rate (FR) is an age-sex specific rate usually derived
from the census or special demographic surveys. This rate is a meas
ure of how frequently women in the fertile age range (15-44 years)
are having babies, so where the CBR is high the FR will also be high.
Developing country populations with an average fertility might have
a rate of about 100-150 births per 1000 women aged 15-44 years per
year; in high-fertility populations it might be around 200 per 1000
and in a population with lower fertility it might be about 60 per 1000.
The crude death rate (CDR) is calculated as :
CDR
total deaths in one year
total midyear peculation (all ages, same year)
x
1000
The CDR commonly ranges from around 10 deaths per 1000
people per year in more developed areas to more than 20 deaths per
1000 per year in poorer populations.
The infant mortality rate (IMR)—which is the proportion of all
liveborn infants who die in the first twelve months of life—is com
monly considered a good measure of health status. It is usually
calculated from the census or special demographic surveys. There are
many technical problems in calculating accurate IMRs and health
workers should not rely on the accuracy of their estimates unless
there is a very good vital registration system in operation. The fol
lowing formula is commonly used:
IMR
total infant (aged <1 year) deaths during one year
x
1000
total births in same year
In many poor populations in developing countries, the IMR often
ranges between 60 and 150 infant deaths per 1000 births per year,
but in severe conditions it may go as high as 200 or more. The dis
trict IMR is an average figure and the actual figures are frequently
higher in some poor, disadvantaged groups and lower in richer
groups. Most of the infant deaths occur during the first month of life,these deaths are called neonatal mortality. The total number of
infant deaths can be calculated as follows:
25
District Population
IMR
No. of infant deaths
iooo
x no. of births
Thus, in a district with a population of 200 000, 9000 births per
year and an IMR of 100, the number of infant deaths would be:
903 cer vear, or approximately 17 per week
x 9000 =
1000
Thethild mortality rate jCMR) is based on deaths between 1 and
4 years of age and is important because malnutrition and infectious
diseases are common in this age group. It is usually calculated from a
census or special surveys since it is not easily calculated with suffi
cient accuracy from district health information.
A neglected death rate is the maternal mortality rate (MMR),
partly because it is difficult to calculate accurately. An approximate
rate for many developing countries is 1-5 maternal deaths per 1000
births per year, which means that a district with a population of 200
000 and a CBR of 40 per 1000 might expect between 8 and 40 mater
nal deaths per year. In this case it is more important to know the
true numbers than the rate, since the actual numbers are so small.
The use of births as the denominator, instead of the number of
women of child-bearing age, may give the impression that the prob
lem of maternal deaths in developing countries is less serious than it
is in reality. For example, even the fact that MMR may be 5 per 1000
in Africa compared to 5 per 100 000 in Europe does not adequately
reflect the much greater risk of mothers dying from pregnancyrelated causes in Africa. This is because the average number of
births per woman is also much higher in Africa and therefore the
risk of a particular woman dying of pregnancy complications is
today about 400 times greater in many developing countries than
in developed areas.
materr.al pregnancy-related deaths in one year
MMR
total b'.rtb.s in same year
The factor is usually 1000 or 100 000.
FOR A DISTRICT OF200 000 PEOPLE THE FOLLOWING
RATESAND TOTALS MIGHT BE EXPECTED:
20-45 per 1000 or 80--70 live births per week
CBR
10-20 per 1000 or 40-50 all deaths per week
CDR
s
MR
MMR
60-150 per 100
or
5-25 infant deaths per weeks
1-5 per 1000
or
1-3 maternal deaths per month
x factor
i
28
Manual of Epidemiology for District Health Management
3.4 Population growth
The population growth in a district depends on the balance be
tween the number of births and people migrating into the district on
the one hand, and the number of deaths and people migrating out on
the other. Occasionally, a district's population may actually be
declining, but this is usually due to migration away from the area,
and not because deaths outnumber births.
The rate of natural increase, which excludes migration, is com
monly between 1% and 3_% per year in many developing countries
and is calculated as follows:
Rate of natural increase
=
CBR minus CDR
This rate largely determines how fast the district population will
grow, as shown in Table 3.2.
Natural growth in the district population
Table 3.2.
Rate of natural
increase
Present district
population
Population increase
in 10 years
in 20 years
total
% increase
total
% increase
1
1%
2%
3%
200 000
220 900
10
244 000
200 COO
243 800
297 200
200 COO
268 800
22
34
22
49
361 200
81
I
Figures calculatec to nearest ICO oeoole and percentages to nearest whole number
Figure 32. Growth in district population due to natural
growth and migration
NATURAL
=!=THS
*!
■
II ■
DISTRICT
ARRIVING
leav:\$
(mi
I
I
I
District Population
(
i
WITH 3% NATURAL
GROWTH THE POPULATION
WILL DOUBLE IN 25 YEARS
Besides the rate of natural increase in population, the number of
people migrating into or out of the district must also be taken into
account. This number will probably have to be estimated, as accu
rate figures on migration are not usually available. Some estimates
may be available from the last census, but for most districts you will
probably have to rely on local knowledge.
Estimates of population growth can be derived from the size of the
population at two or more points in time. The simplest way of esti
mating population growth is to obtain the difference between the
population size at two points in time and then to divide this differ
ence by the number of years7 interval between them. This yields the
average growth in the number of persons per year.
Example: If the population in an area was estimated to be 7830 on
31 March 1985 and 8450 on 30 September 1989, then the average
increase per year is estimated to be (8450 - 7830) divided by 4.5 =
138 people. The estimated population on 30 September 1990 is
therefore 8450 + 138 = 8588.
This method assumes that the increase in the number of people
per year is constant. However, when used for projections over a
longer period of time, this method tends progressively to underesti
mate the total population, as populations tend to grow at a constant
rate of growth rather than by a constant absolute increase per year
(as illustrated in Table 3.21.
3.5 Sources of population information
A knowledge of the number of people living in the district, with
additional information on their age, sex and geographical distribu
tion, is necessary’ for several aspects of planning and evaluation of
health services. DHMTs will need population estimates for the
district to provide:
• Total population by age and sex groups and other relevant criteria.
• Total number of expected live births and deaths per year.
The published sources for such information are:
Reports on the census. If a population census has been taken re
cently and the data for the area in question are available, the
census report can be invaluable. Often the information is not
readily available on a local area basis and other ways to obtain it
will have to be explored.
Reports of other studies. A study carried out by somebody else in
your district may yield useful population data that can be useful.
For example, there may have been an agricultural or economic
survey in the area, or a survey in connection with social research.
There may also have been studies previously carried out by other
health authorities or research organizations. Extensive mapping
and census data are often collected for malaria and other largescale disease control programmes.
27
28
Manual of Epidemiology for District Health Management
Other sources. Other possible sources of information are the au
thorities concerned with the provision of other services in the
area, for example, housing, education, law enforcement and public
utilities. Valuable information may be obtained from socioeco
nomic development schemes carried out by sectors other than
health, e.g. ministries of agriculture, water, labour, social welfare
and rural development. Religious organizations are often well
informed and may have properly documented data.
Assistance may be sought from senior officers in the medical and
health services, and from research personnel and demographers or
statisticians in the government and in universities. It is possible that
similar requests for assistance may have been received from other
parts of the country, and a programme might then be drawn up at a
central or regional level for obtaining more population data or for
updating old information.
If the required population data cannot be found from any of the
sources previously mentioned, then the health team may have to
obtain them directly or rely on the best estimates available.
Two quick ways of estimating small local populations are:
• To ask all local leaders how many people they are responsible for
and to add all the responses to obtain a total. However, beware of
being misled for various reasons, such as fear of taxation.
• Ask for the total number of houses, or count them, and multiply
by the known average household size for your district. If the
average is not known a useful method is to visit every 10th or
20th household and ask for details on all the people who normally
live there. Visit between 100 and 150 households. The average is
commonly around five people per household in many developing
countries in rural areas. It may be lower or higher for urban
households.
3.6 Accuracy of data
When considering how to obtain the necessary data, it would be
wise to keep in mind that the accuracy of population data is limited,
not only by the resources available for their collection, but also by
the level of socioeconomic development and the educational and
cultural sophistication of the population itself. Because of the
numerous factors that can contribute to inaccuracies in population
data, it is very important to have built-in mechanisms for validating
various important items of information. For example, to elicit infor
mation on the number of children a mother has had, a common
method is to obtain the birth order of all her children, starting with
the first bom, and ask about pregnancies in the long gaps,- a preg
nancy about every two years is a frequent pattern in many countries.
29
District Population
How accurate and complete population data must be before they
can be useful depends on the purpose to which they are to be put.
Sometimes even very rough estimates may be good enough and
resources would be wasted on efforts to obtain more accurate data. In
any case, data are rarely useless simply because they are not abso
lutely accurate.
In general, the more developed the medical and health services
are, the greater will be the need for detailed and accurate data for
planning and evaluation. However, in such services, it will also be
easier to obtain more accurate data.
The target in terms of accuracy can never be 100%. However
desirable it may seem to be, it is in practice neither possible nor
necessary to attain this degree of accuracy. In practice the aim
should be to obtain:
• Data that are as accurate as reasonably possible, given the re
sources available.
• An estimate of the nature and extent of inaccuracies in the data.
It is as important to try to estimate what inaccuracies are present
as it is to try to obtain data as accurately as possible.
To improve the accuracy of the information on various population
characteristics, it will be helpful to consider the following particular
examples.
Age
Age is a basic variable that is particularly required, as it is closely
related to disease patterns. It may also be one of the most difficult
variables to ascenain accurately. Although people often do not know
their age in years, they may know in which year they were bom.
Hence it may be worthwhile to ask both questions and see which
gives the most reliable answers.
In situations where birth records are not kept, age can sometimes
be estimated according to a calendar of notable events that occurred
in the community. In such a situation, for example, a person's age
might be estimated by asking:
"How many seasons after the earthquake were you bom!"
A more precise method of estimating age in the younger age
groups is usually required. For those in the age range of 6-24 months,
age can be crudely estimated from the following formula, provided
the child has not suffered from severe malnutrition:
:
Age (in months)
=
6 + number of erupted teeth
W-- -
A considerable proportion of people will give different ages on
separate occasions. For instance, it was found that 43% of those
enumerated in Ghana in 1960 by two interviews were classified in
different five-year age groups, older age groups being more inaccurate
1
I
30
Manual of Epidemiology for District Health Management
than younger ones. One form of age error, which is very widespread,
is a tendency to round off ages and say one is 30 or 40. Sometimes
men wish to seem older as this brings more prestige and to report
their wives as being younger. In order to avoid the distortion of the
age structure which may arise from wrong information about age, it
is best to group people by five- and ten-year age intervals e.g. 0-4,
5-9, 10-14, 15-24, 25-34, 35-44, 45-54, 55-64, 65+.
Because of these inaccuracies, it is important to use strict criteria
when recording age, to specify how age was calculated, and to men
tion the source of the information.
Sex
HOW ACCURATE ARE THE
DATA?
Sex is another important characteristic to record, because of the
different physiological and behavioural patterns in the two sexes.
Marked differences between the sex composition of a survey popula
tion and the general population could result in the survey results not
being valid. Considerable care is needed to avoid this. For example,
in some, countries it is a common practice for women to be secluded
in their homes and interviews may be very difficult to obtain, so that
women are under-enumerated in censuses.
Ethnic group
People in the same ethnic group tend to have similar social and
cultural practices and some of these practices may result in either
higher or lower disease frequency than in other groups. It may be
important, therefore, to study different ethnic groups and to under
stand their particular sociocultural patterns, as this could provide
clues to how disease may be reduced in the community.
Marital status
The precise definition of what is meant by a "married" or "sepa
rated" state may raise problems. Such definitions vary considerably.
Often such distinctions in marital status do not play as important a
social role as in traditional western society. A thorough understand
ing of the population and of its customs and life-styles is thus impor
tant, for it will help in deciding whether or not to include such
variables in the study.
Occupation
In obtaining information about the person's occupational status,
decide whether present or past occupations are to be recorded. For
example, a person who has recently started his or her present job
may have worked for the previous 10 years in a totally different
occupation. It may be more useful to record the occupation in which
the person has spent the longest time, as well as his or her present
one. A worker might have spent the past 15 years working in a
granite stone quarry, but as a result of pulmonary silicosis was un
able to continue and recently obtained employment as a watchman.
1
District Population
If the records show only his present employment, it will be difficult
to understand how a watchman contracted silicosis. Changes in
occupation from one season to another are also common in rural
areas. Some villagers may farm during the rainy season and fish after
the harvest. The possibility of exposure to more than one risk must,
therefore, be kept in mind.
Other variables
Data on other variables such as parity, religion, social class, place
of residence and mobility may be required. For census and registra
tion systems, nomads can pose difficulties. Village heads rarely
know the whereabouts of all the nomads from their villages. A
problem encountered in some parts of Africa is that the men leave
their camps early in the morning to look after cattle and the women
are not permitted to mention their husbands' names. These ex
amples show it is necessary to adapt methods of collecting informa
tion to the local context.
3.7 An example: the Malumfashi
Endemic Diseases Project
In this project on endemic disease in Nigeria, maps were required
before census enumeration could be done. Because of the inadequacy
of existing maps, basic maps at a scale of 1:20 000 were drawn, based
on recent air photography. These maps were then checked in the
field when the households were assigned numbers; compounds that
had been omitted on the base map were added and those that had
been abandoned were deleted. Such maps were adequate for scattered
settlements, whereas new large-scale plans were drawn for densely
populated settlements.
The enumeration questionnaire used in Malumfashi was similar
to that used at Macbakos in Kenya, which was a modified version of
the form designed for the 1969 Uganda census. The first half of the
enumeration questionnaire used at Malumfashi asked for such basic
information as age, sex, and residence. The second half consisted of
more detailed questions on survivorship, fertility and mortality. The
main sections of the questionnaire were as follows:
!
I
• Each individual was identified by name, a survey number and the
geographical location. The construction of individual survey
numbers intentionally contained the location so that the number
alone provided a guide as to where the individual was resident.
• Data were collected on residence. For demographic purposes it
was essential to distinguish between residents and visitors when
defining the population. Enumeration was carried out over some
months, and a de jure definition of population was used, i.e. based
on a population that comprises all persons who usually reside in
the defined area.
31
32
Manual of Epidemiology for District Health Management
• The date of birth and sex of each individual were noted. It was
intended that the date of birth should be accurate to the month
for children under 5 years old and accurate to the year from age 5
years onwards.
• Cultural group and stated religion were recorded.
• Marital status was noted for each individual.
• Questions were asked on the survivorship of parents, first spouses
and oldest siblings, and deaths in the household during the previ
ous 12 months were recorded, to provide estimates of mortality.
• Each female aged 15 years or over was asked questions about the
number of births she had had, and also about her most recent live
birth and whether the child had survived or not. Fertility rates
and infant and child mortality rates were computed from this
information.
3.8 District population checklist
• Obtain or draw a large-scale map of the district and mark on it all
health facilities and large villages.
• Determine the present total population for the district and its
density in different areas.
• Assess migration patterns and estimate the annual district growth
rate from official published figures.
• Calculate the total number of people in the 0-11 months and 1-4,
5-14, 15-44 and 45+ years age groups and the total number of
women aged 15-44 years.
• Obtain the official district figures for the crude birth rate, crude
death rate, infant mortality rate and maternal mortality rate.
• Work out the expected total number of births and deaths per week
and per year for the district. Do this for all deaths, as well as
infant, child and maternal deaths separately.
I
I
33
C HAPTER4
Epidemiological Health
Information
) I
I I
4.1
I
library
4.1
Health status assessment
33
4.2
Important diseases
34
4.3
Sources of epidemiological information
35
4.4
Morbidity patterns
37
4.5
Mortality patterns
38
4.6
Seasonality
40
4.7
Using morbidity and mortality rates
41
4.8
4.9
Death registration and certification
42
District health information checklist
43
Health status assessment
An assessment of the health status of the community, based on
information about health problems and diseases, is necessary’ for
planning and evaluating the health services. However, since the cost
of obtaining such information, in terms of resources and time, can be
considerable, it is very important that the DHMT selects the data
that are most feasible to collect and that will be most useful for its
work in the district. In addition to this information, the DHMT will
also need information on the health resources of the district and how
they are being used. The use of this information in district health
planning and management is dealt with further in Chapter 13.
Useful information on health status can often be obtained fairly
easily from the information reported to the district headquarters.
This information can be made of greater value, not by a detailed or
elaborate analysis, but by looking at it in its proper context. For
example, if in a reported outbreak of a dozen cases of jaundice there
have been two or three deaths, this indicates immediately that the
problem may be yellow fever and not hepatitis. Similarly, a general
1
I
increase in reported episodes of diarrhoea in older children and
adults may be the start of a cholera outbreak. District health workers
are in the best position to put such locally collected information mto
its proper context.
34
Manual of Epidemiology for District Health Management
Each country, region and district will have to decide on the spe
cific items of information required for planning, management and
evaluation of its health services. Each type of information has its
uses and limitations. A useful concept is that of the ^iceberg phe
nomenon", which emphasizes that the mass of health problems lies
below the surface and is frequently not fully recognized (see Figure
4.1).
In most countries a routine health information system already
exists, but health workers may have to take the time to uncover and
analyse the "hidden" data it contains. This system will provide
useful information or at least a good "guestimate". Make the best
use of the data being collected, rather than rejecting it as unsuitable.
On the other hand, the collection of useless and unnecessarily com
plicated data should definitely be discouraged.
Figure 4.1. The iceberg phenomenon—routine information comes mainly from
people who attend the health services
PREVENTIVE CARE
CURATIVE CARE
Pol lo
Drugs
VACCINES
PEOPLE
SEEN
PEOPLE
NOT
SEEN
Measles
1
________________ _ _______ _________ __ ___________________________
L———— —‘
4.2
Important diseases
Which are the important diseases? From an epidemiological view
point there are two factors that are indicative:
• Frequency—high incidence or prevalence, including potentially
epidemic diseases.
• Severity—causing much disability and a high mortality.
35
Epidemiological Health Information
IMPORTANT AND
CONTROLLABLE DISEASES
SHOULD HAVE THE
HIGHEST PRIORITY
For example, falciparum malaria is important because it can have
a high incidence and lead to many deaths. Similarly, malnutrition
can have a high prevalence and a high mortality. Some diseases are
important because they are potentially epidemic, such as cholera,
meningococcal meningitis and trypanosomiasis. Many outpatient
attendances are for minor self-healing illnesses and, although they
are important to the individual, they are not particularly important
to the community as a whole.
Diseases that have a high frequency and are severe, and which are
preventable or controllable, should receive the highest priority in
planning health programmes.
I
I
I
4.3 Sources of epidemiological information
The routine district health information system commonly has
information on the frequency and distribution of the locally impor
tant causes of morbidity and mortality. However, often this informa
tion is not presented in a way that is easy to understand and use.
Morbidity information gives the overall picture of ill health in die
community. Although this information is often deficient in quality
and quantity in developing countries, the district sources include the
following:
• Hospital inpatient records
• Workplace records
• Outpatient records
• Schools
• Disease notifications
• Special surveys
Hospital inpatient records
Analysis of hospital and clinic records can provide high-quality
information on the most important causes of major illness in a
community, but for them to be useful as an indicator of the health
status of the whole population, allowance must be made for the
strong tendency for inpatients to come from among people living
near the hospital, the wealthier and the better educated. In some
countries, many seriously ill patients never reach a hospital if they
live far away.
Outpatient clinic records
Records of people attending health centres and health posts as
outpatients may provide information, but there are problems in
collecting the necessary data. For example, diagnoses are frequently
given in terms of the chief complaint or symptoms; attendances are
given in terms of total visits rather than by new and repeat visits;
and people attending for immunizations or other preventive services
may be recorded together with those who have come because they
are ill. The information suffers from selection biases similar to those
3B
Manual of Epidemiology for District Health Management
■
mentioned for hospital records. Although health centres and health
posts may cover a wider population than hospitals, patients who live
near a facility and who can afford the time and fees, if any, are the
people most likely to attend. Such records, however, do provide
information about the use made of outpatient facilities and the most
frequent complaints, and do help to describe the pattern of disease in
a community.
Disease notifications
Notification systems are restricted to a selected list of ''important
diseases", which may differ from one country to another. By and
large, these diseases are the infectious ones which require prompt
action for control. Medical practitioners and other health personnel
may have a special responsibility and may be legally required to
provide such notifications. The health officer in charge of the district
is usually responsible for receiving the notifications and taking the
most appropriate action.
Workplaces and schools
Workplaces may provide data on absences due to sickness as well
as the results of any periodic health examinations. Data on those
employed reflect the situation in a selected sample of people who
work; people who are ill may not have been employed in the first
place, or they may have had to leave employment because of ill
health.
Schools can provide data on absences due to sickness as well as
the results of screening programmes by school health services. In
countries where school attendance is low the information may be
substantially biased and likely to miss those children who are so
cially and economically disadvantaged.
Special surveys
Ok-
There are two ways in which important diseases may be under
recognized by district information systems. If the disease has a low
frequency (incidence or prevalence) there may be too few cases for
them to show up clearly in a poorly functioning health information
system—leprosy is a good example. The other way is when the ill
ness does not produce clear symptoms and signs that are easily
recognized by the people themselves or the health workers. Schis
tosomiasis, filariasis and malnutrition are good examples.
Special surveys and research studies are often required in these
'•
circumstances, particularly for subclinical and chronic infections
such as malaria, African trypanosomiasis, Chagas disease, filariasis,
leprosy and schistosomiasis. The same applies to chronic physical
and mental disabilities and eye diseases. The importance of polio
myelitis and neonatal tetanus in many developing countries was
demonstrated mainly by such special surveys.
'
;
I
Epidemiological Health Information
37
4.4 Morbidity patterns
ROUTINE INFORMATION
SYSTEMS MAINLY REPORT
COMMON AND EASILY
RECOGNIZED DISEASES
Because of the high infant and childhood morbidity rates and the
high percentage of children in developing country populations, a
high proportion of the total illness (up to 50%) occurs in the younger
age groups. This means that a great deal of outpatient and inpatient
work at hospitals, health centres and health posts will be concerned
with children.
The most readily available source of data on serious morbidity is
hospital records, but, as indicated above, due caution is needed in __
interpreting such data. For instance, in malarious areas, malaria and
diarrhoeal disease alone often account for about a quarter of all
outpatient attentances. The ten most frequent causes of admission
to hospital are shown below for the United Republic of Tanzania.
This list can be considered fairly typical for a developing country. It
should be noted that:
• A fifth of all admissions are for pregnancy, its complications and
the delivery of babies.
• A fifth to a quarter of all admissions are for common diseases
such as malaria, pneumonia and diarrhoeas.
• Measles is frequently among the first ten causes of admission.
Table 4.1.
Hospital-reported morbidity for the United
Republic of Tanzania
Causes of hospital admission
1.
TEN CONDITIONS ARE
COMMONLY RESPONSIBLE
FOR ABOUT TWO-THIRDS
OF ALL ADMISSIONS TO
HOSPITALS
I
Deliveries, complications of pregnancy,
childbirth and puerpenum
% of total
admissions
22
2.
Malaria
3.
4.
Pneumonia
8
7
Diarrhoea
7
5.
Anaemia
6.
6
4
7.
Measles
Hookworm
8.
Bronchitis, emphysema and asthma
2
9.
External injury
2
10.
Ascariasis
1
Total
All other causes
62
38
3
38
Manual of Epidemiology for District Health Management
4.5 Mortality patterns
The main causes of mortality are different in developing and
developed countries, as shown in Figure 4.2. These pie charts show
that the infectious diseases and "other" causes, which include mal
nutrition, account for three-quarters of all deaths in developing
countries, whereas cancers, circulatory disorders and traumatic
injuries account for over half of the deaths in developed countries.
Figure 4.2. Distribution of the causes of death in developing and developed
countries
DEVELOPED COUNTRIES
DEVELOP?,G COUNTRIES
Canc
Circulatory
_ diseases
Circulatory
1 s ea s e s
Traunotic
.injury
\
\
Cancer
Infectious
diseasesi
Others
infectious
diseases
Traumatic
injury
In countries without an effective system for registering deaths and
certifying the causes, information on mortality patterns is largely
based on deaths occurring at health facilities, despite the recognition
that these deaths are a selected sample. Some countries have desig
nated sample population areas which are kept under surveillance for
vital events such as births and deaths. However, classifying deaths
that occur in the community by cause requires the details to be
gathered retrospectively from relatives. This is not an accurate way
to certify the cause of death, although "verbal autopsies" by doctors
can be helpful. Accurate information on deaths often has to be col
lected by special demographic and health interview surveys.
Mortality data are based on death registration (number and distri
bution) and death certiEcation (cause). In many developing countries
a large proportion of deaths are unregistered and death certification,
often of doubtful accuracy, is generally limited to patients who have
been admitted to hospital. Frequently, only about 10-20% of deaths
39
Epidemiological Health Information
are certified by cause, and many of these registered deaths will be of
urban middle-aged adults who were relatively well-off. There is
commonly an under-representation of rural people, infants, the
elderly and the poor. Despite these faults, however, such information
does provide some idea of the major causes of death. The informa
tion is rarely available on a district level and the DHMT will often
have to make its own estimates for the district population. A further
difficulty in using death certificates is the quality of the information
on the underlying cause of death, which is often poorly and inaccu
rately entered by medical officers (see Section 4.8).
!
I
i
I
I
I
I
IN MANY DEVELOPING
COUNTRIES 40-50% OF ALL
DEATHS OCCUR IN
CHILDREN
In some countries special samples in the national census are used
to provide age- and sex-specific death rates’/ sometimes by state or
region. These rates are likely to be far more accurate than those
based on death registration, but they usually provide no information
concerning the causes of death.
Table 4.2 shows a list of the ten most common causes of death in
hospital in the United Republic of Tanzania. This list may be consid
ered typical for developing countries. It is clear that:
• Pneumonia, measles and diarrhoea, particularly in children, account for about a quarter to a third of all hospital deaths.
• Chronic conditions such as malnutrition, tuberculosis, heart
disease and anaemia are also very important causes of death.
Table 4.2.
Common causes of death in hospitals in
the United Republic of Tanzania
Diseases
% of total deaths
1.
Pneumonia
16
2.
Measles
11
3.
4.
Diarrhoea
10
Conditions of early infancy
7
5.
Malnutrition
5
6.
Tuberculosis
7.
Tetanus
5
4
8.
9.
Heart disease
Malaria
4
4
10.
Anaemia
3
Total for 1-10
69
All other causes
31
Manual of Epidemiology for District Health Management
40
4.6 Seasonality
SEASONALITY AFFECTS
MORBIDITY,MORTALITY
AND USE OF HEALTH
SERVICES
In many developing countries health programmes are more diffi
cult to organize during certain months or seasons. However, what is
not so well appreciated is that disease frequency, health behaviour
and use of health services also show considerable seasonal variation
or seasonality.
Seasons are accompanied by marked climatic changes, all of
which affect household and agricultural activities, including the
availability of paid work. Seasonality particularly affects the urban
poor and the rural subsistence farmers, who are vulnerable because
of their low family income and the seasonal availability of food
supplies. Malnutrition is, therefore, frequently more common at
certain times of the year. The insect vectors of some communicable
diseases, such as the mosquitos transmitting dengue, filariasis and
malaria, begin to increase with the onset of the rainy seasons. Dur
ing the colder months people tend to crowd together, which pro
motes the spread of airborne and contact infections, such as respira
tory diseases and measles. An example of a seasonal variation in
mortality rate is shown in Figure 4.3.
Figure 4.3. Variation in the percentage of deaths
occurring per month, showing a peak
during the rainy season, largely
attributable to deaths from malaria
(based on data from Indial
I
I
I
y Auqust
Deaths
o
/
10
I
I
I
I
5
a.
co
□z
S
5
a
>—
I
\
\
\
- I
I i
Rainfall
I
300
:
o
200 2 2 I
UJ
100 >
<x o
0 a: -
JFMAMJJASONDJFM
MONTHS
In addition, people are less Likely to use the health services during
the difficult months, the supervision of community health workers
and health posts will be less effective and the availability of essential
drugs will probably be lower.
i
i
41
Epidemiological Health Information
It is very important, therefore, for the DHMT to analyse all the
available health information by month or season of the year and to
take seasonality into account when making district health plans.
Greater efforts may be needed during the "unhealthy" seasons.
4.7 Using morbidity and mortality rates
If reasonably reliable disease-specific incidence or prevalence rates
are known for the district, or national ones are available, it is pos
sible to calculate the approximate number of cases that the district
might expect in one year. This applies particularly to chronic dis
eases such as leprosy and tuberculosis, whereas acute diseases are
Hable to show too much seasonal or annual fluctuation in incidence
for these rates to be particularly useful at the district level. Although
the number of cases derived from known rates may not be very
accurate, the information is sufficient for most health planning,
management and evaluation purposes.
For instance, if the prevalence of pulmonary tuberculosis is stated
to be 5 per 1000, then there should be about 1000 cases on the regis
ter in a district population of 200 000, as follows:
No. of tuberculosis cases
=
prevalence rate x 200 000
5
1000
=
THE EXPECTED NUMBER
OF CASES CAN BE
CALCULATED USING
RELIABLE RATES
i
i
I
I
x 200 000
1000 per year
If there arc only 250 patients on the register, the health services
arc treating only about a quarter of all pulmonary tuberculosis cases
in the district. This then raises the question: How effective are the
services? The example also demonstrates why it may be dangerous
to rely on incidence or prevalence rates calculated from cases re
ported by the district's routine information systems.
Disease-specific mortality rates for the district population are
difficult to calculate with any reasonable degree of accuracy, but
in the planning of health programmes against chronic diseases,
national or other reliable rates can be used to calculate the number
of expected cases. A good knowledge of the expected total number
of deaths is useful for certain epidemic diseases, such as cholera,
typhoid, meningitis, typhus and African trypanosomiasis. A small
increase in the total number of cases or deaths may signify’ that an
undetected epidemic is already under way in the district.
A measure of disease severity is the case fatality rate, usually
expressed as a percentage, as follows;
Case fatality rate =
number of patients dying from a disease
number of cases of the disease diagnosed in same period
*
42
Manual of Epidemiol
CASE-FATALITY RATES
CAN MEASURE DISEASE
SEVERITY AND QUALITY OF
MEDICAL CARE
This rate can also be used as a measure of the quality of the treat
ment being given by the medical services. For instance, if a hospital
admits 200 cases of malaria in one year and 30 die, the hospital case
fatality rate is 15%, which might be considered too high. However,
interpretation of this figure must also take into account the mixture
of cases on admission, that is, how many were mildly ill and how
many severely ill.
It is also useful to calculate the fatality rate for cases arising in an
epidemic, as it gives a clear indication of the severity of the illness
and can point to which disease is causing the epidemic.
4.8 Death registration and certification
Death registration is the official recording of a person's death and
usually includes name, age and sex of the person, and the date of
death. In most countries it is a legal requirement, but it is often not
enforced and frequently not a responsibility of the health system.
Death certification records the cause of a person's death, as stated by
a doctor or another responsible health worker. Information from
death registrations can be useful in compiling mortality statistics
and in the surveillance of specific diseases, particularly in countries
where registration is reasonably complete. Analysis of known deaths
may also be useful in evaluating disease control activities and a rise
in certified deaths from a particular cause may indicate a serious
outbreak.
Information on mortality has the advantage that the data tend
to be more accurate than those for morbidity reporting (for such
elements as diagnosis and personal data), but it has the disadvantage
that only fatal cases are included.
In many developing countries, the data may be deficient because
certification of deaths is not done by a doctor and there is consider
able under-registration.
An outline of the medical certificate on the cause of death
(based on recommendations by the World Health Organization) is
shown opposite.
It should be noted that the certificate is in two parts. Part I shows
the sequence of events leading to death and proceeds backwards
from the most direct cause in 1(a). This does not mean the mode of
dying, e.g. heart failure. It means the actual disease, injury or compli
cation that immediately caused the death. Then come the underly
ing causes. In compiling statistics, deaths are classified by underly
ing cause, which is frequently not the same as the most direct cause.
A
43
Epidemiological Health Information
Figure 4.4. Medical certificate of the cause of death
Cause of death
Approximate time between
onset and death
I
Disease or condition directly
leading to death
(a)
due to (or as a consequence of)
Antecedent causes
Morbid conditions, if any,
giving rise to the above
cause stating the underlying
condition last
(b)
due to (or as aconsequence of)
(c).
II
Other significant conditions
contributing to the death, but
not related to the disease or
condition causing it
4.9 District health information checklist
• Review of the collection and analysis of basic data:
- symptoms and diseases included
- diagnostic criteria used
- reporting by which facilities
- reporting frequency and regularity
- efforts made to analyse data.
• Distribution and use of information:
- use by district health management team
- distribution of information within district
- feedback to health facilities
- reporting to regional and central authorities.
• Improvement of the routine information system:
- common and obvious faults
- effectiveness of reporting for important items
44
Manual of Epidemiology for District Health Management
- effects of seasonality
- more standardization of procedures
- inclusion of in-service training for primary health care workers.
• Presentation of epidemiological information:
- ten commonest diseases in outpatients and inpatients
- ten commonest causes of death in hospitals
- distribution of common and epidemic diseases throughout
district
- importance of other health problems, such as those of preg
nancy and delivery, child-spacing, nutrition, water supphes and
health behaviour.
• Frequency, distribution and importance of the following diseases
in the district:
- obstetric and perinatal problems such as:
prolonged labour, haemorrhage, retained placenta, puerperal
pyrexia, respiratory and feeding difficulties in newborn, neo
natal tetanus
- common diseases such as:
accidents and injuries, diarrhoea, intestinal helminths,
leprosy, malaria, malnutrition, measles and tuberculosis
- other locally endemic diseases such as:
filariasis, rabies, schistosomiasis, sexually transmitted
diseases, tetanus, trachoma and trypanosomiasis
1
- potentially epidemic diseases such as:
cholera, meningitis, plague, relapsing fever, typhoid, typhus
and whooping cough.
-|
45
Chapters
Reporting and Surveillance
Systems
j
f
5.1
I
5.1
Routine health information systems
5.2
Surveillance
5.3
Definition of cases
5.4
Sources of information
45
47
49
50
5.5
Additional sources of health information
53
5.6
Analysis and presentation of data
5.7
Communicating and using the information
5.8
5.9
Effectiveness of reporting or surveillance systems
54
56
56
57
District reporting systems checklist
Routine health information systems
Each country will have its own system of collecting routine
health information and reporting it from the periphery to the centre.
District health workers need to be familiar with this system and to
appreciate what is required to make it work efficiently. An outline of
such a system is shown in Figure 5.1.
Information is collected when people visit the health facilities and
the data are written down or recorded in various ways. The collected
data are then analysed and included in reports, which may be com
municated by the DHMT to health workers and to other district
organizations. The national ministry of health is responsible for
collating the information for the whole country.
Unfortunately the health information systems in many countries
are frequently characterized by elaborate forms filled in by clerical
workers in outpatient clinics and hospitals and sent for analysis at
central headquarters. This analysis is often carried out many months
or even years later, with no feedback to those collecting the original
data. The result is data that are of limited value to the DHMT.
For the most part, district staff may have little say in the number
or format of the forms to be used, but they can influence how well
they are used. It is important to keep in mind that information is
4fl
Manual of Epidem
Figure 5.1. Routine district health information or surveillance system
Detection of new cases by:
• health facilities
• community
• special means
Implement necessary changes
Collection of information
Improve health planning
Analysis and presentation
Reporting and dissemination
of information:
• Regional health office
•Ministry of health
needed by virtually everyone involved in health and that the effi
cient and timely communication of that information is essential.
Properly designed forms are necessary to accomplish this.
For the primary health care level, emphasis should be on collect
ing the minimum amount of necessary data in the simplest possible
way. An important principle is that any data recorded by health
workers should be useful to the workers themselves and to their
supervisors in their duties. The major purpose of collecting such
information should be to support the management and evaluation of
the health activities being carried out at the worker's level. The
DHMT is in a good position to make the best use of this health
information system and to keep these principles in mind.
Controlling the quality of the collected data is of vital importance
and checks will have to be made to ensure comphance with the
methods laid down. One way to do this is to make certain that the
forms are directly useful to the health workers and supervisors in
planning and evaluating their own district health services and pro
grammes.
Districts may participate in and use national reporting and sur
veillance systems, but they may also need to set up their own local
system. It is important to realize that for a system to be useful it
i I
47
Reporting and Surveillance Systems
does not have to detect all cases. Good estimates of incidence can be
invaluable for planning and evaluating health programmes and,
provided the proportion of cases detected remains reasonably con
stant, trends over time can be forecast. Incomplete data are certainly
much better than none, provided that the problems and defects in
the data are understood. This is well illustrated by the number of
poliomyehtis cases detected in Brazil before and after the onset of a
polio immunization campaign (see Figure 5.2). Even though all cases
were not detected the trend suggests very strongly that the incidence
of poliomyelitis has been markedly reduced since the start of the
campaigns.
Figure 52. Decline in the incidence of paralytic poliomyelitis in Brazil
following mass immunization campaigns
500n
Poliomyelitis by 4-week periods
Brazil, 1975-83
o
o
1
400 -
H-
! 3 300 1
<
u-
200
I
U.
100
i
o
z
0
£
____ £
1975
1976
1977
1978
1979
YEAR
5.2 Surveillance
The term surveillance is used in two rather different ways. First,
surveillance can mean the continuous scrutiny of the factors that
determine the occurrence and distribution of disease and other
conditions of ill health. It is thought to be essential for effective
control and prevention, and includes the collection, analysis, inter
pretation, and distribution of relevant data. Such a broad definition
almost equates surveillance with routine health information systems
and the two can therefore be considered together.
The second use of the term refers to a special reporting system
which is set up for a particularly important health problem or dis
ease, for example the spread of communicable diseases in a natural
disaster, nutritional status in a famine, or an epidemic. Such a sur
veillance system is often organized for a limited period and is closely
48
Manual of Epidemiology for District Health Management
integrated with the management of a health intervention pro
gramme. It is organized when information on incidence is urgently
needed but reliance cannot be placed on routine information sys
tems. This may be because the coverage of the population by the
routine reporting system is low, because the reporting is too slow, or
because the system itself is faulty and cannot quickly be improved.
Districts would be wise, at least initially, to limit a new surveil
lance activity to diseases and problems for which there is a continu
ing programme, for instance an expanded programme on immuniza
tion or a specific control programme for cholera or malaria.
Surveillance aims to provide quickly information which can be
analysed to determine frequency (usually incidence), and to answer
the questions: who? where? and when?
Epidemiological surveillance has a number of major uses:
• To identify outbreaks and epidemics and to ensure that effective
action to control the disease is being taken.
• To monitor the implementation and effectiveness of a specific
control programme by comparing the extent of the problem before
and after the implementation of the programme.
• To assist in the planning of health programmes by showing which
health and disease problems are significant and therefore worthy
of specific intervention. This also assists in deciding on priorities.
• To identify high-risk groups (e.g. by age and occupation), geo
graphical areas where the problem is common, and variations over
time (e.g. seasonal and year to year). This also assists in planning
programmes.
• To increase knowledge of vectors, animal reservoirs and the
modes and dynamics of transmission of communicable diseases.
I
Examples of events that may require surveillance are:
• Epidemic diseases, e.g. yellow fever, dengue, meningococcal
meningitis.
• Nutritional status and malnutrition.
• Animal reservoirs and vectors of communicable diseases.
• Environmental pollution, particularly of water.
• Demographic events, such as births and deaths.
Three diseases (cholera, plague and yellow fever) are covered by
the International Health Regulations, while a further five (epidemic
typhus, influenza, malaria, poliomyelitis, and relapsing fever) are
subject to global surveillance. Six diseases (filariasis, leishmaniasis,
leprosy, malaria, schistosomiasis and trypanosomiasis) are included
in the UNDP/World Bank/WHO Special Programme for Research
i
Reporting and Surveillance Systems
f
and Training in Tropical Diseases, and immunization against
six other diseases (diphtheria, measles, poliomyelitis, tetanus,
tuberculosis and whooping cough) is part of the WHO Expanded Pro
gramme on Immunization. Certain diseases will, of course, be of
particular regional or local interest.
5.3 Definition of cases
i
I
DEFINE THE CRITERIA FOR
REPORTING A CASE. AND
THEN TRAIN HEALTH
WORKERS TO APPLY THEM
I
The criteria for recording a particular patient or event as a case
must be clearly defined and practical. The criteria for reporting a
possible, probable and definite case must also be clear. With malaria,
for example, should any person with a fever be reported, or only a
young child with no other apparent cause of fever? Should a positive
laboratory diagnosis be required as well? The criteria chosen must
be realistic and easily understood. Some examples of the use of
symptoms and signs for reporting possible cases of certain diseases
by primary health care workers are given in Table 5.1. These are
only suggested criteria, which should be modified to suit different
situations.
Health workers need to be trained in how to apply such criteria; it
cannot be assumed that they will automatically do this properly. An
excellent way to train staff is to give each worker a case description
and then ask each of them separately whether the criteria are satis
fied or not and hence whether the case should be reported. There is
usually sufficient disagreement for a good discussion, which raises
awareness and clarifies the issues.
Reporting cases
The district office should receive information on each case in the
district through the reports. The details will vary according to local
circumstances and what follows can only serve as a guide.
I1
The essential information concerns the diagnosis and the fre
quency of the cases. In many situations further information may be
needed on any of the following details: name, age (even if approxi
mate), sex, address (at least name of village), occupation, vaccination
or treatment status (if applicable), date of onset and duration of
disease, death or not, place of infection (if known), source of infec
tion (if known), and names of people exposed to infection who may
heed to be followed up.
In certain circumstances, such as when the disease is covered by
quarantine regulations or is potentially epidemic, speed is needed
and the reporting person may have to inform the district office by
telephone or by messenger. Full information can follow later,
with reporting forms or reports sent by post, e.g. at weekly or
monthly intervals, although this will vary according to the local
circumstances.
4J
50
Manual of Epidemiology for District Health Management
Table 5.1.
Examples of diagnostic criteria, based on symptoms and signs, for
reporting of possible cases by community health workers
Possible diagnosis
Symptoms and signs
Measles
Fever with red rash, red eyes, disappearing within a week
Poliomyelitis
Fever with paralysis
Trachoma
Chronic inflammation of the eyes, leading to shrinkage and turning-in
of lids and blindness
Sudden and severe watery diarrhoea with massive and rapid
dehydration
Cholera
Leprosy
Chronic hypopigmented skin lesions, loss of sensation, thickening of
ear lobes, deformities of fingers, toes and face
Tuberculosis
Cough for 4 weeks or longer, loss of weight, bloody sputum, low fever,
night sweats
African trypanosomiasis
Fever, swollen glands in back of neck, lassitude, headache, sleepiness
Cutaneous leishmaniasis
Chronic, round, slowly healing ulcers, often on the face or exposed
parts of body
Malaria
Fever, rigors, headache, body aches and inability to carry out normal
daily activities
Lymphatic filariasis
Fever, painful groin swellings, inflamed streaks in legs, elephantiasis,
swollen genitals
Onchocerciasis
Itching of skin, nodules under skin, eye lesions, blindness
Schistosomiasis
IS. haematobium)
Blood m urine of schoolchildren and teenagers
Ascanasis
Roundworms expelled
Guinea worm infection
Pamfui legs, skin ulcers with worm protruding
Tapeworm infection
Segments expelled in faeces
5.4 Sources of information
Health facilities
The main sources of information are often morbidity reports from
health facilities including hospitals, health centres and clinics,
private practitioners, and traditional practitioners. Traditional practi
tioners should be encouraged to use diagnostic criteria as discussed
above. In these cases a confirmed diagnosis may not be possible
without further investigation and reliance may have to be placed on
criteria for possible and probable cases.
Ideally a senior person at each health facility should be respon
sible for periodically (e.g. each month) sending in information on
1
52
Manual of Epidemiology for District Health Management
The community
Community surveillance of a limited number of key diseases and
problems can be an integral part of primary health care by linking
community health workers, traditional birth attendants and o er
groups to primary level health facilities (see Figure 5.4). Such surveil
lance can also be useful to staff employed in special programmes,
such as malaria and family planning.
Simple and standardized diagnostic criteria need to be used and
the examples in Table 5.1 are a useful guide for simplified reporting
by the community. Many of these diseases are of course well recog
nized by local people, who give them names in their local languages.
For some selected problems or diseases members of the general
public may be asked to make reports. If so, efforts must be made to
collaborate with the community and secure its cooperation. Of
special importance in this regard will be schoolteachers, religious
leaders and government administrative officers. Radio broadcasts
may be used for publicity, especially in remote areas. Cases reported
by the public may also need to be checked by a health worker or a
laboratory.
Figure 5.4. Some ways of organizing community surveillance
School Teachers
Religious Leaders
womens Groups
village Leaders
Households
111 people
Conmunlty Health workers
Traditional Birth Attendants
Traditional Practitioners
Primary Healtn Care Services
District Health Office
t
Ministry of Health
|^n_/oo
08316
Cj C
J
A*'0* 2 /
53
Reporting and Surveillance Systems
5.5 Additional sources of health information
Special searches
1
Special searches can be made in schools, markets, particular
villages and high-risk households in order to improve case detection.
These searches should be done by health workers, or people who are
specially trained. In schools, teachers and children can be asked
whether or not they know of any cases. In both schools and markets
it is useful to show a picture of a person with the disease, with the
symptoms and signs described. When a worker is told of a possible
case, details of the person's name, age, sex and address and the infor
mant's name and address should be noted down. This information
can then be followed up and, if the diagnosis seems reasonable, full
information obtained.
o
Investigation of outbreaks
Outbreaks of disease will need to be investigated and all possible
cases identified. The investigation of an outbreak is described in
Chapter 6. In an outbreak it is important to organize active case
finding, if necessary by carrying out surveys in surrounding villages.
Details of all known cases in an outbreak should be recorded and
notified.
Surveys
Surveys can be very useful for the periodic surveillance of some
conditions. They can also be used to estimate the coverage of the
regular reporting or surveillance systems and to monitor changes in
completeness of reporting. (For an example, see Section 5.8.)
I
When the survey is analysed, it is important to keep in mind
that there may be both over-reporting and under-reporting. Over
reporting, because cases are falsely diagnosed, can be checked by
investigating reported cases and determining whether they have been
correctly diagnosed or not. However, under-reporting can only be
determined by checking whether the cases identified in a special
survey had been notified by the routine or surveillance system (using
identifying information such as names and addresses). Accuracy of
reporting (taking into account both over-reporting and under
reporting) is estimated by comparing the number of cases reported
with the number of cases estimated from the survey.
.
.
.
»■
.
In summary, the main sources of informationi are:
. • Health facilities.
• Special searches.
• Death certificates.
• Outbreak investigations.
• Laboratories.
• Surveys.
• The community.
I
54
Manual of Epidemiology for District Health Management
5.6 Analysis and presentation of data
The collected data need to be analysed and presented in a simple
but clear way. How this is done will depend on the local situation;
what follows can serve only as a guide.
A register is a book or file containing the recorded data. Each case
may be entered in one register, or a separate register may be used for
each condition. An example of a page from a register for one disease
is shown in Figure 5.5. The details, especially name and address,
should be checked against previous entries so as to avoid counting a
case more than once.
Figure 5.5. Register for recording details of reported cases of a particular
disease
Reporting period
Disease or problem
Serial no.
Date
Name
Age
Village
Sex
Death
or not
/
At the end of the reporting period (e.g. monthly) tables can bemade of the cases by who? (e.g. age and sex) and where? (e.g. village
of onset), as shown in Figures 5.6 and 5.7. The number of deaths can
be reported if required. Ideally, cases should be reported by date of
onset, but in practice it may only be practical to report by date of
diagnosis.
Figure 5.6. Suggested table for analysis of cases by age and sex
Reporting period
Disease or problem
Sex
Male
Female
Total
Age group (years)
0-4
5-14
15-44
45+
Total
55
Reporting and Surveillance Systems
i
Figure 5.7. Suggested table for analysis of cases by place of onset
4
Reporting period,
Disease or problem
Number of cases by place (e.g. village)
B
A
Totai
E
D
C
For a specified reporting period (e.g. one year) the district office
can indicate where the cases occurred by using pins in a map of the
district. This is called a spot map and it will readily indicate where
diseases are occurring most commonly. In particular, the clustering
of cases may indicate that a disease is more common in a particular
locality or that there is an outbreak.
I 5
ANALYSE ALL CASES BY
“WHO? WHERE? AND
WHEN?"
The district office should also keep a check on the incidence or
total number of new cases occurring each month. This is best done
by entering cases on a chart or graph at the end of each reporting
period (by month, for example, as in Figure 5.8). An analysis of inci
dence over time by who? factors, such as age, sex and occupation,
may help to identify high-risk groups. Time trends over longer peri
ods can be brought out by continuing the analysis over several years,
as in Figure 5.2. When interpreting time trends it is important to
consider seasonal variation and remember that when a new reporting
Figure 5.8. Analysis of hospital admissions for all diarrhoeas and cholera by
month {based on data from the International Centre for Diarrhoeal
Diseases Research, Bangladeshi
•—• All diarrhoeas
Cholera
|
l
GO
2:
1500-
o
GO
GO
L
sz 1000-
Q
C
-J
□>
GO
o
\
500-
X
\
cz
J
F
M
A
M
J
MONTH
J
A
s
0
N
D
Manual of Epidemiology for District Health Management
58
or surveillance system is implemented, the completeness of report
ing may increase, thereby giving a false impression of an increasing
frequency. Conversely, when a system has been m operation for
several months or years, the reporting tends to deteriorate and there
is likely to be under-reporting, or even falsification of reports Tius
usually calls for improved supervision and retraining of health staff.
In practice the total number of cases rather than a rate is usually
used This is sufficiently accurate providing the size of the popula
tion is reasonably stable. However, for the companson of different
districts or different population groups withm the district, it will
probably be necessary to calculate rates, using as the denominator
the actual or estimated size of the population at risk.
5.7 Communicating and using the information
The information from a reporting or surveillance system together
with what that information means, should be communicated to all
relevant people, in particular to:
. Regional and national staff, so that they are informed of local
situations and can compile information for larger areas.
• Primary health care workers involved in sending in the original
data and reports.
• Health workers involved in organizing community health pro
grammes, particularly all district health staff.
• Village councils and other local organizations.
• Nongovernmental and voluntary organizations.
WILL THE INFORMATION
BE USED? WHAT
DIFFERENCE WILL IT
MAKE?
• Local mass media, such as local radio stations.
However, the main reason for having such reporting and surveil
lance systems must be that the information is used to improve the
planning of health programmes and disease control activities If
information remains unused, organizing a reporting or surveillance
system can be a waste of staff, time and money.
5.8 Effectiveness of reporting or surveillance systems
Example: Suppose the DHMT is interested in finding out the
frequency of attacks of malaria in the district and the completeness
of the reporting or surveillance for this disease. The diagnostic
criteria for an attack are fever and chills which prevent people from
doing their normal work. (The same principles can be apphed to
other examples, such as the reporting of births or immunization
coverage.)
The population consisted of 50 villages with a total of 8500
houses and an average of 5.1 persons per household.
57
Reporting and Surveillance Systems
The sample consisted of 20 randomly selected villages. Within
each of the selected villages one house was chosen randomly as the
starting point, and then the next 19 houses in any one direction were
taken. This gave a cluster of 20 houses in each of the 20 villages,
making 400 houses or an estimated 2040 people in all.
Assessment of the incidence of malaria was done by paramedical
workers, previously trained by the health officer, who visited the
houses and asked the question:
“Over the last 4 weeks has any member of this household had an
attack of intermittent fever, with chills, preventing him or her from
doing normal activitiesi,f
(Alternatively, the local name for malaria could have been used if
a malaria attack is a well recognized condition.)
The total number of such cases was recorded.
Results: 30 persons were found to have had an attack of probable
malaria, starting in the previous 4 weeks, an incidence rate of nearly
15 attacks per 1000 people per month.
The estimated incidence of probable malaria per month for all
households is:
30
-------------------------- x
no. of houses in sample
total no. of houses
30_
400
x 8500 = 638 cases
From the records it was found that, over the same period of 4
weeks, 82 cases were reported by the surveillance system from the
8500 households.
The completeness of reporting is therefore:
cases reported
82
cases estimated from survey
638
x 100
13%
Conclusion: Only about one in seven of the probable cases of
malaria was being reported. Why is this? The DHMT now needs to
conduct an inquiry, starting by questioning cases at home and then
examining all steps in the information system in order to detect all
possible ways of correcting this under-reporting.
•
7
7
7.
5.9 District reporting systems checklist
• Diseases or health problems being reported:
- which cases, episodes or attendances?
- diagnostic criteria and working definitions being used
I
r
- estimates of under- and over-reporting.
58
Manual of Epidemiology for District Health Management
• Sources of health information:
- health facilities
- death registration
- laboratories
- the community
- special searches
- outbreak investigations
- surveys.
• Analysis and presentation:
- registers
- files
- monthly graphs
- spot maps
- special reports.
• Communication of findings:
- ministry of health and regions
- primary health care workers and district staff
- village councils and organizations
- nongovernmental and voluntary organizations
-
local mass media, radio.
• Use of information in health planning:
- coverage of reporting and surveillance system
- improvements to community health programmes
- improved district health plans
- use of information in community health education
- changes in district health status indicators.
59
Chapter 6
Controlling an Epidemic
Definition of an epidemic
59
6.2
Confirming the epidemic
60
6.3
Describing the epidemic
62
6.4
6.5
6.6
6.7
6.8
Case-control analysis
65
Environmental assessment
67
Control of epidemics
67
68
69
6.1
6.1
Reporting on the epidemic
District epidemic checklist
Definition of an epidemic
An epidemic is commonly defined as the occurrence in a commu
nity or area of cases of a disease that are clearly in excess of what is
expected.
Although epidemics of different diseases happen in different
ways, the district management team will need to follow a reasonably
systematic approach in order to avoid confusion. This chapter sug
gests an orderly sequence for investigating and controlling an epi
demic, but how the steps and procedures arc actually used will differ,
naturally, according to the disease and the local circumstances. The
approach is summarized diagrammatically in Figure 6.1, which
shows the importance of the two main components - investigation
and control.
Serious epidemics are uncommon and the DHMT will probably
only have to handle one every few years or so. However, the early
stages of an epidemic may be shown by the reporting or surveillance
systems and the DHMT is then required to bring the potential
epidemic under control. Most important epidemics are due to com
municable diseases with a short incubation period that are easily
transmitted. Food-borne diseases and cholera are good examples.
However, epidemicity is relative to the previous incidence of the
disease in the same area, among specified population groups and
at different seasons of the year. The appearance of two cases of
plague in an area may constitute an epidemic, whereas a high inci
dence of diarrhoeal diseases during the peak diarrhoeal season may
be considered the normal frequency. Measles and influenza are nvo
other diseases that can show marked seasonal and annual variations
in incidence.
BO
Manual of Epidemiology for District Health Management
Epidemics may commonly be due to:
• Food-borne outbreaks, e.g. enteritis due to Escherichia coh,
staphylococcal infection, salmonellosis.
• Communicable diseases with short incubation periods,
e.g. dengue, cholera, influenza, malaria, measles, plague,
yellow fever.
• Communicable diseases with longer incubation periods,
e.g. African trypanosomiasis, viral hepatitis, kala-azar.
• Toxic substances, e.g. contaminated foods, insecticides and
agricultural chemicals.
An epidemic may be detected by:
• Community leaders, such as politicians and teachers.
• Health workers in primary health care facilities.
• District health information and surveillance systems.
• Hospitals.
It is important to recognize a potential epidemic, and then to
determine the existence and size of the outbreak and to develop
ideas about the cause, method of transmission and best methods of
control. A judgement as to whether an epidemic exists or not can be
a highly political issue and it would be wise for district staff to
consult with their district and ministry of health colleagues before
making public announcements.
6.2 Confirming the epidemic
The main steps in the investigation and control of an epidemic are
shown in Figure 6.1. The initial step is to review the reported cases
in order to diagnose the problem. This can usually be done by analys
ing the clinical case histories and laboratory tests. Specimens, e.g. of
blood or faeces, may be taken and sent for further laboratory tests. If
laboratory facilities are few or non-existent, action should be taken
on the basis of a clinical diagnosis before the results of the tests are
available. It may be necessary to consult a more experienced health
worker or to request advice.
At this stage it is important to review the diagnostic criteria that
would be needed to differentiate non-cases from cases and to classify
the latter as possible, probable or defimte cases. This is particularly
important when cases are not easily diagnosed clinically and a dis
ease may be transmitted through subclinical or asymptomatic infec
tions. The best information about why the epidemic has occurred is
most likely to come from the analysis of the probable and definite
cases. It is also important to make these criteria clear before search
ing for other cases, even though the criteria may be modified later in
the light of further experience.
J
Cl
Controlling an Epidemic
Figure 6.1
Outline of the investigation and control of an epidemic
Review routine
A
information, surveillance,
clinical cases, community
information and reports }
WHAT ARE THE
DIAGNOSTIC CRITERIA FOR
A POSSIBLE, PROBABLE
AND DEFINITE CASE?
Information
regarding possible
epidemic
'Criteria for
establishing presence
of an epidemic
| Check records and
I seasonal incidence
—I
Is there an epidemic?
Disease control
component
Investigation
component
Isolate and
treat cases
Confirm diagnosis
—
Attack source and
transmission
Conduct case finding
i
i
Institute prevention
Trace contacts
Compile information
concerning epidemic
Conduct environmental
assessment
i
I
z
Continue surveillance
I
Process and
analys^ data
Communicate findings
I
i
Institute health plans to
prevent recurrence
Interviewing is a technique that requires skill, and health
workers need to help suspected cases feel at ease when giving details
of their illness. Interviewing cases also helps to identify contacts or
additional cases and to recognize special circumstances that might
help explain the outbreak. In order to standardize the interview pro
cedure, special case-history forms may be designed after the initial
cases have been thoroughly interviewed and examined.
62
Manual of Epidemiology for District Health Management
An epidemic can be confirmed by comparing the incidence of the
disease with that in the recent past or at a similar time in previous
years for the same community. Alternatively, an outbreak may be
confirmed if a number of cases are clustered and come from the same
place at a similar time. It should be noted that with some diseases,
such as yellow fever or cholera, only a few cases will need to be
investigated to confirm whether an epidemic exists and later a spe
cial search can be undertaken to find any umeported or unsuspected
cases. This extremely important step is called active case detection.
Some cases will be obvious but others may be mild and identified
only by such detailed inquiries.
Once the source of the outbreak is known, further cases may be
discovered by contact-tracing, that is, following up all the people in
contact with infectious cases or the same source of infection in the
outbreak. This is essential when all cases need to be treated or iso
lated. If the incubation period is a long one, contact-tracing can be a
difficult task.
6.3 Describing the epidemic
Information should be obtained on the age, sex, residence and
occupation of known cases, as well as the date and time of onset of
the illness and the whereabouts of cases during the period of incuba
tion of the disease. Other relevant questions need to be asked. For
example, malaria is usually transmitted by a night-biting mosquito
(Anopheles) and so residence is important, whereas dengue fever is
transmitted by a day-biting mosquito (Aedes) and so place of work
may be important. A knowledge of the disease concerned can help
greatly in focusing the investigation.
The basic questions about an epidemic that need answering are:
• What is the disease causing the outbreak?
• What is the source?
• What is the mode of transmission?
• How can the epidemic be explained?
WHAT DO ALL CASES HA VE
IN COMMON?
To find these answers it is important to analyse all the informa
tion on who? where? and when? contained in the case interviews to
see in what way all cases are similar. Early in the analysis, the actual
number of cases may be used, but later age- and sex-specific attack
rates will probably be needed.
Epidemic incidence curve (when})
A graph that plots cases of the disease by the time of onset of the
illness is called an epidemic incidence curve and it is an essential
part of the analysis. This graph can indicate the nature of the out
break and the probable source.
r
63
Controlling an Epidemic
A point-source or common-source outbreak is one where there
has been a simultaneous exposure of many susceptibles to a patho
genic agent resulting in a rise in the incidence of cases of the disease
over a short time, approximating to the incubation period of the
disease. This is an important clue. This type of outbreak is character
istic of water-borne diseases, such as cholera and typhoid, and foodbome diseases. A typical epidemic curve is shown in Figure 6.2.
The graph may be modified in epidemics where the point source
provides continuous exposure over a longer period of time—an
extended point-source outbreak. In this situation the onset will be
abrupt, but the incidence of cases will be spread over a greater period
of time than one incubation period.
Figure 62. Epidemic curve of a point-source outbreak
r
50CO
UJ
co
MO-
u_
30-
ex:
20-
Ig 10
1
2
3
M
5
6
7
DAY OF ONSET OF ILLNESS FOLLOWING EXPOSURE
INA POINT-SOURCE
EPIDEMIC MOST CASES
WILL OCCUR WITHIN
ONE INCUBATION
PERIOD
In the example in Figure 6.2 the exposure is assumed to have
taken place on day 0 and the incubation period therefore appears to
be between 2 and 6 days, with an average of 4 days. This epidemic
curve is very similar to that expected in a cholera outbreak.
(
Figure 6.3. Epidemic curve of a propagated epidemic
____
50]
GO
LU
CD
MO
JO-
ex
LU
CQ
20-
10-
T 1T 18' 20' 22 2M 26
DATE IN MONTH OF ONSET OF ILLNESS
28
30
In the example in Figure 6.3 the source case is unknown and
the first detected cases appeared on the 13th day of the month,
64
Manual of Epidemiology for District Health Management
■
with subsequent cases appearing at about 3-day intervals. This is
reasonably typical of a propagated epidemic due to person-to-person
transmission, such as occurs in shigellosis.
The shape of a propagated epidemic curve (as in Figure 6.3) will
depend on the incubation period of the disease and the suitability of
the environment for transmission. The longer the incubation period
the more spread out the cases will be. For contact-spread outbreaks,
the degree of crowding and intimacy of contact will determine
the rapidity with which the epidemic reaches a peak, while the
proportion of the population that is susceptible will influence the
extent of the outbreak. For vector-borne diseases, the time the organ
isms take to develop in the vector and conditions favouring the
development of the vector itself will also affect the shape of the
curve.
An epidemic curve can provide further information. If the organ
ism, and therefore its incubation period, is known, then the probable
time of exposure can be determined. For example, if Figure 6.2 repre
sents a cholera outbreak, it could be deduced that exposure occurred
between 2 and 5 days (the usual incubation period of cholera) before
days 3 and 4 when most cases occurred. Similarly, if the epidemic in
Figure 6.3 were due to a shigellosis outbreak, it could be estimated
that the first cases on the 13th day of the month had been exposed to
a source case about 3 days previously (the usual incubation period of
shigellosis). This use of a known incubation period helps to trace the
source of the infection.
Conversely, if the time of exposure is known, the incubation
period can be calculated and this is a clue to the causative organism.
This applies particularly to food-borne disease outbreaks where the
time of exposure is usually known. For example, if the epidemic in
Figure 6.2 was a food-poisoning outbreak and the meal had been
taken at midday on day 2, the median incubation period could be
calculated as around 24 hours, which would indicate salmonellae
rather than staphylococci as the probable infecting orgamsm. It
should be noted here that incubation periods are best expressed in
terms of the median and range. The mean or average is not used
because there may be some extremely long or short incubation
periods in many outbreaks, which tend to pull the mean away from
the central clustering of cases.
Analysis of cases by who! and where!
Analysis of the cases by personal factors such as age, sex and
occupation may also give clues to the source of the infection. For
example, if the initial cases are mainly in children, the source may
be in the vicinity of their school, whereas if adult men are affected,
the source may lie at work, in the fields, or elsewhere. Again it may
be necessary to calculate rates in order to compare attack rates in dif
ferent population groups.
i
85
Controlling an Epidemic
Figure 6.4. District spot map showing clusters of epidemic cases
TOWN
Marking known cases on a "spot map" (see Figure 6.4) may indi
cate a possible source of infection. Often only known cases are
marked on the map, as in this example. However, it may also be
important to know the distribution of the general population. For
example, if 70% of the population in this district live in the town,
then the apparent cluster in town is actually a relative deficiency,
indicating that the disease is mainly rural.
When all cases occurring within a short time period (e.g. 2-3 days
for shigellosis, one week for cholera and two weeks for typhoid) are
marked on the map in one colour, and other cases beyond that time
in different colours, the spread of the epidemic may become appar
ent. The clustering of cases together in one area and at about the
same time, also called space-time clustering, indicates a localized
epidemic.
6.4 Case-control analysis
Descriptive analysis by who? where? and when? may provide
sufficient information about the source of the outbreak for appropri
ate control action to be taken immediately. In other outbreaks,
however, further analysis may be necessary.
A case-control study is often used for this analysis. Patients are
questioned to determine what contacts they have had with possible
sources, as suggested by the incubation period. Exactly the same
questions are then asked of a group of control people who live in the
same area but who do not have the disease. There should be at least
one control person for every case. A useful way of selecting controls
is to interview a person of the same sex and age (within 5 years) as
the case who lives in a neighbouring household in which there are
no known cases. Try to avoid asking leading questions in the inter
view,- however, since people may have difficulty in remembering
what they previously ate, for example, a checklist approach is often
helpful. If possible, interviewers should not know the explanation for
the epidemic so as to avoid the recording of biased answers.
I
87
Controlling an Epidemic
6.5 Environmental assessment
A systematic analysis of the data may indicate an environmental
source for the outbreak. This can be confirmed by obtaining samples
of suspect food or water for examination in a laboratory (if facilities
are available] for toxic chemicals or faecal contamination. Breeding
sites for disease vectors may also need to be investigated. The help of
a local health inspector who knows the area may be useful in such
investigations.
6.6 Control of epidemics
When the causative organism, its source and the route of trans
mission are known it will probably be easy to explain why the epi
demic occurred. Control measures depend on the individual disease
concerned. The main strategies for the control of communicable
diseases can be summarized under three headings, as in Table 6.2.
Table 6.2.
Main strategies for the control of an epidemic due to a
communicable disease
Attack source
Interrupt transmission
Protect susceptible
people
Treatment of cases and carriers
Environmental hygiene
Immunization
Isolation of cases
Personal hygiene
Chemoprophylaxis
Surveillance of suspects
Personal protection
Control of animal reservoirs
Vector control
Disinfection and sterilization
Notification of cases
Restrict population movements
Better nutrition
Primary prevention is achieved by all the measures listed under
"interrupt transmission" and under "protect susceptible people,"
together with control of animal reservoirs. If all these are properly
performed the number of new cases should be greatly reduced. Thus
clean water supphes and the correct disposal of faeces could prevent
the spread of cholera, control of anophehne mosquitos could reduce
malaria transmission, and immunization could protect young chil
dren against measles.
Secondary prevention can be achieved by finding subclinical cases
and carriers and by contact-tracing and surveillance.
!
i
1
Tertiary prevention is by the treatment of cases or carriers so that
they do not spread the organism any further.
88
Manual of Epidemiology for District Health Management
Case-control study techniques involve:
. Interviewing both cases and controls using exactly the same
questionnaire to identify possible sources of infection.
. Analysing data from cases and controls to find the percentage of
each group that had contact with each of the possible sources.
• Looking for any significant differences.
Example: Three sources of drinking-water were considered the
possible origin of a cholera outbreak. By questioning the cases and
similar number of controls about the water they had Jeen
2 to 3 days before the outbreak, the ^v^tlgators/ne^t^e6te1rXre
of
incubation, period.
Table 6.1.
Case-control study of 18 cholera cases by
water source used
Wafer source used
Cases
Controls
Total
7"
B
c
18
17
16
6
18
14
3
17
It can be seen from this table that if only the 18 cases had been
Questioned the likely source of infection could have been either
water source A or B. The addition of controls shows that far fewer of
them had taken water from B and thus this source is implicated
Sometimes it may be necessary to perform tests of statistical signifi
cance if the differences are not clear-cut. It may be asked w en
looking at the table, how it was that there were 2 cases of cholera
who hfd not taken water from source B. In such an investrgati ,
there are always some errors or people who give an incorrect his ry_
Likewise lhe 3 controls who said they had taken water from sour B
mav^ fact not have done so or they may have taken the water but
did not succumb to the disease. It is the difference between the cases
I
.(
i
and controls that is important.
Another useful method, particularly for food-borne epidemics,
is to compare the attack rates for illness among
those having
eaten pScular foods. In this case the attack rate should be high
among the exposed population, but it will be lower for e
unexposed population.
88
Manual of Epidemiology for District Health Management
The main elements in the control of an epidemic are, therefore, as
follows:
Attack source and mode of transmission. Contaminated water
should be prohibited or sterilized, infected food destroyed, and
vector breeding sites dealt with. Health education has a large part
to play in this work and even legislation may be necessary.
Treat and isolate all cases. The treatment given will vary with the
disease and the facilities and supplies available.
Increase resistance of local population. Some communicable dis
eases can be prevented by chemoprophylaxis (for example malaria)
or immunization (for example, poliomyehtis and measles). It
should be borne in mind that in epidemics of some diseases, such
as typhoid and cholera, vaccination is relatively ineffective.
Continue surveillance. During the acute phase of the outbreak, it is
necessary to keep suspects at special risk under observation. Once
the epidemic is under control, surveillance for new cases should
be carried out to ensure that the control measures have been
effective. The routine reporting system may not be adequate to
show this and special surveillance may be needed. Community
surveillance may then be an important means of recognizing and
reporting any new cases.
6.7 Reporting on the epidemic
The DHMT should report early on a possible epidemic to col
leagues in the ministry of health and to other district officers so that
the disease control authorities can institute their own procedures. A
brief report on the epidemic should be written and should include
recommendations for measures to prevent any similar outbreaks in
the future.
The report should cover the following points:
• Causative organism and probable routes of transmission.
• Description of the epidemic curve, the geographical distribution
and main features of the cases.
• Explanation of the reason for the epidemic.
• Disease control measures that were introduced.
• Recommendations for improvements to prevent the epidemic
occurring again.
Copies of the report should be distributed to the regional health
officer, ministry of health, senior district government officers, health
workers in charge of district health facilities, and community leaders
and other local organizations.
i
I
6.8 District epidemic checklist
• Collect information to answer the question: Is there an epidemic?
I
i
I
- review cases for probable diseases and define diagnostic criteria
for possible, probable and definite cases
- check health information system for cases
- search for missed cases
- review previous levels of endemicity and local knowledge.
• Describe the epidemic:
- when? epidemic incidence curve
- where? mapping of cases
- who? characteristics of cases
- collect information on population at risk to establish denomi
nators.
• Answer question: What caused this epidemic?
- causative agent
- source and transmission
- exposure
- susceptibles and high-risk groups
- use case-control method to test explanation
- collect additional specimens for laboratory investigation.
• Institute control measures for the particular disease:
- attack source
I
ih^.
- interrupt transmission
- protect susceptible people
- notify authorities
- write and distribute report.
70
Manual of Epidemiology for District Health Management
I
I
I
I'
I
I
1
I
*
. 5
J
1 .
_71
T-fi-
C H A PIE R 7
LIBRARY
Epidemiological Surveys ’
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
7.9
7.10
7.11
7.12
7.13
7.1
Uses of surveys
Cross-sectional and longitudinal surveys
Survey objectives
Selecting the sample
Sample size
Response rate
Accuracy of measurements
Questionnaires
Variables
Repeatability
Validity
Ethical issues
District survey methods checklist
71
74
75
76
78
78
79
79
81
83
83
84
86
Uses of surveys
Surveys offer a very useful way of collecting additional informa
tion that is not available from the routine health information or
surveillance systems - a common situation in many developing
countries. However, as surveys consume staff, time, and money, the
DHMT should be convinced that the required information is not
available froQi reports or other ministries and that a survey would be
the most desirable way to obtain it.
The DHMT will encounter many situations in which it may be
desirable to undenake an epidemiological survey. In some countries
the most frequent reason for such investigations will be to form part
of a nationally organized survey undertaken, for example, to find out
the prevalence of leprosy, blindness or some other chronic condition
or to assess the effectiveness of a particular health programme. The
DHMT will not usually be directly involved in the planning and
preparation of the protocol, but may well need to implement it
locally and should understand why the survey is being done and the
reasons for each step. For example, Figure 7.1 shows how district
health staff might become involved in a national nutrition survey
and what might be expected from them.
72
Manual of Epidemiology for District Health Management
Figure 7.1. Possible district involvement in a national nutrition survey
Ministry of health
Medical research
institute
Nationa nutrition
survey headquarters
International
agency
Regional health
authorities
District health
management team
I
Nutrition study team
Medical
studies
Environmental
studies
J
Support team
I------------
I
F
I
Laboratory
studies
Administration
and staffing
Logistics and
supplies
There will also be occasions when the DHMT wishes to organize
an epidemiological survey to estimate, for example, immunization
coverage. Other examples might be to find out why one group of
householders is not cooperating with the antimalarial spray team or
to screen schoolchildren for a communicable disease such as leprosy
or schistosomiasis. It is a good principle always to offer some health
service to the people being surveyed - no survey without service!
Surveys are commonly made to:
• Estimate the incidence or prevalence of important diseases, e.g.
leprosy, malaria, malnutrition, schistosomiasis.
I
I
i
• Screen population groups for treatment of important diseases, e.g.
pregnant mothers, young children, schoolchildren, plantation or
factory workers.
• Provide health information about households and their members,
e.g. excreta disposal, water supphes, food habits.
• Find out about local beliefs, customs and health behaviour, e.g.
use of local foods, breast- and bottle-feeding, smoking.
• Evaluate how effective the health services are, e.g. antenatal
attendance, immunization coverage, utihzation of outpatient
clinics.
There are five main stages involved in an epidemiological
investigation:
• Clarification of the need for the survey and statement of
objectives.
• Determination of the sample and methods.
• Organization and implementation of the survey.
J
73
Epidemiological Surveys
• Analysis, interpretation and presentation of findings and
recommendations.
• Use of findings in health planning and disease control.
These are illustrated in more detail in Figure 7.2.
Figure 12. Outline of the main steps in an epidemiological survey
Identify problem for investigation J
I
I
I
Determine priority of problem
,
Formulate hypothesis
[Determine objectives and methods
Survey aspects:
• Survey design
• Sampling
• Observations
/Organizational aspects
• Transport
• Clearance
• Schedules
• Manpower
• Communications
• Materials
• Accommodation
l • Finance
• Measurements
• Instruments
• Training
♦
Conduct survey
I
I
I
I
T~~
Collect data
c
Analyse and interpret data
Write report
Communicate findings
Incorporate information
into health planning
I
Evaluate effectiveness of
new health programme
f1
'X
7.2 Cross-sectional and longitudinal surveys
CROSS-SECTIONAL
SURVEYS COLLECT MAINLY
ALENCEDATA
Cross-sectional surveys examine people at one point in time and
therefore provide prevalence data, particularly for infections or
conditions that last a relatively long time such as malnutrition and
African trypanosomiasis. The "point in time" may last up to several
weeks or months, provided that the incidence of the condition being
investigated does not change much during these weeks. However, for
a disease or event that shows ^considerable seasonal fluctuation, a
careful
care^l choice should be made of
ot the best time of year to undertake
survey.
Longitudinal surveys collect information about all the new cases
----------------------------- or events occurring over a period of time and therefore supply inci
dence data, e.g. new cases of measles or tuberculosis, or the number
of pregnant mothers attending the antenatal clinic for the first time.
The time period for collecting the incidence data must be defined.
Longitudinal surveys may be organized in a similar way to the sur
veillance of communicable diseases.
Senior district health staff should be able to organize their own
cross-sectional surveys for common problems, but longitudinal
surveys are more complicated and for these they should seek further
advice. Some incidence data can be collected during cross-sectional
surveys by asking people about events in the previous two weeks. If
repeated cross-sectional surveys are carried out on the same popula
tion and all new cases occurnng after the first survey but before the
second one are found, then an estimate of the incidence of the dis
ease or event can be calculated. However, this is possible only for
chronic diseases that do not come and go between the two surveys.
Cross-sectional surveys are rarely repeated within one year. A repeat
survey should be carried out in the same month as the initial survey
to avoid seasonal changes that might affect the frequency of the
disease or event.
The main points about cross-sectional surveys are that they:
• Mainly provide information on prevalence.
• Are useful for descriptive information, screening and estimation
of use of services.
• Are not useful for rare diseases or events.
• Are not useful for diseases or events of short duration.
• Are fairly quick and easy to organize
• Give prevalence data that may be difficult to interpret.
• Can give an estimate of incidence if two cross-sectional surveys
are carried out.
I
1
75
Epidemiological Surveys
7.3 Survey objectives
At an early stage it is important for the DHMT to be clear about
precisely what questions are being asked and then to decide if a
survey can indeed provide the necessary answers. The question
should be simply and clearly stated, for example:
What is the prevalence of malnutrition among villagers in
this district!
I
The objective might be to determine the prevalence of mild,
moderate and severe malnutrition among children aged 12-36
months living in the district.
The objectives will determine the population to be surveyed.
Criteria will have to be stated for mild, moderate and severe malnu
trition. Lack of clarity in the precise question and the formulation of
objectives is a very common fault in surveys.
I
I
At this time, the major end-results expected from the analysis
should be worked out in the form of "dummy tables/' For example,
suppose it is planned to carry out a survey to relate haemoglobin
level to the presence of hookworm infection, according to the age
and sex of the patient. The preparation at the outset of the dummy
tables shown in Figure 7.3 would help to clarify the objectives. These
tables show the key relationships that are expected to emerge from
the investigation.
Figure 7.3. Sample dummy tables
The distribution of hookworm-infected subjects by haemoglobin level.
i
Haemoglobin level
Hookworm infection
present
Less than 10 g per 100 ml
10 g per 100 ml or more
Total
% with anaemia
absent
Total
% with hookworm
78
Manual of Epidemiology for District Health Management
Figure 7.3. Sample dummy tables (continued)
Distribution of the villagers by age, sex and haemoglobin levels
Total
Age group (years)
Haemoglobin level
(g/IOOml)
0-4
5-14
15-45
45+
M F
M F
M F
M F
M F
6.0
7.0
8.0
9.0
10.0
11.0
12.0
13.0
Total
7A Selecting the sample
EACH INDIVIDUAL SHOULD
HAVE AN EQUAL CHANCE
OF BEING IN THE SAMPLE
All the people being investigated are called the reference popula
tion, but it is unusual to investigate each and every one of them. It is
more common to select a sample, called the study population, in
such a way that each person in the reference population has an equal
chance of being included in the study. In this way selection bias is
avoided and the study population will probably be representative of
the reference population. Incorrect or poor sampling is another
common fault in surveys.
I
i
i
Studying the entire population can involve too much time, man
power and money. Moreover, the very size of such a large study
might introduce errors. In some instances, however, examination of
the entire population is unavoidable, for instance when information
is required on all cases during an epidemic or when selection of a
group of people for the study would create a strong feeling of dis
crimination in the population.
In organizing a survey it is important to define which people
are to be surveyed: for example, which sex, age group, occupation
and area. This helps in selecting the sample and keeping the amount
of survey work down to a minimiim. There are two main methods
of drawing a sample—random and systematic. For statistical
reasons, randomsampling is more likely to provide a study popula
tion that is representative but systematic sampling may be easier to
do in practice.
I
Epidemiological Surveys
The following are the main steps for drawing a simple random
sample:
• Decide what is to be the unit of sampling. This may be a person,
household or village, as appropriate.
• Make a list of all available sampling units from which the sample
is to be drawn. This is called the sampling frame.
• Decide the number of units that need to be randomly chosen from
this sampling frame. This is called the sample size, and this num
ber divided by the total number of available units in the sampling
frame is called the sampling fraction.
• Pick the required number of units from the sampling frame by a
method which ensures that all units have an equal chance of
being picked, i.e. randomly. This can be done either by drawing
lots or by using a table of random numbers (see Appendix 3,
page 177).
A systematic sample may be chosen as follows:
• The first unit is chosen randomly.
• Then choose the next units in a systematic manner, e.g. every
fifth person on a list, every third hospital admission or every
tenth house on a street.
• Often different random starting units are selected to give several
clusters.
CLUSTER SAMPLES ARE
COMMONLY USED FOR
CONVENIENCE
It is often not possible to obtain a complete sampling frame of
individual people. Such detailed information is often not available
and to obtain it would be costly and time-consuming. One solution
is to use randomly selected villages or households in place of indi
viduals. A cluster sample is recommended for collecting information
on a variable that is common; one way of obtaining such a sample
is by the "30 clusters of 7" technique. Thirty individual villages—
a cluster of households—are chosen randomly and then within
each cluster seven households or people are again chosen randomly.
This technique was originally developed for estimating immuniza
tion coverage, but has now been widely used for descriptive crosssectional surveys. This sampling method is suitable for use when a
relatively common condition is being investigated, but it does not
give sufficiently accurate estimates for rare conditions. It is also
not a suitable sample to use to measure changes in health status
(see Sections 2.8, 2.9 and 13.3).
Cluster samples have several advantages:
• Only a simple sample frame is needed, e.g. number of villages.
• It is easier and faster to do the survey because people are grouped
together.
• It is often more acceptable to the local community.
77
78
I
Manual of Epidemiology for District Health Management
7.5
Sample size
In general, the larger the sample the more reliable will be the
estimate of prevalence obtained. Table 7.1 provides a guide to the
sample sizes required for different prevalence rates. For instance, if
the expected prevalence rate of a particular condition is about 40%,
then the prevalence for a random sample of 50 people will probably
be between 26% and 55%. If we examine a sample of 200 subjects
this range falls to 33%-47%. Clearly there is a significant gain in
accuracy in going from 50 to 200, but not a great deal more accuracy
C_is obtained by examining a sample of 500 subjects. For many pur
poses, a sample of 100-200 subjects for a common condition will be
sufficient. However, when greater accuracy is required or conditions
with a low prevalence are being investigated, a much larger sample
size will be needed. For a more detailed discussion on estimating
sample size see Appendix 2, page 175.
Table 7.1.
Relationship between expected prevalence and range of
prevalence estimate determined by survey, according
to sample size
Number of people in sample
Previous estimates
of prevalence
%
50
200
100
1000
500
Probable range for prevalence estimates
7.6
1
0
5
0.1
4
0.3
3
0.5
5
2
11
2
9
3
8
4
18
6
15
7
13
8 12
16
24
18 23
10
3 22
5
20
10 34
13 29
2
7
30
18 45
21
40
15 26
24 37
26
35
27 33
40
50
60
70
80
90
26 55
30 50
33 47
35
45
37 43
36 64
40 60
50 70
60 79
43 57
45
55
47 53
53 67
55
65
57 63
65
76
82 95
85 94
87
74
84
93
67 73
87
63 76
74 85
45 74
55 82
66 90
78 97
71
77 82
88 92
Response rate
Even if samples are well chosen, surveys can still give misleading
results if a high proportion of the households or individuals are not
contacted or they do not answer the questions. This is called the
non-response rate. A bias can be introduced by selecting those who
are seen and by missing out those who are not seen. For instance, a
village survey in a rural area done during the day may miss the
young men and women who are working in the fields or paddies.
5
Epidemiological Surveys
FOLLOW UP ALL NON
RESPONDERS AT LEAST
ONCE
In leprosy surveys, leprosy patients may be deliberately elusive and
not attend, thus giving a low prevalence. Conversely, people may
only attend a survey if they think there is something to be gained, as
in malnutrition surveys leading to free food supplements. Those who
are not seen may be as important as those who are seen. In surveys
of common conditions this non-response may not be as critical as it
is with some of the rare diseases. The problems of poor sampling and
poor response rates apply equally to all surveys.
To reduce the non-response rate it is necessary to:
• See at least 80% of the original sample.
• Follow up all non-responders at least once.
7.7 Accuracy of measurements
Measurements can easily be inaccurate. This is commonly be
cause the survey worker makes a faulty measurement, and not
because of faulty instruments or unreliable subjects. This is known
as observer error. There may, however, be instrument errors if they
are not checked regularly, e.g. the adjustment of the zero reading on
weighing scales. Another common source of error is faulty recording
of information on the record forms or questionnaires.
Inaccuracies can be reduced by:
• Thoroughly training all staff and checking frequently to see that
the methods are being carried out correctly.
• Following standard and agreed guidelines, e.g. how to weigh an
infant; how to ask questions in a questionnaire.
OBSERVERS ANO
INTERVIEWERS ARE THE
MAIN SOURCE OF
INACCURACIES AND
ERRORS
L
• Using "blind" methods wherever possible. This means that the
subject and/or the observer does not know important items of in
formation that might encourage them to bias their answers or
their techniques, e.g. the main purpose of the survey or whether a
child is thought to be malnourished or not.
• Survey workers should sign their name against any case history,
physical examination, measurement or laboratory test, so that it
is clear who did it. This encourages more accurate work and
makes the checking of records much easier.
• Checking all instruments at least once per day using a known
standard, e.g. infant weighing scales can be checked against a 10
kg weight.
7.8 Questionnaires
Questionnaires may look simple but in fact they are surprisingly
difficult to design. They are used for collecting information, usually
with an interviewer, about such items as what people have been
doing in their recent past, what foods they eat, whether they have
any illnesses, which people have died recently, and where they go to
get medical help. Such information would be difficult or impossible
79
80
Manual of Epidemiology for District Health Management
Figure 7.4. An interviewer administering a questionnaire
to a family during a household survey
to get in any other way. For instance, it is easier to ask people where
they get their domestic water than to observe them to find out.
However, it must be remembered that such information is what
people say they do, which might be quite different from what they
actually do.
The following are some common problems with questionnaires:
• Poor questions, which are unclear, badly worded or which really
contain more than one question. Each question must be simple,
clear and non-threatening.
• Leading questions can influence responses. Questions should not
suggest that a particular answer is correct.
■i
• Sensitive and personal questions may produce evasive answers.
Ask general questions first, then go on to more sensitive issues
later.
• People may not remember events that occurred long ago. In gen
eral, a recall period of two weeks is the maximum that can be
relied upon (except for major events such as death or admission to
hospital).
• Interviewers left too free to interpret the subject's answers. Record
directly what the subject says or use pre-coded answers.
• Too many questions, so that subjects and interviewers get bored.
After basic questions on name, age, sex, etc., a further 10-15
Epidemiological Surveys
r
questions allow for an interview lasting about 15-20 minutes,
which means that three to four people can be interviewed in
one hour.
• Use of inappropriate and non-standardized translations of ques
tions into local languages. Many problems can arise when a
questionnaire is constructed in one language but the questions
are actually asked in a second language. This can be overcome by
getting one group of people to translate the draft questionnaire
and then for another group to translate it back again. This reveals
discrepancies that arise during the translation, which can then
be remedied.
• Subjects are usually worried about why the questions are being
asked and who will be given the information. A reasonable and
full explanation should be given by all interviewers, followed bv s
strong assurance as to the confidentiality of the respondent's
name and the subsequent use of the information. People may be
particularly worried about their name being passed on to other
authorities.
PILOT-TEST ALL
QUESTIONNAIRES
THOROUGHLY TRAIN
AND SUPERVISE ALL
INTERVIEWERS
After being designed, it is vital for a questionnaire to be pilottested using interviewers and a small group of people for whom i: is
intended. This is essential to correct any mistakes or ambiguities
and to train all interviewers in the new agreed version. Role-plavizg
by the interviewers and interviews done under the critical eye or
colleagues are two very useful means of obtaining a standardized
technique by all the interviewers.
Interviewers need to be aware of how they can influence the
answers to questions, which should be asked in a neutral and non
threatening way and without any indication that certain answers are
"correct." The interviewer should, therefore, not show agreement ?r
disagreement or distaste or pleasure with the replies. This abilitv
only comes with careful training. Health workers often make poet
interviewers since they have difficulty in remaining neutral and m
refraining from giving advice. For this reason, it may be better to
recruit interviewers from among other groups of educated people
such as teachers, extension workers or secondary school children
rather than the district health staff. In many cultures it is very im
portant, especially for sensitive subjects like family planning and
childbirth practices, for women to interview women and men to
interview men. The interviewer's sex can markedly influence the
respondent's replies to questionnaires.
7.9 Variables
Choice of variables
Variables are characteristics that are measured either numeric;/ly
(e.g. age) or in categories (e.g. absence or presence of disease!. The
observations for each variable must be reproducible for them to be
useful and considerable time should be spent during the planning
stage of a survey to ensure that standardized methods are used in
measuring and categorizing all variables.
81
82
Manual of Epidemiology for District Health Management
Which variables should be included in a study? Obviously the
selection should be based upon their relevance. If a variable is of no
use in the subsequent analysis, then it should not be included. Be
quite clear what information you hope to obtain before including a
particular variable, because considerable effort and manpower could
be needed to collect and process the extra data. Each item to be
included in your survey must be justified and made to "pay its own
way". When considering which variables to include in the study, the
answer is as many as necessary7 but as few as possible!
On the other hand it is a mistake to Emit the number of variables
so that it is impossible to analyse certain aspects of the survey be
cause relevant information was not collected in the first instance. As
a routine, therefore, you should first produce a comprehensive list of
all the variables that you think would be required for the study and
then go through each item, in detail, to see how the information you
wish to collect for that particular variable may be used in your ulti
mate data analysis. At this stage it is useful to draw up dummy
tables that will be required in the analysis (see Section 7.3).
Measuring variables
Once the variables have been chosen, the next step is to plan how
to measure them under field conditions.
There are two requirements tor every7 variable:
• A good definition.
• A method of measuring it.
Illness has different meanings for different people. For example,
what you call a "common cold" may be interpreted as "influenza"
by someone else. Such differences in perception can lead to situ
ations where the measurement of variables by different people will
produce different results, i.e. the findings are not repeatable.
It is necessary, therefore, to define all variables clearly and by a
method that permits them to be objectively measured. For example,
malaria could be defined as the presence of Plasmodium parasites
in the blood stream of a patient as identified from a single thick
blood film, or as a child with splenomegaly, or as a fever with chills,
or a combination of these. This is the operational definition. In
formulating the operational detinition of variables one must always
keep in mind that only simple and limited, standardized techniques
can be applied on a mass scale. More detailed examination tech
niques, such as those that are available in hospitals, are often not
practical. It has to be accepted that such simplified techniques might
miss a small percentage of cases or include non-cases, but it is just as
important to ensure that your findings are repeatable.
In choosing the methods by which the variables can be measured,
two facets should be considered. These are the precision or repeata
bility and the validity of the measurement.
i
Epidemiological Surveys
7.10 Repeatability
Even the simplest measurements have their errors, sometimes to
a surprising degree. The variations caused by experimental errors,
including those related to test performance, determine the repeata
bility of the measurement, whereas those that are inherent in the
method itself determine its validity. Thus it is possible to have a
method which gives highly repeatable results that are not valid.
The repeatability of a measure is its ability to reproduce consis
tently the same information when repeated examinations of the
same population are made.
The more reliable the method, the more repeatable the data are
likely to be. If the variability in a method leads to random fluctua
tions in values above and below the true mean value, a relationship
that actually exists may be missed, but false conclusions about a
relationship will not occur. On the other hand if there is a consistent
over- or underestimate of the true value, called a bias, faulty conclu
sions are likely since readings will be consistently lower or consis• tently higher than they should be.
The repeatability of a measurement can be affected by:
• Observer variation. This can occur in observations made by a
single person (intra-observer variation) or by different people
(inter-observer variation). An example of this is the variation in
the ability of technicians to determine the presence of malaria
parasites in the same blood slides.
• Subject variation. For example, the response to a question may be
affected by the subject's motivations and beliefs and the place of
the interview.
• Instrument and method variation. Some are obviously more
reliable than others.
7.11 Validity
Validity refers to the extent to which the test is capable of cor
rectly diagnosing the presence or absence of the condition or disease
concerned.
The strict definition of a case of the disease or event under study
is of utmost importance in obtaining a high validity because words
may have different meanings for different people. An accurate diag
nosis is as important to epidemiologists as it is to clinicians. But the
clinician's task is to answer the question: "What condition does this
patient have?" A clinician is free to perform additional tests until
the proper diagnosis becomes clear. By contrast the epidemiologist
has to preselect the diagnostic criteria to answer the question: "Does
this individual in my population sample have the condition I am
studying or not?"
83
84
Manual of Epidemiology for District Health Management
The diagnostic criteria the epidemiologist uses may require stan
dardized interviews, physical examinations, laboratory examina
tions, as well as such examinations as radiography (tuberculosis),
electrocardiography (Chagas disease), slit lamp examinations of the
eye (onchocerciasis), sonography (of the liver and spleen in malaria
and schistosomiasis) and histopathology (leprosy). In the selection of
the diagnostic criteria to be used, the epidemiologist has to consider
the accuracy or validity of all the different methods.
There are two important aspects of validity, referred to as the
sensitivity and specificity of the test. For example, a test is said to
have a sensitivity of 90% if it correctly gives a positive result in 90%
of people who actually have the disease. On the other hand, a test is
said to have a specificity of 90% if it correctly gives a negative result
in 90% of people who do not have the disease.
The predictive value of a test, which depends upon the disease's
prevalence as well as on the sensitivity and specificity of the test, is
the most important measure for determining the test's usefulness
under field conditions. The positive predictive value measures the
likelihood that a test-positive person actually has the disease.
Neglecting cultural factors may have a considerable effect on
the validity of a questionnaire in surveys and may lead to wrong
conclusions about the frequency and distribution of the variable.
For example, consider the following survey of two villages in an area
endemic for schistosomiasis. A urine sample was examined for eggs
of S. haematobium and the questionnaire asked: "Do you have blood
in your urine?" In the first of the two villages surveyed this question
was considered offensive by the women, leading to marked under
reporting by them compared with the results of the parasitological
unne examinations carried out at the same time. In the second
village, where no such taboos existed, there was good agreement
between the interview results for both men and women on the one
hand and the unne examinations for eggs on the other. Without
the knowledge provided by the laboratory test, the questionnaire
results might have led to the conclusion that schistosomiasis in
the first village was more common in men than women. This
example demonstrates the importance of checking the validity of
all questionnaires.
Repeatability, validity and predictive value are considered in more
detail in Appendix 5, page 189.
7.12. Ethical issues
These can play a very important part in deciding whether to
undertake a survey or not. However, most of the major issues con
cern full-scale research studies; surveys undertaken as a part of the
85
Epidemiological Surveys
■
work of the district health services would be viewed less rigorously.
A much more detailed consideration of the issues is given in
Appendix 1, page 169.
In many developing countries where health care services are
limited, the population quite reasonably will expect some help with
its own health problems. Some service should be a part of any survey
and people found to be suffering from a disease or known not to have
received a preventive health measure, such as immunization, should
be treated or referred to a clinic.
Figure 7.5. No survey without health care
1
S’
■?
I'M SORRY, WE CAME ONLY TO COUNT YOU
Another issue concerns informed consent. Each individual in
cluded in the survey should have been informed about its purpose
and have given his or her consent to being included, even though
this may be difficult to organize. Particular care should be taken
when a potentially harmful procedure is used, such as a drug with
dangerous side-effects. Provided the survey procedures are well
established, it is common practice first to brief community leaders
and local organizations and to follow this by some kind of general
88
Manual of Epidemi
announcement that allows individuals to withdraw if they wish.
Ideally at the time of being seen, each individual should be given a
full explanation and then asked to sign a declaration to say that he or
she agrees to being included in the survey. However, in many cul
tures people would be very reluctant to give such written consent.
CONFIDENTIALITY IS VERY
IMPORTANT
The next issue concerns confidentiality. All information is given
in confidence and it should not be possible to identify any individual
in the analysis or in subsequent reports. In addition, no information
about any individuals contained in the records should be passed on
to other-people or organizations without the subject's consent.
In order to overcome some of these difficult ethical issues, it
would be wise to discuss the survey with other senior district offi
cers, community leaders and district health workers and to seek
their comments, advice and permission.
7.13 District survey methods checklist
In order to avoid problems in carrying out surveys or in interpret
ing the findings, particular attention should be given to the need for:
• Clear, quantified objectives.
• Good definition of cases and events.
• Proper sampling procedures and adequate sample size.
• Low non-response and refusal rates.
• Avoiding bias by using standardized techniques and good
equipment.
• Well designed and translated questionnaires.
• Well trained and supervised interviewers.
• Good pilot trials of methods, questionnaires and equipment.
• Using the same methods throughout the survey.
• Good commumcations with the population on consent and
confidentiality.
The most serious fault, however, often lies with the people who
carry out surveys because frequently the findings are never publi
cized or used in health planning. So why was the survey done at all? M
87
Chapter 8
Organizing Investigations and
Surveys
8.1
8.1
Preliminary plans
87
8.2
8.3
8.4
Organizing the fieldwork
89
jsgistics and support
90
District investigation and survey organization checklist 92
Preliminary plans
The organization of epidemiological investigations and surveys is
complicated and needs careful planning and much more time than
might be expected. If the organization proves too difficult and the
proposed investigation or survey is thought to be unfeasible, some
changes may have to be introduced, such as a reduction in the
sample size or omission of particular questions or tests. A pilot trial
is essential for testing the methods and the organization of the
fieldwork. A more detailed and practical guide to organization is
given in Appendix 4, page 179, which should be read by all district
staff who will be responsible for organizing such field investigations.
The survey organization should cover such details as:
• Personnel - number of workers and skills required.
• xMatenals - records, questionnaires and equipment.
• Finance
- purchases, salaries and allowances.
• Time
- preliminary organization, fieldwork and analysis.
An outline of the necessary preliminary steps is illustrated in
Figure 8.1 and a suggested timetable is shown in Figure 8.2.
i
The number and type of personnel required depends on the duties
that have to be carried out, e.g. professional, technical or administra
tive. .An analysis of the tasks to be undertaken for the survey is
essential and an estimate of the minimum educational levels re
quired for each job is useful in recruiting staff. As women in some
communities live in a secluded manner and male staff are not per
mitted to enter compounds or talk directly to women, female interviewing staff may be essential. Although a husband may answer
88
Manual of Epidemiology for District Health Management
Figure 8.1. Preliminary steps in organizing epidemiological investigations
Is a change in the
objectives or scope
necessary?
Can resources be
borrowed or o'brained
from elsewhere?
A PILOT TRIAL IS
ESSENTIAL FOR ALL
STAFF, METHODSAND
EQUIPMENT
r Consider objectives
and scope of the
proposed investigation
^or survey
| • Objectives • Population
• Methods • Time
I • Location
Consider availability
of resources
• Manoower • Time
• Materials • Money
• Equipment
Are resources sufficient
for the study?
Further logistics
planning
Consider organizational
needs for carrying out
survey
Consider also back-up,
logistics and
contingency plans
Use planning checklist
questions for his wife, the answers could be very7 misleading. Suffi
cient numbers of staff are required if the survey is to be completed in
a reasonable length of time.
The amount of materials required also depends upon the nature of
the survey. It is wise to estimate the minimum amount required and
then to top this up with extras to allow for unforeseen developments
or for underestimation.
Finances are usually required for such items as salaries, overnight
allowances, accommodation, food and petrol, or for unforeseen needs
such as new equipment, spares or additional transport.
Once the preliminary planning has been done and the different
stages identified, the time needed to complete each stage should be
estimated (see Figure 8.2). This will give an indication of the total
time required. A pilot trial at this stage is essential.
The dates for the actual fieldwork will have to be fixed some time
in advance. The time required for preparations and pilot-testing may
I
? I
eg
Organizing Investigations and Surveys
i
I
overlap with training, but all three must be completed in time for
the fieldwork. Similarly, analysis and consultations should be com
pleted before the final report is produced.
Figure 82. Example of organizational plan and timetable for proposed survey
Week
• I
I
February
12 3 4
March
1
2
3
4
1. Preparation of sample, forms
and questionnaire
Obtain staff and equipment
- I
January
12 3 4
2. Pilot-testing of methods and
all equipment
3. Training of staff
4. Fieldwork
5. Analysis
6. Consultation on findings
7. Final writing and distribution of report
Total time required
2 weeks
1 week
1 week
2 weeks
2 weeks
1 week
1 week
10 weeks
8.2 Organizing the fieldwork
If the investigation or survey involves a series of different meas
urements on each selected individual who attends a survey centre, it
is important to break the work up into stages that can be handled by
different staff. An example of a nutritional survey of young children
is shown m Figure 8.3. The sections on the record form should be in
the same sequence as the different stations (numbered 1 to 7) in
order to avoid confusion. Also, to avoid observer bias each assistant
should obtain information or take measurements before being able to
look at the rest of the record form. If the assistants see the form first
they will be tempted to write down information that fits well with
what is in the other sections. Each assistant should sign his or her
portion of the form in order to make supervision and correction of
faults easy and to reduce observer errors during the survey.
If the investigation is a household survey, it is important to be
realistic about the number of households that an interviewer can
visit in one day. Remember that it is comparatively easy to see the
first respondents, but the difficulty comes when chasing up the non
responders. This can require much staff time and effort. With a
household visit and interview that takes 30-45 minutes per house, it
is realistic for an interviewer to visit 5-7 households in one day.
When following up on non-responders, even fewer households may
be seen in one day by one interviewer.
Tl
00
Manual of Epidemiology for District Health Management
Figure 8.3. Example of organizational plan for a survey centre carrying out a
nutritional survey
Station 1
12 assistants)
Start
Registration and identification of individual on sample list and
allocation of survey identification number.
Station 2
(1 assistant)
Entering of name, address and other information on record forms;
labelling of specimen containers.
Station 3
(2 assistants)
Interview of patients, assessment of age and completion of
health questionnaire.
Station 4
(1 assistant)
Measurement of height, weight and arm circumference; eye test.
Station 5
(1 physician,
1 nurse)
Physical examination and recording of signs of malnutrition.
Station 6
(1 assistant)
Specimens taken for laboratory tests.
Station 7
tl medical
assistant)
Dispensary clinic for treatment and drugs. Check subject’s
form is fully completed before saying thank you and goodbye.
Finish
8.3 Logistics and support
A full checklist of logistic and support items is given in Figure 8 4
and what follows is a brief summary. For more details refer to
Appendix 4, page 179.
Transport is usually essential in most developing countries,
ransporting people and equipment by road, water or air will require
careful planmng. Allow extra time for mishaps and pack all equip
ment very carefully. If possible, have back-up transport in case of
emergencies.
Specimens for laboratory investigation may require special con
tainers, labelling, storage and transport. A cold chain is needed for
biological reagents, such as vaccines.
Accommodation and cooking facilities need to be arranged in
advance md employing a cook wiU save on staff time. Food may
need careful storage and cooking in order to avoid contamination
Water for drinking may need to be punfied, filtered or boiled Facili
ties for refuse disposal and toilets may also be needed if survey staff
stay m rural areas.
Loss of stores and supplies, particularly due to theft can be a
major problem It is best to appomt one staff member as storekeep
erz
so that he or she alone is in charge of stores and issuing of items/
91
Organizing Investigations and Surveys
Good financial management is essential for staff morale. Salaries
and allowances need to be paid on time, and petty cash should be
available. A detailed record of expenditure is necessary, together
with receipts, as the most senior staff member will have to account
for all funds issued and spent.
Figure 8.4. Survey planning checklist
1. Proposed study
Title
Type
Duration
Location
Person in charge
Contacts
2. Clearances
Local authority
Police
Government
Higher authority
3. Location
Climate
Geographical features
Maps
Road conditions
River conditions
Airstrips
4. Data collection
Type
Regularity
Timing
5. Staff requirements
Functional categories
Number
Existing/new staff
i
i
6. Accommodation
Location
Survey team
Support group
Females/males
Tents
Electricity
Water
7. Supplies
Immediate
Replenishments
Stockpile
Order/indents
Food/cooking
Water/purification
Petrol
8. Transportation
Vehicles, bicycles
Boats
Aircraft
Maintenance
Tools
Spares
9. Equipment
Survey equipment
Stationery
Chemicals
Generator
Waterproofing
Other equipment
10. Specimens
Reception
Pick-up schedules
Refrigeration
Containers
Instruction slips
11. Special points
Emergencies
Back-up
Communications
Medical care for staff
Medicines and drugs
Records
Photographic eqipment
Tape recorders
1
82
Manual of Epidemiolog
8.4 District investigation and survey organization
checklist
The following activities are listed in the order in which they
might be done:
1
• Planning:
- Decide why survey results are needed and how they will be
useful.
- Consult people with the relevant experience in local district
government, community leaders and health workers.
j
- Visit local villages to discuss the survey and hear comments
from the people about their culture and local environment
- Decide which observations and measurements are needed and
standardize the techniques.
- Choose an appropriate population sample.
- Design and pilot-test record forms and questionnaires.
- Make arrangements for staff, equipment, transport, finance,
accommodation, etc.
• Organization:
- Obtain cooperation from local leaders.
- Train survey staff.
- Arrange for laboratory procedures.
- Draw up a daily work plan for all staff.
- Pilot-test all organizational details, including staff, methods
and equipment.
• During the fieldwork:
- Supervise all staff to ensure a high standard.
- Ask local leaders to help with organization and checking for
attenders and non-responders.
- Make random checks on staff at the survey centre or interview
ers during household visits.
• Analysis and communications:
- Analyse the data as soon as possible, preferably daily
- Discuss results and their meaning with health workers and
community leaders to obtain their comments
- Write report, incorporating comments, and make recommends
tions for new or improved health programmes
- Distribute report and discuss recommendations with relevant
local committees and organizations and with local media, if ap
propriate.
- Plan to evaluate any changes introduced as a result of the
survey, to estimate their effectiveness.
I
93
Chapter 9
Record Forms and Coding
9.1
9.1
Use of forms
93
9.2
94
9.3
Design of forms
Coding of information
9.4
Individual number
96
9.5
Layout of record forms
97
9.6
Coding column
97
96
Use of forms
Two types of forms may be used for the collection of information:
one is given to the person, called the respondent, to complete by
themselves and is known as a self-administered questionnaire. The
other type requires an interviewer to ask for the information from
the respondent and is called a record form or questionnaire. Which
approach is used depends, to a large extent, on the literacy of indi
viduals. In many developing countries interviewers are used much
more frequently than self-administered questionnaires.
Epidemiology often uses information from several different
sources, such as clinic registers (e.g. outpatient and mother and child
health clinics), lists of patients under treatment (e.g. in tuberculosis
and leprosy), records of special investigations undertaken (e.g. chest
X-rays, blood-slides examined), or special registers set up to collect
information on a particular disease (e.g. registers for cases of cancer,
malaria or trypanosomiasis). However, in district epidemiology it is
important to assemble all the information on each of the individuals
concerned on one record form. This is particularly true for special
investigations or surveys.
Record forms often contain four kinds of data:
• Basic or preliminary data, including identification.
• Questionnaire information.
• Results of physical or medical examination.
• Information from special investigations.
B4
Manual of Epidemiology for District Health Management
9.2 Design of forms
Designing forms is not easy. Care should be taken to ensure that all
relevant information is included while at the same time all repetitions
and irre evancies are left out. The form should be clearly laid out in
a sensible stepwise manner, keeping in mind the educational back
ground of both the interviewers and respondents. Before a survey can
be undertaken, therefore, careful planning of the survey forms is
necessary and pilot-testing is essential.
When designing record forms four main rules need to be followed:
A. separate fom should be used for recording information on each
individual to facilitate analysis.
All the required information must be clearly requested on the form.
The form must be easy to use and clearly laid out, with each
section in its proper sequence.
Each form should be laid out to make data processing and analysis
easy.
“ the
a set of instructions is necessary
for the interviewer to refer to if and when a problem arises. This is
ownaS/r"60658/17 fo[interviewers wh° may be working on their
avoid eoX p»ble™
SUP'"iS“'
Hp t0
Some guidelines on the design of record forms:
• Avoid the use of very lengthy forms. Much time and energy is
wasted in collecting information that has no relevance to the ob
jectives of the survey. Furthermore, a lengthy form with too many
questions can be annoying to the respondents, who are under no
ODugation to participate in the survey.
’
ONLY ONE INDIVIDUAL
PER FORM AND ALL
INFORMA T10N ON ONE
FORM
the ^estions m o^er of difficulty. It is wise to save
difficult or embarrassing questions for the end, when the inter
viewer should have estabhshed a closer rapport with the responPhrase questions in clear and simple language. Avoid the use of
t^hmoai or ambiguous terms. Try to phrase questions in such a
ay that they wdl sound as though the interviewer is having a
conversation rather than conducting an interrogation.
Make sure that respondents can answer your questions. There is
no point in asking questions on matters that are outside the remmembered^6"161106
t0° 10ng ag° t0 be proPerly
II
95
Record Forms and Coding
-
Figure 9.1. Example of the first page of a precoded record form used in
Malaysia
COMMUNITY HEALTH SURVEY OF VILLAGE
(PERIOD: 1 April to 30 April 1988)
Date of interview:
INSTRUCTIONS:
/
Interviewer:
/
Fill in all the blank spaces
Circle the appropriate answer in sections E-H
Do not omit any item of information
Do not fill in the boxes in the coding column
II
I
Coding column
(D
Village:
□□□
(2.3,4)
House no:
12 3 4
Household no:
□□
(6.7)
Individual no:
(5)
5 6 7
PERSONAL PARTICULARS
A Name:
B Address:
: I
C Sex:
2 Female
1 Male
8
/
D Date of birth:
Day
/
Month Year
□□
Age last birthday (calculated from date of birth):,
9 10
I
E
Present marital
1 Not married 2 Married
status:
4 Divorced/separated
3 Widowed
5 Other (specify)
11
F
Ethnic group:
1 Malay
2 Chinese
3 Indian/Pakistani
4 Other (specify)
12
G Place of birth:
1 Malaysia
2 Singapore
4 China
5 India/Pakistan
3 Indonesia
6 Other (specify)
1
:
13
H Religion:
1 None
4 Ancestral
2 Islam
5 Buddhism
3 Christianity
6 Hinduism
7 Other (specify)
14
SB
Manual of Epidemiology for District Health Management
9.3
Coding of information
Although the use of precoded answers is not necessary for all
surveys, processing of data (see next chapter) can be made easier if
the answers have been numerically coded. Coding consists of assign
ing a numerical value or code to a specific item of information. For
example, we may use the numerical value 1 to denote a "Yes" an
swer for a particular question and 2 to denote a "No" answer for the
same question. There is no hard and fast rule on the choice of nu
merical codes. However, standard replies like "Yes", "No" "Don't
know", and "Not known" should preferably have the same codes
throughout the recording form, to avoid confusion and errors in the
analysis stage.
As an example, the different categories for the usual source of
drinking-water may be numerically coded as follows:
1__
tap in house
2_
tap or pump in yard
3__
tap or pump in public place
open well
4_
5_
6_ spring, river or lake
7_ other sources—specify
8_ (not used for a code)
9_ unknown
rainwater
It is useful to provide a category labelled "others" which can be
used for all those answers that do not fit into one of the precoded
categories. The interviewer should be instructed to write down, or
specify, these "other" answers so that they can be analysed later.
PILOT-TEST ALL
QUESTIONS AND
ANSWERS BEFORE USE
IN SURVEYS
When the information is numerically coded, it is preferable to
have a single-digit code for each item of information as this makes
the data processing easier, particularly when manual techniques are
used. However, coding for exact age is a typical example where a
two-digit code is needed. If a 75-year-old man has to be coded 75, a
6-year-old girl has to be coded 06. Exact age should be collected if
possible, not just the age group. This makes analysis more flexible
and permits the selection of particular individuals in the sample,
such as women aged 15-44 years or children below 5 years old.
I
Alternatively, ages can be grouped into 5-year intervals such as
0-4, 5-9, 10-14, 15-19 and so on. The age group 10-14, for example,
will comprise all children from 10 years to just below 15 years.
Children less than 1 year old are often separated from the 1-4 year
group, particularly in countries with high infant mortality. Larger
age intervals are also used, e.g. 15-44-year-old females for the repro
ductive age interval and 60 years and above for old people.
9.4 Individual number
The individual's survey number depends on the sample size used.
If this does not exceed 99, the first individual is given the number
01 and the ninety-ninth the number 99. The number can be more
4
97
Record Forms and Coding
f
complex. For example, in a large-scale survey involving randomly se
lected clusters of villages, it may be necessary (as in Figure 9.1) to
identify a person by his or her village, house and household using a
series of codes:
I I
INDIVIDUAL SURVEY
NUMBERS HELP TRACE
PEOPLE
i
1
2 3 4
House
Village
6 7
Person
5
Household
The first digit is used to identify the village in which the person
lives, the next three digits for the house, the fifth digit for the houses
hold should there be more than one household in that house. The
sixth and seventh digits give the identification number of the person.
A serial number like 3-126-2-08 is used to reference the eighth per
son belonging to the second household of house number 126 in
village 3. Serial numbers can be very useful for tracing individuals.
9.5 Layout of record forms
In the design of a record form, there is a variety of formats to
choose from. The precoded answers to a question like "How many
living children do you have?" can be arranged as shown:
i
(
0_
None
1_
1 child
2_
2 children
3_
3 children
4_
4 children
5_
6_
7_
5 children
8_
9_
8 or more children
6 children
7 children
Unknown
Using a tick (/) or a cross (X) in boxes makes the coding of the
information neat and easy. However, in the absence of boxes, the
numerical code for the appropriate category should be circled as
shown below.
To reduce space on the form the replies can be rearranged on one
or two lines:
Number of children:
j
i
i
I
O(none)
1
2
3) 4
5
6
7
8 or more
9 (unknown).
9.6 Coding column
Provision is usually made for a coding column on the right-hand
margin of the form, where the appropriate numerical code is entered
(see Figure 9.1). The purpose is to facilitate data processing, particu
larly when the data have to be punched on to punch cards or entered
directly into a computer. Once the coding boxes within the coding
column are filled, processing of the data can proceed efficiently and
speedily simply by sorting the numbers in the appropriate boxes in
the column without referring to the answers themselves.
98
Manual of Epidemio
It is easy to make mistakes in transferring the precoded informa
tion from the question to the appropriate coding box in the coding
column. For this reason, it is common practice to fill in the coding
column after the interviews have been completed, when the coding
can be done at a more leisurely pace. This coding should be done by
only one or two people and each coding should be double-checked by
another person to make sure it is correct. Coding errors need to be
corrected before the analysis is started.
I
i
•w
r
99
CHAPTER 10
Data Processing and Analysis
I
10.1
10.2
10.3
10.4
10.5
10.6
10.7
10.8
Processing and analysis
Hand-tallying and sorting
Steps in data analysis
Simple tabulations
Cross-tabulations
Summarizing statistics
Correlation
Standardization
99
100
101
102
107
110
110
111
10.1 Processing and analysis
Data only become really useful when they have been processed
and analysed. Health data may come from routine information
systems, surveillance systems, investigation of epidemics or special
surveys.
Once the data have been collected and coded, the next step is to
process them to provide information on all the important questions
or variables. Data processing within districts will probably be under
taken by hand-tallying, hand-sorting or microcomputing.
For simple descriptive epidemiological data on up to 300 individu
als, data processing and analysis can be done easily, and quite
quickly, by two or three people using hand-tallying and/or hand
sorting. Two advantages of these methods are that recording and
coding mistakes can be corrected immediately and that they give a
very good 'Teel" for the data and their meaning. After the initial
tallying, the totals, percentages and rates can easily be calculated
using a small electronic calculator. The use of microcomputers, with
the coded data entered directly via the keyboard, is recommended
only for large and complicated surveys. Many developing countries
are introducing the use of microcomputers for handling health infor
mation, but as yet few have good and reliable microcomputing facili
ties for district health teams. Edge punch cards used to be popular
but they are now rarely recommended because microcomputers have
largely taken their place.
I
100
Manual of Epidemiology for District Health Management
10.2 Hand-tallying and sorting
Hand-tallying involves counting the number of times a particular
category of information appears in all the record forms. If the data
have not been pre-coded, all the possible answers must first be listed.
For example, when processing data on the symptoms reported for a
particular disease, first list all the possible symptoms and then go
through all the record forms. As a particular symptom is encoun
tered, count it or record it by making a vertical mark against that
particular symptom on the tally sheet thus:
Figure 10.1. Example of a tally sheet
Symptoms
Tally
Nausea
//
2
Total
Diarrhoea
wz //
7
Constipation
W TM /
11
Arthralgia
W /
6
For easy counting, every fifth mark is drawn diagonally across the
preceding four marks to produce the notation (7//Z), which indicates a
group of five items. In the example just given, we can quite readily
see the totals for each symptom.
When the number of forms to be analysed is larger than about
100, tallying becomes prone to error, often as a result of fatigue on
the pan of the person doing it. Common errors in tallying include
the misclassification of an item, e.g. "nausea" when it should be
diarrhoea; double-counting an item or missing one altogether. The
risk of errors is higher when tallying for a two-way tabulation, as in
Figure 10.2:
Figure 102. Hand-tallying for a two-way table, showing the age and birth
order of children
Birth order of child
Age
1
2
3
4
(years)
5 or
Total
more
0-4
7W
5-9
10-14
/
Tf// /
///
W //
/
15-19
20-24
7//ZWZ/
Total
17
//
9
14
///
//
11
3
///
w
//
////
14
/
11
10
18
5
56
Hand-sorting is similar to tallying in that counts are made of
specific items of information as they appear on the record form. In
hand-sorting, the first step is to decide on the different levels and
then to sort all the record forms according to the item of information
and count the total in each pile. The totals for all piles should add up
to the total number of forms being analysed. An example for highest
educational level achieved is shown in Figure 10.3.
Figure 10.3. Hand-sorting forms by highest educational level achieved
I'
(4)
i
NONE
(4)
(8)
SECONDARY
PRIMARY
(2)
TERTIARY
Like tallying, this procedure is susceptible to errors. However,
unlike tallying, errors arising from misclassification can be more
easily checked and rectified. For example, if in checking the pile for
primary level, we come across a secondary level form, all we need to
do is to transfer it to the "secondary" pile. Hand-sorting can be a
tedious procedure, particularly if the sample size is over 100 indi
viduals.
10.3 Steps in data analysis
When the data are processed, each variable (or question) for each
subject can be counted and these counts summarized as tables.
Graphs and diagrams cannot be drawn without the use of tables.
The main set of tables for the expected end results (called
"dummy" tables) should have been developed early on in the investi
gation when the objectives of the study were agreed upon (see Sec
tion 7.3 on survey objectives). These dummy tables act as a guide in
the analysis, and may be reviewed and modified as the analysis
progresses. JWhen constructing tables three conditions must be satisfied:
• All data should be in a form that can be classified into categories:
e.g.
spleen palpable: Yes or No
age or haemoglobin: actual numbers or values.
oo
06316
102
Manual of Epidemiology for District Health Management
• All table categories must be mutually exclusive
i.e. no individual can be classified twice in one table.
• Each table should include all the ''raw" data
i.e. no individual can be left out of a table, except under special
circumstances.
These conditions mean that if the observed value of a variable,
such as age or haemoglobin, is entered directly on to the record
forms, a classification must be designed before the analysis can take
place. Where precoded categories are used, the classification obviously already exists. When designing a classification it is useful to
remember that there must be at least two categories, but as a general
rule there should not be more than 10.
Statistical analysis can involve complicated techniques. However,
district epidemiology makes use of three separate stages in analysis
that it is best to go through in sequence.
The three essential steps in data analysis are:
• Analyse each variable separately, one after the other, for the
distribution of counts in each category. This is called simple or
one-way tabulation. What do the distributions look like? What do
they tell us ?
• Analyse pairs of variables that are relevant, such as incidence of
cases by age group, sex or month of the year. This is called twoway or cross-tabulation. Are there any findings or associations
that stand out clearly?
• Calculate the appropriate means, percentages and standard devia
tions. These are called the summary statistics.
By the time these three stages of the analysis have been com
pleted the person doing the analysis and the DHMT should have a
good "feel" for the data. In drawing conclusions it is important to
look for associations between variables and to check whether any
apparent differences are real or not, that is whether they are significant or not. Tests of statistical significance may be needed at this
stage, the results of which are commonly expressed in terms of the
probability that such an association could have occurred by chance.
Often an association is arbitrarily considered as significant if the
probability (P) is less than 5% (P < 0.05), i.e., it would occur by
chance less than 1 time in 20.
10.4 Simple tabulations
W
Table 10.1 shows a set of raw data obtained from a cross-sectional
survey of a random sample of 100 villagers for haemoglobin levels
and hookworm infection. Additional information was also collected
on the age and sex of the villagers. This set of raw data will be
used to illustrate the various kinds of statistical tables that can
be constructed.
i
Data Processing and Analysis
103
Table 10.1. Raw data on haemoglobin levels (grams per
100 ml) and hookworm infection (positive
or negative) for 100 villagers, obtained by a
cross-sectional survey of 10 randomly
selected individuals in each of 10 randomly
selected villages from the whole district
I
Individual
Age
Sex
(years)
1
2
3
4
5
6
7
. 8
9
10
11
12
13
14
15
16
I
17
Ii
18
19
20
21
22
■ I
23
24
25
26
27
28
29
30
31
32
33
34
46
9
8
30
45
2
12
33
4
57
38
51
42
38
41
8
16
0
1
4
2
13
8
12
59
19
3
4
32
27
43
4
62
33
F
F
F
F
F
M
M
F
F
M
M
F
M
M
F
F
F
M
M
M
F
F
F
M
F
M
F
F
F
M
F
M
F
M
Haemoglobin
Hookworm
(g/WOml)
infection
10.3
8.4
9.0
9.4
8.4
10.0
8.3
10.9
8.8
10.9
10.9
7.3
8.3
9.0
9.8
9.3
9.7
13.8
8.9
7.5
10.9
10.7
10.5
9.6
6.8
11.6
9.2
8.4
10.9
11.5
11.4
9.4
10.3
12.7
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
LTBRABY JJ ftM
104
Manual of Epidemiology for District Health Management
Table 10.1 (continued)
Individual
Age
Sex
(years)
Haemoglobin
Hookworm
(g/IOOml)
infection
+
35
8
M
10.2
36
37
37
M
11.2
56
M
12.2
38
39
F
10.0
39
37
M
12.2
40
6
M
10.1
41
2
14
F
11.0
M
10.3
+
F
+
42
43
44
45
46
47
48
49
+
48
62
0
F
8.4
8.9
M
13.1
16
F
9.3
+
F
6.4
+
+
50
9
24
20
68
51
12
M
10.8
52
24
M
11.8
53
71
M
6.6
+
54
45
M
+
55
3
M
8.7
9.8
56
52
M
12.4
12.0
M
11.7
F
10.0
F
9.1
57
21
F
58
59
60
33
F
10.6
7
F
10.8
42
F
11.0
61
5
M
10.0
62
36
F
10.9
63
64
1
F
7.4
65
66
67
68
69
70
+
+
+
+
8
29
4
0
2
F
M
10.2
+
+
13
‘ F
10.4
+
35
M
10.8
+
F
11.6
F
10.8
F
9.8
8.6
I
105
Data Processing and Analysis
I
Table 10.1 (continued)
Individual
Age
Sex
(years)
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
Haemoglobin
Hookworm
(g/IOOml)
infection
63
M
7.0
+
35
M
11.0
+
42
M
10.2
+
3
F
10.2
+
4
M
23
M
9.4
6.8
+
34
M
12.2
41
F
10.0
+
1
M
51
M
9.4
9.6
+
36
M
11.2
8
F
11
M
7.9
9.2
14
M
12.6
38
F
10.2
56
F
9.1
+
19
F
9.4
+
27
F
10.8
9
M
12.7
36
F
6.2
+
14
F
10.1
+
42
M
12.9
10
M
11.2
39
M
8.4
11
F
11.6
8
F
9.7
23
M
11.9
13
M
17
M
7.8
9.8
41
F
12.1
+
+
+
+
+
+
+
+
i
All tables must have a proper title which should be complete,
concise and self-explanatory. The title should give information on
the sample being analysed and on the variables being presented.
Table 10.2 shows the age distribution by broad age groups of the
sample of 100 villagers. As the tabulation involves only one variable,
108
Manual of Epidemiology for District Health Management
namely age, we have a simple or one-way table. In this example,
age is treated as a discrete measurement and broad age groups have
been used because the number of subjects is fairly small. To group
the villagers, the two opposite extreme values are found from the set
of raw data, (in this case, less than 1 year old and 71 years old). The
first category in the table includes the smallest value and the last
category the highest. Having obtained the classes, we can then
process the raw data by hand tallying or sorting.
If a larger sample were being analysed, the use of a larger number
of age classes might be necessary. In that case, the age groups might
be: under 1 year old, 1-4, 5-14, 15-24, 25^34, 35-44, 45-54, 55-64 and
65+ years.
Table 102. Age distribution of the sample
of 100 villagers
Number
Age group
lyears)
0-4
19
25
40
16
100
5-14
15-44
45+
Total
Table 10.3 shows the age and sex distribution of the sample of 100
villagers as an example of a two-way table. The age-sex distribution
of the sample can now be compared to that for the district. If the two
are similar, the sample is probably reasonably representative of the
whole district population.
Table 10.3. Age and sex distribution of the sample of
100 villagers
Age group
Male
Female
Total
10
9
13
20
9
51
19
25
40
16
100
lyears)
0-4
5-14
15-44
45+
Total
12
20
7
49
I
107
Data Processing and Analysis
For a continuous measurement variable, such as haemoglobin
level, a simple tabulation starts by grouping the data in categories, as
shown in Table 10.4.
Table 10.4. Distribution of haemoglobin
levels in 100 villagers
I
Haemoglobin level
I
Number of
lg/100 ml)
villagers
6.0- 6.9
7.0- 7.9
8.0- 8.9
9.0- 9.9
5
22
10.0-10.9
29
11.0-11.9
14
12.0-12.9
9
2
13.0-13.9
Total
6
13
100
To make up the categories, the two extreme haemoglobin values
(i.e. the lowest and highest) are first identified; the difference be
tween them gives the range. In this example, the lowest reading in
the sample is 6.2 and the highest 13.8; the range is therefore 7.6
grams/100 ml. The range is then divided by 10 to obtain a crude
value for the class interval. (Since about 10 categories or groups
usually gives a good distribution, dividing the range by 10 is a good
starting point.) This would give an interval of about 0.8 g, but as a
whole number is preferred, intervals of 1 g are used. If the number of
categories is too small or too large, it may be difficult to make sense
of the distribution of the variable.
10.5 Cross-tabulations
i
Cross-tabulations involve the use of at least two variables. In its
simplest form, a cross-tabulation consists of 2 rows and 2 columns
(excluding the row and column totals). Table 10.5 gives an example
of a two-by-two table, which is often used when both variables are
classified as "present" or "absent"
108
Manual of Epidemiology for District Health Management
Table 10.5. The distribution of hookworm-infected subjects by haemoglobin
levels of less or more than 10 g/1OO ml.
Haemoglobin level
Hookworm infection
present
Total
% with
hookworm
absent
Less than 10 g/1OO mi
35
11
10 g/1OO ml or more
24
30
46
54
Total
59
41
100
% with anaemia
59.3%
26.8%
76.1%
44.4%
A haemoglobin value of less than 10 g/100 ml can be used as an
arbitrary cut-off point for the diagnosis of anaemia, which affected 46
out of the 100 villagers. Over half of the villagers, 59 in all, were
found to be infected with hookworm. Of those villagers with hook
worm infection, 59.3% were anaemic compared with 26.8% of those
with no hookworm. A further analysis by age and sex could be made
using the Chi Squared test, which is a useful test of statistical sig
nificance to apply to z/2 by 2Z/ tables. Use of the test on these data
would show that in these villages anaemia was statistically associ
ated with hookworm infection (P, or probability, <0.05).
Table 10.6 gives a further example of a cross-tabulation involving
the two variables haemoglobin and sex, but this time with more
than two categories for haemoglobin. The grouping of the data for
haemoglobin has been done in the same manner as in the simple or
one-way tabulation shown in Table 10.4.
An example of a three-way tabulation is illustrated by Table 10.7.
The three-way table is constructed by processing all the data on the
three variables, namely haemoglobin, age and sex, given in the set of
raw data. From this three-way table, one-way or two-way tables can
be obtained by looking at the totals. For example, in the three-way
tabulation given in Table 10.7, the row totals give the simple tabula
tion for haemoglobin levels.
I
109
Data Processing and Analysis
Table 10.6. Distribution of the 100 villagers by haemoglobin levels and sex
i
Total
Sex
Haemoglobin level ____
(g/100 ml)
Males
Females
(both sexes)
6.0- 6.9
7.0- 7.9
8.0- 8.9
2
3
3
5
9.0- 9.9
10.0-10.9
11.0-11.9
12.0-12.9
13.0-13.9
10
10
Total
3
6
13
7
12
19
6
22
9
7
2
2
0
29
14
9
2
49
51
100
5
Table 10.7. Distribution of the 100 villagers by age, sex and haemoglobin level
(g/IOOml)
1
I
Total
Age group (years)
Haemoglobin level
0-4
5-14
15-45
45+
M F
MF
MF
M F
M F
6.0
0
0
0
1
1
1
1
1
2 3
7.0
1
1
1
1
0
0
1
1
3 3
8.0
1
3
1
1
2 0
2
3
6
9.0
5
1
2 10
1
2
2
10 12
10.0
4 2
2 2
11.0
0
1
0
0
12.0
0
0
8 2
4 2
1
0
13.0
2 0
0
0
0
0
9 5
7 2
2 0
Total
10
12 13
7 9
49 51
9
2 3
5 5
1 2
2 0
0
3
0
20 20
7
10 19
110
Manual of Epidemiology for District Health Management
10.6 Summarizing statistics
All the data on a variable can often be expressed in a much briefer
way by using summarizing statistics or indices.
The most useful indices are:
• Percentage of subjects infected,
e.g.
schoolchildren infected with S. haematobium or malaria.
• Percentage of subjects above or below a certain cut-off point,
e.g.
adults with a systolic blood pressure of 160 mm Hg or more
in hypertension and pregnant women attending antenatal
clinic at least once.
• Mean and range. The average is another name for the mean and
the range expresses the difference between the lowest and highest
values in the raw data.
e.g.
mean birth weight and average number of visits made by
children to child health clinics.
• Standard deviation (often called SD) can be used when there is a
reasonably normal or bell-shaped distribution about the mean.
(For an example see Figure 11.6.) The SD can easily be calculated
using some scientific electronic calculators. A large SD indicates a
wide scatter of individual values on either side of the mean or
average while a small SD value indicates a narrow distribution of
values.
e.g.
severe malnutrition defined as those children more than two
SDs below standard weight-for-age,-
normal range for many laboratory tests is often expressed as
the mean plus or mmus 2 x SD.
10.7 Correlation
When two quantitative variables are associated or correlated with
each other, an increase or decrease in one is associated with an
increase or decrease in the other. For example, the newborn baby's
weight is positively correlated with mother's weight and the propor
tion of children with lower than normal weight-for-age is often
negatively correlated with family income.
Whenever there are two continuous variables, often measured.on
the same individuals, a very useful technique is to draw a scatter
diagram, or scattergram, as shown in Figure 11.12. The more
closely correlated the two variables are, the closer the dots will
appear to be on a straight line. This correlation can also be
calculated using statistical formulae and it is then expressed as a
correlation coefficient.
11
Data Processing and Analysis
Whenever two variables are thought to be correlated it does not
necessarily mean that one variable is the cause of the other. Great
care must be taken in interpreting the relationship between vari
ables. For instance, a mother's height may partly determine the
baby's weight but the reverse is biologically not true.
10.8 Standardization
To make comparisons, for example between the prevalence of
malnutrition in the district 10 years ago and at present, or between
disease prevalence rates in the district and in the country as a whole,
it is necessary to standardize for the different populations. This
avoids comparisons between two populations with different shaped
population pyramids and age-sex distributions.
To handle increases in population size, we commonly use a per
centage or a rate per 1000 people, which can be made more accurate
by using an age-, sex- and disease-specific percentage or rate, e.g.
20% of male children 5-9 years old were infected with Bancroftian
filariasis in 1986 compared with 10% in 1976.
I
Where the age-sex structure or pyramid of the district is noticeably
different from the national one, statistical adjustments should be
made before comparing district with national rates. A technique
called direct standardization can be used to do this. The technique
and an example are explained in Appendix 6, page 193. It is impor
tant to standardize for the age structure of the district population
where a variable, such as the risk of death, is unevenly distributed
according to age and for the sex structure where the variable, such
as number of births, is associated with one sex. The crude death rate
(CDR) could be much higher in one district because there is a higher
percentage of children living (and dying) in that district compared
with another one nearby. In this case comparing the CDRs for dis
tricts would not be valid; the same is true if the district rates for
today are compared with those of 10 or 20 years ago. Further details
on using disease-specific rates and direct standardization are given in
Appendix 6.
111
112
<
Manual of Epidemiology for District Health Management
I
113
t
C H A P T E R11
i
II
i
I
I
Presenting Health
Information
11.1
Tables and figures
113
11.2
Graphs
115
11.3
11.4
Frequency histograms
118
Bar chans
119
11.5
121
11.6
Pie charts
Scatter diagrams
11.7
Maps
124
122
J
!
11.1 Tables and figures
Tables are the essential means of presenting an organized set of
analysed data, particularly numerical or quantitative data. Figures,
graphs and maps are also frequently used because they can present
visual information much more clearly than tables, for example when
we want to show comparisons, patterns or time trends. Figures are
also very useful for qualitative, or non-numerical, types of informa
tion.
(
i
The most frequently used methods for presenting health informa
tion are:
• Pie charts
• Tables
• Graphs
• Scatter diagrams
• Histograms
• Maps
• Bar charts
The following are some important points about using tables and
. figures.
I
!
J
Titles should always be concise and self-explanatory, expressing
clearly all the information that is being presented. The meaning
of the title should be immediately obvious to the reader, without
having to refer to the text for an explanation. For example: " Preva
lence rates by age for Schistosoma haematobium in five provinces
of Zambia, ^SO".
114
Manual of Epidemiology for District Health Management
■
Rows and columns must be clearly labelled and, where appropri
ate, all the categories should be clearly shown. For examples, see
Tables 10.2 and 10.5 on pages 106 and 108, respectively.
Axes of graphs and diagrams should be properly defined, and
clearly labelled with their scales. The vertical axis of a graph is
known as the Y-axis, or the ordinate, while the horizontal axis is
known as the X-axis, or the abscissa. For example, in Figure 11.1 the
horizontal axis shows birth order, beginning from a value of 1 and
increasing by 1 until the highest recorded birth order in the sample is
reached. The vertical axis shows the frequency of individuals, pre
sented in intervals of 10 individuals. Thus the first bar (labelled A)
of the diagram indicates that there were 50 individuals in the sample
of infants that were of birth order 1 and only 5 were of birth order
5 or more.
Figure 11.1. Distribution of 100 consecutive live births by birth order
50
oo
n:
40
A
CQ
LxJ
>
30
-J
O
cr
□J
20
CQ
sz
=5
10
I
0
1
2
3
4
BIRTH ORDER
5
6
7
Keys or labels are necessary in graphs with more than one line, i.e.
when information on more than one group is presented. The labels
identify the different groups being presented for comparison (see
Figure 11.2).
Footnotes are used to indicate the source of the original informa
tion. In reports it is quite common for a diagram or chart to be repro
duced or adapted from another source to illustrate the issues being
discussed.
l
115
Presenting Health Information
11.2 Graphs
These are the most commonly used type of figure, particularly for
showing numerical data such as deliveries per month, percentage of
children immunized by year or number of new cases per month of a
disease, such as trypanosomiasis or kala-azar.
Figure 11.2 shows a graph of time trends of the total number of
different health workers over a 10-year period. These show that the
number of nurses has increased most, from less than 3000 to nearly
6000, with midwives and doctors next. Note, however, that in per
centage terms, the pharmacists have increased fastest, showing a
nearly 3-fold growth from about 100 to nearly 300.
Figure 112. Growth in number of registered health workers, 1975-1985
6000
i
/ NURSES
5000
tn
<r
o
M000
Q
>
. I
Q
Z
3000
MIDWIVES
o
an
LU
2000
DOCTORS
co
sz
zz>
z:
1000
0 Ln
1975
DENTISTS
J
J
iPHARMAQISTS
1977
1979
1981
1983
1985
YEAR
I
i(
i
Graphs are useful for showing two or more distributions, provid
ing the difference between the lines is clearly shown. Figure 11.3
shows the number of mothers delivered each month in a district by
trained traditional birth attendants (TBAs) during one year compared
to the number delivered by professional midwives in the district
health centres and hospitals. Assuming that there were a total of
9000 deliveries per year in the district (see Section 3.3), the trained
TBAs supervised about 3000 or 33% and midwives a further 1900 or
21% of all deliveries between them; thus, 54% of all births were
attended by a trained health worker. The graph also shows that the
118
Manual of Epidemiology for District Health Management
number of deliveries undertaken by midwives each month rose
towards the end of the year, whereas the number undertaken by
TBAs remained fairly constant.
Figure 11.3. Number of reported deliveries undertaken
each month by trained traditional birth
attendants (TBAs) in homes and by nurse
midwives in health centres and hospitals
during 1386
300-i
TBAs
Midwives
200m o
LU £2
cn ex:
UJ UJ
> o_
Zj .-Z
a g 100u_ t—
O az
o
o z
J
F
M
A
M
J
J
A
S
0
N
MONTH
How to draw a straight-line frequency graph
• Draw the horizontal axis (the X-axis). Mark off the scale using
equal units. Use the mid-point of each interval to represent all
measurements lying within that interval.
• Mark off the vertical axis to show the frequency, commonly as a
number, percentage or rate.
• For each class of the grouped data mark a pomt where the vertical
(frequency) and the horizontal (scale) values intersect.
• Join the marked points with straight hues. The lines can be ex
tended beyond the first and last classes to touch the X-axis. This
kind of graph is called a frequency polygon.- an example is shown
in Figure 11.4.
Graphs can be used effectively to compare two frequency distribu,tions, e.g. birth weight by sex. Figure 11.4, for example, suggests that
there were more low-birth-weight female than male babies.
Cumulative frequency graph
This graph expresses a cumulative distribution, often expressed as
total numbers or as a percentage. This is a useful graph for showing
progress in implementing a planned activity, such as immunizations.
I
I
(
117
Presenting Health Information
I
Figure 11.4. Frequency distribution of 100 live births by sex and birth weight
14
A
I
I
\
I
12 i
oo
zz
cn
t
I
I
FEMALES •
CQ
UJ
I\
10-
>
8-
o
6 -
I
MALES
QZ
□J
CQ
I
/
sz
4
/\
2
0
I
i
i
i
2.2 2.4 2.6 2.8
3
i
i
i
i
3.2 3.4 3.6 3.8
i
' I_____ p\
4
4.2 4.4 4.5
BIRTHWEIGHT (IN KG)
Since the frequencies are progressively accumulated, i.e. they
always increase (or at least remain constant) over time, the cumula
tive frequency graph never dips downwards. If no occurrences are
added to the cumulative frequency over a time interval, the graph
line merely flattens out to give a plateau effect.
An example is given in Figure 11.5. Suppose a district has 40
community health workers (1 per 5000 people) and the DHMT has
decided that they should be visited once every two months, making
20 visits to be undertaken per month and a total of 240 in one year.
Figure 11.5 shows that the visits schedule started off well and then
slowed down, but then a special effort was made to carry out more
visits in order to catch up towards the end of the year.
How to draw a cumulative frequency graph
• Obtain a cumulative frequency distribution by adding the figure
in each class of the frequency distribution to all the frequencies in
all the preceding classes (for an example see Section 13.10).
I
• For each class, plot the cumulative frequency at the end of the
class interval on the horizontal axis.
• Join the points with straight lines to produce the cumulative
frequency graph.
5
1
a
118
Manual of Epidemiology for District Health Management
Figure 11.5. Cumulative graph showing number of visits
to community health workers by district
headquarters staff
200cn
oo
>
U-
o
> 100_l
3
0
J
F
M
A
M
J
J
MONTHS
A
S
0
N
D
11.3 Frequency histograms
These diagrams are commonly used for presenting information.
Figure 11.6 shows a typical frequency histogram. An important
characteristic is that the bars of the histogram are contiguous, that
is, one bar immediately follows another with no space between. This
shows that the scale on the horizontal axis is a continuous measure
ment scale.
The shape of the distribution in Figure 11.6 is bell-shaped which
is the sign of a normal distribution. With this type of distribution it
is valid to calculate the mean and the standard deviation for all the
individual haemoglobin values. For a one-sided distribution, such as
that shown in Figure 11.1, it is valid to calculate the mean, but not
the standard deviation (see also Section 10.6).
How to draw a frequency histogram
• The horizontal axis, the X-axis, gives a continuous scale of the
measurement variable while the vertical axis, the Y-axis, shows
the frequency.
• For each class of the grouped data, a bar or rectangle is drawn. The
width of the bar is the same as the class interval used.
i
119
Presenting Health Information
Figure 11.6. Histogram showing the distribution of haemoglobin levels for 1400
adult men and women
I!
Mean Hb 10.4 g/100 ml
25
20
CL
o
CT
JZ
o
15
03
<U
C
0)
CT
2 10
C
03
Q
0)
CL
5
J
2
2
0
5
6
7
8
9
10 11
12
13
IM
15
16
Haemoglobin in grams per 100 ml
11.4 Bar charts
These resemble the frequency histograms in appearance, but they
differ because the bars are not joined together, but separated by a
space. This diagrammatic arrangement is used when the horizontal
axis deals with information that is qualitative or non-continuous in
nature.
; i
Figure 11.7 shows a simple bar chart. It is usual to have the van
able or attribute on the horizontal axis and the frequency on the
vertical axis. When percentages are used, the sum of the heights of
all the bars should be equal to 100%. Occasionally, a bar chart is
drawn in which the frequency is represented on the horizontal axis,
as shown in Figure 11.8.
When two or more distributions involving descriptive variables
need to be compared, a multiple bar chart gives a visual comparison
of the two. Figure 11.9 compares the distribution of registered health
workers in government and private practice. Each group of workers
is represented by a pair of bars, one showing workers in government
service and the other those in private practice.
120
Manual of Epidemiology for District Health Management
Figure 11.7. Bar chart showing distribution of married couples practising
contraception by the main method used
40
30 %
20 -
10
0
BARRIER
METHODS
PILL
STERILIZATION
OTHERS
METHODS OF CONTRACEPTION
Figure 11.8. Bar chart showing the main source of information on health
matters, as reported by individuals in a household survey
PERCENTAGE OF PERSONS
0
r
RADIO/TV
S
NEWSPAPERS
o
z
FRIENDS &
RELATIVES
SCHOOLS
2
o
UJ
1
GO
Z
2
MEDICAL
PERSONNEL
POSTERS
OTHERS
10
"T"
20
—I
i
30
|-
40
50
121
Presenting Health Information
Figure 11.9. Multiple bar chart showing numbers of registered health
workers in government employment and in private practice in
December 1987
3000
<EY
Government
Private
az
2000 -
LU
CQ
51
ZD
Z
o
1000
0
DOCTORS
DENTISTS PHARMACISTS
NURSES
MIDWIVES
TYPE OF HEALTH WORKER
11.5 Pie charts
f
Figures 11.10 and 11.11 show typical pie charts. These are circular
diagrams cut up into several segments or pieces, representing the
frequency distribution of the various groups or divisions of a descrip
tive variable. Pie charts often use percentage distributions, so that a
hemisphere represents 50% (half of the pie) and a quadrant 25% and
so on. To draw a pie diagram requires the use of a compass, and a
protractor for marking out the segments.
The pie chart can also be used for comparing two or more distri
butions (see Figure 4.2). Pie charts are useful for explaining informa
tion clearly to people who are not used to handling numbers.
I
I
4
i
122
Manual of Epidemiology for District Health Management
Figure 11.10. Percentage of the district health budget
spent on primary health care facilities
during one year
PRIMARY HEALTH
CARE 22%
HOSPITALS
58Z
OTHER
22Z
Figure 11.11. Main source of calories for infants aged
6 to 12 months
OTHER 15Z
SUGAR 3Z
CEREALS 4ZA
COW'S MILK 5X
BREASTFEEDING
53Z
STAPLES 20Z
11.6 Scatter diagrams
These are very useful for displaying information on two connected
variables that show a bivariate distribution. For example, when
information is obtained on both the baby's birth weight and gesta
tional age, the two distributions are said to be bivariate.
I
f
123
Presenting Health Information
I
A scatter diagram is formed when the bivariate distributions are
plotted, with birth weight on the vertical axis and gestational age on
the horizontal axis (Figure 11.12). The name comes from the scatter
or spread of the individuals in the sample with respect to the two
variables. In drawing the scatter diagram, each dot on the diagram
represents the pair of measurements made on one baby. Thus the
point marked with a circle in Figure 11.12 represents an infant whose
gestational age was 35 weeks and whose birth weight was 3 kg.
Scatter diagrams are used because they show visually whether an
association or correlation exists between the two variables. The
example in Figure 11.12 shows that there is a positive association
between birth weight and gestational age. An infant with a high
gestational age tends to be heavier at birth than an infant with a low
gestational age. The scatter diagram can only suggest such an asso
ciation: statistical techniques are necessary to measure and test the
actual strength of this correlation (see Section 10.7).
Figure 11.12. Scatter diagram showing the distribution of 50 live newborn
infants delivered in a hospital by birth weight and gestational age
5
cn
E
O
v_
CT
O
O
o 3
@ •
•
X
:e
2 -
Lui
zc
>—
CH
CQ
1 “
0
i
28
30
i
I
32
34
I
I
36
,
I
i
i
i
38
40
42
44
WEEKS OF GESTATION
!
Figure 11.12 also shows that 5 infants had a birth weight below
2500 g, that is, they were low-birth-weight babies. In this sample,
therefore, the low-birth-weight rate is 5 out of 50, or 10%.
Manual of Epidemiology for District Health Management
124
11.7 Maps
Maps of a district are extremely valuable, particularly for showing
a geographical distribution (see Figures 11.13 and 13.3). They can
show, for example, the local distribution of particular diseases (e.g.
leprosy, schistosomiasis) or health programme activities ^e.g. clinic
sites, protected water sources). Maps can show clearly the geographi
cal distribution of cases in an epidemic and the pattern of spread can
suggest which disease is causing the epidemic (see Figure 6.4). Maps
published by central government may show which diseases occur in
the district or demonstrate comparisons between districts for-such
items as infant mortality rates or immunization coverage.
Figure 11.13. Local map showing the transmission sites of Onchocerca volvulus
and Schistosoma mansoni close to villages A, B, C and D
I
RIVER
ShntLlium flies present and
transmi tting Onchocenca
j
Villages
A
B
D
C
|o^W
ROAD
Heal th
Centre
CREEK
B-awhaUiia snails present and
transmitting Schhstosorna mansoni | '
H
0
SCALE F
1
4
2
4 Km
Figure 11.13 demonstrates the value of local maps. During an
analysis of the health centre records it was found that most cases of
onchocerciasis came from village A and some from B, whereas most
of the cases of Schistosoma mansoni infection lived in villages C and
D, with some others coming from village B. An investigation showed
that Simulium flies (the vector of onchocerciasis) were breeding in
the main river, whereas the Biomphalaria snails (an intermediate
host for schistosomiasis) were living in the creek. Once a detailed
local knowledge of disease transmission has been obtained, control
measures can be implemented at the appropriate sites.
I
125
CH A PIER 12
Communicating Health
Information
12.1
Importance of communications
125
12.2
Health reports
126
12.3
Suggested length of reports
129
12.1 Importance of communications
!
?
Too often information on district health matters, whether it
comes from the routine system or from special investigations, re
mains poorly used and unpublished. The findings and conclusions
are not put into a written report and are rarely communicated to
other people. Just as serious is the fact that the information is often
not used by the district management team itself. Thus, sadly, the
full potential of this newly collected information is frequently unex
ploited. Good information, besides being useful, can also be very
influential and persuasive. If DHMTs recognized this and communi
cated their findings and recommendations more effectively, district
health plans could get much greater support from the community
and nongovernmental organizations, as well as from district-level
government and the ministry of health. In this way DHMTs would
be better advocates for health.
Information can be communicated in three main ways:
II
• By writing and disseminating full repons.
GOOD INFORMATION CAN
BE VERY INFLUENTIAL
• Through meetings and discussions with local organizations.
• Through local and mass media.
A report should communicate the findings and recommendations
^;to the:
District authorities.
• District health workers.
• Local community organizations.
• Nongovernmental health organizations.
Manual of Epidemiology for District Health Management
120
• Regional health authorities.
• Ministry of health.
I
I
A written report is fundamental to communicating health infor
mation obtained from analysis of the district health data or from an
investigation or survey, and a good report carries a great deal ot
influence and status.
12.2 Health reports
The style and content of the written reports will depend on their
nature and purpose and on who is likely to read them.
1 ' •.
V-
The following points should always be considered when writing a
report:
• The report should be clearly written using simple language with
short sentences and paragraphs.
• The title should clearly explain what the report is about.
• If possible, the main part of the report should not be longer than
10 pages and include only the most relevant points.
Communicating Health Information
• The draft report should be discussed with colleagues and commu
nity leaders before the final report is written.
I
I
I
i
• A number of drafts may have to be written before the final ver
sion.
A report usually contains the following sections, each of which
should be clearly labelled. Depending on the circumstances, sections
may be subdivided or combined and items may be moved to more
appropriate sections if indicated.
Title page. This should show the full title of the report, the names of
the authors and their positions and addresses.
!
I
■
The title must be clear, e.g. "Malaria in Amber District: the
prevalence of Plasmodium falciparum infection in ten villages in
July 1986."
Summary. This is best placed immediately after the title page. It
should be limited to the main findings and recommendations and
in order to catch the eye of the reader it should be less than two
pages long.
Introduction and purpose. This section should set out the back
ground information and purpose of the study or investigation, to
gether with the reasons for it. Any relevant literature or previous
work can be summarized here.
Objectives must be clearly stated. For example:
• To measure the prevalence of Plasmodium falciparum infec
tion in ten villages in Amber district during 1986.
• To identify high-ask groups for malaria infection with regard
to age, sex, place of residence, and use of mosquito nets and re
pellents.
Methods. The study population should be described first and then
the sampling method, stating the sampling frame, procedure and
sample size. If controls are used, the method of selection should
be explained.
The response rate - the percentage of the original sample eventu
ally seen - should be given, together with the possible reasons
for non-response. The representativeness of the sample and its
comparability to the whole district population should be briefly
discussed.
*
All the variables should be described. This applies not only to the
disease or event being measured, but also to other variables such
as age, sex, ethnic group and occupation.
The operational definition of the variables should be given next,
such as the criteria used to diagnose the diseases, assess age, or to
127
128
Manual of Epidemiology for District Health Management
place a person into a particular occupational group. The proce
dures to elicit these criteria need to be described, for example, the
method of palpating for the spleen in the diagnosis of malaria or
of relating a person's birth to a well-known event to estimate age.
All this is important for the reader to judge the suitability of the
methods and the comparability of the findings to other studies.
The method of information collection should then be described. If
a questionnaire was used, the type (open or closed) should be
mentioned. If possible, a sample of the record form and question
naire should be given in an-appendix to the report. The interview
ers and the training given to them to minimize observer variation
should be described. If observer variation was measured, the
results can be given.
The pilot survey carried out to refine the methods should be de
scribed and the consequent amendments mentioned.
Results. All the relevant findings should be explained in the text.
Clear and brief tables should be used to present the important
data and figures should illustrate particularly important findings.
Tables and figures should be self-explanatory, with clear titles,
legends and footnotes, and each one should show only the one or
two most important points. For the tables, check that the totals
(especially percentages) add up correctly. Numbers normally need
be given to an accuracy of one decimal place only. When there is
more than one line in a graph, each one should be clearly labelled
and differentiated.
Discussion and interpretation. This is where the results are inter
preted and the conclusions made clear. Comparisons can be made
with other relevant studies. The limitations of the study (and
there are invariably some) should be indicated. It is in this section
that the authors can give their own opinions and suggest explana
tions. They may also give their reasons why any controversial
decisions were made.
Recommendations. All recommendations should be listed and num
bered. These will probably include:
• Action to be taken to control a disease or improve a preventive
health programme.
• Other recommendations for future action, which can be di
vided according to the levels at which they are directed, such as:
community organizations;
district authorities;
regional governments,-
ministry of health.
\
129
Communicating Health Information
WHAT ACTION HAS
BEEN TAKEN ON THE
RECOMMENDATIONS?
References, acknowledgements and appendices. These last pages
make up the final details of the report. Any material that you feel
should be communicated, but which is not essential in the main
part of the report, can be placed in an appendix. It is most impor
tant to acknowledge all the people who helped and supported
the work described in the report. To do so is a strength, not a
weakness.
12.3 Suggested length of reports
i
Obviously the actual length of a report will depend on the scope
of the work, but as a guide the following is suggested. Since reports
nearly always come out longer than originally intended, start by
aiming for a maximum of 10 pages.
Number of pages
»
! '
Title page
1
Summary
1
Introduction
2
4
Methods
2
4
Results
3
6
Discussion
2
4
Recommendations
1 - 2
Main body of report
10 - 20
(excluding title page and summary)
130
Manual of Epidemiology for District Health Management
I
131
i
Chapter 13
Epidemiology and District
Health Planning
I
13.1
13.2
13.3
Primary health care and district planning
13.4
Developing district priorities
13.5
13.6
High-risk groups
Improving the provision of health care
Health plans
Present health situation
13.7
Estimating access
• 13.8
13.9
Estimating coverage
Developing the district health plan
13.10 Evaluating progress
13.11
Summary of district health profile
131
133
135
136
138
139
141
141
143
144
146
n
13.1 Primary health care and district planning
The epidemiological responsibilities of the DHMT were outline
in Chapter 1 where they were divided into those concerned with
health information and those concerned with health planning, man
agement and evaluation. This chapter is concerned with the use oi
epidemiology to support these latter functions.
Health planning, management and evaluation for primary health
care are complicated matters, and their organization will vary
considerably from country to country. This chapter can only
present an outline for the use of epidemiology for these activines,
as the details will depend on how districts are organized in each
individual country.
In health planning there are three important questions:
• Where do we want to be in future? - "there"
• Where are we now? - "here"
• How do we get from "here" to "there?"
132
Manual of Epidemiology for District Health Management
The first question—Where do we want to be?—requires the devel
opment of national health policies and plans, stated in operational
terms, based on clear goals, and specific objectives and targets.
To answer the second question—Where are we now? involves
assessing the health status of the population, determining the avail
able resources in terms of health facilities, staff, equipment and
finances, and finally evaluating the access, coverage, efficiency and
effectiveness of the health facilities and their health programmes. It
is most important that community perspectives are also taken into
account in answering this question.
The third question—How do we get from "here' to there ?
involves determining the priorities to be tackled, the health
activities that have to be organized and the management support
that is needed.
In district planning and management, information is needed to
answer these three questions, but the available information is never
as complete or as accurate as would be desirable. In many cases the
DHMT will have to make decisions on the basis of estimates sometimes frank guesses - and it will probably have to set aside
resources to improve the information for next year. However, what
ever the quantity and quality of health information currently avail
able, the DHMT will need to set objectives for programmes for the
coming year.
District planning for primary health care needs to take into ac
count at least the eight main elements of PHC shown below. Some
countries have expanded this list, while others have decided to give
greater priority to some elements than to others. However, all the
elements are important for PHC and the aim should be to make
them all available to the whole district population.
Which of the following elements are being implemented by
the district?
• Education on prevailing health problems and methods of prevent
ing and controlling them.
• Provision of food supplies and promotion of proper nutrition.
• Adequate supply of safe water and provision of basic samtation.
• Maternal and child health care, including family planning.
• Immunization against the major infectious diseases.
• Prevention and control of locally endemic and epidemic diseases.
• Appropriate treatment of common diseases and injuries.
• Provision of essential drugs and supphes.
! •
Epidemiology and District Health Planning
i
The degree to which strategies for the implementation of PHC
in districts are realized frequently depends on national and local
health plans.
The following are commonly adopted strategies:
• Training and use of community members as health workers.
• Community participation in planning and implementing health
programmes.
• Intersectoral coordination, particularly between agriculture,
education, housing, sanitation and water supplies.
I'
• Collaboration between health organizations, particularly govern
mental and nongovernmental agencies and traditional and private
practitioners.
• Decentralization by the ministry of health and the strengthening
of the district health system.
13.2 Health plans
Broad health policies and long-term development goals are gener
ally the responsibility of the national government. Most countries
have accepted PHC as the means of achieving their health goals and
have developed national health plans that take into account their
own particular health problems and the resources available for
achieving health for all.
For example, the ministry of health may state its long-term goals
as follows:
i
• To maximize the total amount of healthy life for the people.
• To ensure that all nationals have ready access to primary health
care.
In order to quantify these goals, the following targets might be set
for the whole country to achieve by the end of the next national
health plan:
• The infant mortality rate should be reduced from the present
national average of 110 to 80 per 1000 live births per year.
• Health facilities should be developed so that 80% of the popula
tion live within 10 km of a facility.
• Supervised deliveries should be increased from the current y
national average of 40% to at least 65%.
i
A target is a specific objective which is quantified and which is to
be achieved over a specified period of time. It is a common practice
to use such targets to indicate what needs to be achieved in the next
I
133
134
Manual of Epide
year or over the next five years. With clearly stated goals and quantihed and time-limited targets, the ministry of health can produce
medium-term or five-year health plans for the whole country incor
porating plans to strengthen the infrastructure of health services and
health personnel development, as well as to improve promotive,
preventive and curative health programmes. It is essential that
targets are formulated as realistically as possible, taking into account
the projected financial resources and available staff. The ministry of
health will need a substantial input from the districts to develop
realistic medium-term plans.
DOES THE DISTRICT HA VE
ANNUAL AND MEDIUM
TERM HEALTH PLANS?
The medium-term plans must in turn be translated into a series of
annual plans, which are used to calculate the budget requests for the
recurrent and capital expenditures for the following financial year.
he DHMT frequently has responsibility for these planning and
budgeting processes.
Some countries do not use plans covering fixed time periods, but
rely on a system of rolling plans. These commonly cover a period of
between three and five years ahead, the details being worked out
only for the next one or two years. At the end of each one or two
years the plans are rolled forward to cover the next year or years
ahead. In this kind of planning system, annual or biennial plans are
the most important.
V'pIanS arC °nly the startinS P°int for improving district
health management. This process can be presented in a simplified
form, as shown in Figure 13.1 below, but it is best understood as a
continuing cycle.
Figure 13.1. Outline of district health management
District
health — ►
P*ans
Allocate
Implement
resources"" ► health
programmes
Improve
Improve
access and — ► health
coverage
status
The district health plans should contain an analysis of the present
situation and the health priorities that need to be tackled On the
basis of this knowledge the DHMT, local government, community
organizations and nongovernmental agencies can then allocate the
money, manpower and facilities needed to implement the necessary
nealth programmes. These programmes should aim to improve ■
access to health care and the coverage achieved, which in turn
should lead to improvements in health status.
Epidemiological skills are thus crucial for members of the DHMT.
These skills are needed to obtain and interpret the health informa
tion necessary for producing the district plans and for guiding the
I
Epidemiology and District Health Planning
allocation of resources, for monitoring the implementation of pro
grammes and for assessing their access and coverage, and finally for
measuring changes in health status.
13.3 Present health situation
The starting point for district planning, for both annual and
medium-term plans, is a detailed analysis of the present situation
in the district in order to produce the district health profile, which
helps to answer the planning question,Where are we now? This
profile allows today's situation to be seen as a starting point for
planning improvements. However, the DHMT needs to be realistic
about where it is now and how much it can achieve over the next
few years.
I
I
lI
I
■
i
To produce a useful district health profile requires a combination
of a good local knowledge of the district and all the available health
information. The DHMT needs such details as those given in Chap
ter 3 on the district population and in Chapter 4 on health status. To
. these must be added information on the district health services and
health programmes, covering the eight essential elements of PHC. A
list of useful indicators has been presented in Chapter 2 and a check
list for producing a district health profile is given at the end of this
chapter (see Section 13.11).
Health indicators are essential for analysing the present situation
in the district, for expressing specific targets and for assessing
whether these targets are being met or not. Keeping the district
health profile up to date should preferably be a continuous activity,
or at least an annual one. The information should be analysed, dis
played and widely communicated - as suggested in Chapters 10, 11
and 12. Visual displays showing progress in the indicators should be
given wide publicity.
The indicators shown in the profile are the basic ones that can be
quantified and should be available. In addition, there is a great deal
of other information about the district, such as its history, geogra
phy, socioeconomic conditions and political system, that the DHMT
will also need to know in order to make realistic health plans.
!
The district health profile lists important indicators under the
following categories:
• District population.
• Health status.
• Health resources.
• Health programmes.
135
13B
Manual of Epidemiology for District Health Management
ANALYSE THE PRESENT
SITUATION USING THE
HEALTH INDICATORS IN
THE DISTRICT HEALTH
PROFILE
The indicators for the district population were covered in Chapter
3. The health status indicators cover nutritional status, morbidity
and mortality- The health resources indicators are for facilities, staff
and finances. The health programme indicators cover pregnancy and
delivery, child care, environmental health and clinical care. A simi
lar listing of basic indicators will probably be accepted by most
ministries of health, but every DHMT will need to work out the
details of any additional indicators it wishes to use.
13.4 Developing district priorities
The next step in district health planning is to answer the
question: How do we get from "here" to "there"? An analysis of the
present situation, as shown by the district health profile, should
provide a basis for this.
The analysis should identify the following:
• Main health problems.
• High-risk groups.
• Access to and coverage by health programmes.
• Organization and management of these programmes.
Districts will always have insufficient health resources; so it is
important that the DHMT decides, on the priorities for developing
PHC in the district. Choices must be made: which population
groups, diseases or underlying health problems should be given
priority? Which health programmes should receive more attention
and more resources? Good epidemiological health information is
necessary7 to help answer these questions.
PLANNING INVOLVES
DECIDING ON PRIORITIES
Making these choices is a difficult process; full consideration
must be given both to the priorities in the national health plan of the
ministry of health and to local priorities decided on by the district
itself. Political, social and economic factors, as well as the present
health status, must enter into the decision-making. Although the
overall mix of health services provided in the district will be deter
mined at the national level, each district will have its own special
problems. Many important decisions on priorities must be made at
the district level, by the DHMT, and involving community represen
tatives, local councils and the planning offices of the local govern
ment.
Health plans must take into account the need for a balance be
tween the following components:
• Promotive and preventive health care, such as community pro
grammes for health education, mother and child health care,
immunization, nutrition and environmental health.
137
Epidemiology and District Health Planning
r
i
• Preventive disease control, with specific programmes for commu
nicable diseases such as tuberculosis, malaria, cholera, rheumatic
heart disease and for selected noncommunicable diseases such as
hypertension and accidents.
• Curative services for ill people, based on primary health care and
care at the first referral level and/or district hospital.
• Reasonable access to these services by the district population,
through community health posts, health centres and other out
reach services.
• Development and upgrading of health workers through training
and retraining programmes.
I
• Adequate transport, communications and supphes for all these
activities.
There is no single "right" way to decide which health interven
tions or programmes should receive priority, but some "rules of
thumb" have proved helpful. Priorities should be developed in terms
of the activities that will have the greatest impact on improving the
■health status of the district. The DHMT will need to develop pro
grammes for carrying out these activities effectively and efficiently.
A priority chart is a useful method for organizing health information
for use in deciding on priorities. Each disease is given a simple score
for its relative importance (based on its frequency, morbidity and
mortality); for the effectiveness of the possible interventions; and for
the costs of these interventions. The total score for the three aspects
provides an approximate guide to the priority of each intervention.
Each of the three aspects is given an appropriate score as follows:
Score
+
Relative importance of disease
low
moderate
high
Effectiveness of interventions
ineffective
moderate
very effective
high
moderate
low
Cost of interventions
Examples are shown in Table 13.1.
It is assumed that priority should be given to diseases or underly
ing health problems that are frequent, severe and cause high morbidity and/or high mortahty, and against which there are effective and
cheap interventions. Epidemiological knowledge and experience are
essential for making use of such a priority chart. Assigning scores is
a matter for consultation and discussion among members of the
DHMT and other health staff, and between community representa
tives, local government officers and staff in other government sec
tors. For convenience this chart has been organized for diseases, but
planning decisions must also be made on how to improve the appro
priate interventions and health programmes. Some programmes,
138
Manual of Epidemiology for District Health Management
such as those for immunization or the provision of water supplies,
cover several different diseases and for these programmes the appro
priate disease scores can be added together to arrive at a priority for
each programme or set of activities.
A major advantage of such a scoring system is that it uses all the
epidemiological information collected in the district and focuses the
discussion on each major health problem in turn. Some communi
cable diseases, such as African trypanosomiasis and dengue, may
have a low frequency but they can cause severe epidemics and hence
long-term control programmes may be given a high priority.
Table 13.1. Chart for determining health priorities, with examples
Disease
Relative importance
(based on frequency,
morbidity and mortality)
Effectiveness
Costs
Priority
score
Measles
9
Diarrhoeas
8
Malaria
Tuberculosis
-H-
Cerebral vascular
accidents
Leukaemia
+
+
-H-
7
-H-
6
4-
5
+
3
13.5 High-risk groups
In working out priorities for annual and medium-term plans
another method is to consider which groups of people should receive
priority in the district. Every man, woman and child is at risk of
getting ill and dying, but some groups are at a much higher risk.
These are called high-risk groups. Not only are some people at high
risk for some diseases but they may also make less use of the avail
able health services - that is, they are also at high risk of not getting
the necessary health care. Epidemiological information is essential
for defining these high-risk groups, which can be recognized by
common features of Who? Where? and When? Health plans can give
priority by selecting certain high-risk groups and thus concentrating
on the development of PHC for these groups.
High-risk groups include:
• Women aged 15-44 years, who comprise about one-fifth of the
total population.
Epidemiology and District Health Planning
• Infants and young children, also about one-fifth of the total popu
lation. A high proportion of all deaths from malnutrition and
communicable diseases, such as measles, diarrhoeas, pneumonias
and malaria occur in this group.
• Some workers, particularly those working with machinery and
using dangerous chemicals, as well as labourers and consmicticn
workers.
• Old people, who suffer from chronic diseases.
• Contacts of people with infectious diseases, such as tuberculosis
and leprosy.
—
• Certain cultural and socioeconomic groups, such as poor famil-rs,
people of low caste, subsistence farmers and recent migrants
• Certain ethnic groups or subgroups that are predisposed to high
risk by their beliefs and customs.
WHO ARE THE PEOPLE AT
HIGH RISK?
• People living far away from health facilities, who are at high risk
of not getting a service they require.
. • People living in areas affected by seasonal and climatic changes
such as areas where the incidence of malaria increases consider
ably in the rainy season .
13.6 Improving the provision of health care
There is no standard way to measure and compare the provision of
health care, but two imponant concepts are those of access and
coverage. Access measures the proportion of the district's populate a
that has a particular facility within reasonable reach, which may be
measured by distance (e.g. 5 or 10 km), time (e.g. 1 or 2 hours' travel
ling), costs (e.g. travel fares and health service fees), and social and
cultural factors (e.g. caste or language bamers). Access to a health
service within a reasonable distance must be a high priority for all
district health planning for PHC.
Coverage is a measure of the percentage of people or households m
need of a health service or facility who actually receive it, for examrie
the percentage of households with a safe water supply or the percent
age of pregnant mothers who have attended for antenatal care.
Epidemiological information is necessary to find out both the
denominator, such as total number of households or pregnant women
in the district, and the numerator, such as the number of houses
with a water supply or the number of pregnant women who have
attended an antenatal clinic at least once. The expanded programme
on immunization (EPI) commonly measures coverage by using the
percentage of young children who have been immunized, with, sav
three doses of DPT (diphtheria-pertussis-tetanus) vaccine.
*
139
140
Manual of Epidemiology for District Health Management
IMPROVING ACCESS AND
COVERAGE IS A HIGH
PRIORITY
It is clear that coverage can only be high if access is high. For
example, EPI coverage will remain low if a high proportion of people
do not have reasonable access to a health centre or subcentre offering
immunizations. In this situation a district might have to use mobile
clinics and mass vaccination campaigns in order to achieve a satis
factory coverage.
Information for calculating the denominators is usually derived
from information on the district population, as described in Chapter
3. Information for calculating the numerators comes from figures
provided by official government pubheations and, in the district,
from the routine information system, surveillance, and from special
investigations and surveys.
Figure 13.2. It takes about one hour to walk 5 km
4
I
ft
/i'
I
141
Epidemiology and District Health Planning
13.7 Estimating access
I
A major priority must be for a high proportion of people to have
reasonable access to a community health worker, a health post or a
health centre, say within 5 or 10 km of their home, which is the
equivalent of 1 to 2 hours' walking. In rural areas distance is com
monly used to reflect access, because of poor communications and
transport. In urban areas and where transport is available, it is fre
quently more important to define access in terms of time, costs or
social factors.
A useful method for estimating geographical access is to draw 5or 10- km circles around each health facility on a district map and
then to calculate the percentage of the total district population who
live within those circles. This is illustrated in Figure 13.3. In this
example, it was found that 75% of the district population lived
within 10 km of a health facility and 50% were within 5 km.
Figure 13.3. District map showing population distribution and those living
within 10 km of a health facility
Scale
0
20
40
60 Km
Heal th
facility r.
at centre.of
:I
13.8 Estimating coverage
To estimate coverage, it is most important that members of the
DHMT learn to use and interpret the health indicators that may
already be available for the district. The most important step is
to have reliable information on the expected number of people,
events or attendances in the district and then to calculate the actual
coverage the district is achieving. The district coverage can then be
1
142
Manual of Epidemiolog
compared with the national situation. Rates derived from special
censuses and surveys are likely to be more reliable than those based
on the routine health information reported by health facilities or on
surveys carried out on a small sample.
The accuracy of routine data is sufficient for most planning,
management and evaluation purposes at district level. Sometimes,
however, special surveys are justified if some important information
is missing.
Example: In a district with 200 000 people, about 4% are less than
one year old (as given by the national census), which gives a total of
about 8000 infants. The district routine health information system
reported that 2400 children under one year of age received 3 doses of
DPT vaccine during the year.
'
Coverage for 3 doses DPT
2400
----- x
8000
100
30%
If the national coverage is known to be 45%, and the long-term
objective is to reach 80%, the DHMT will need to ask itself some
serious questions, first about access to health care and secondly on
the management of the district immunization programme.
The achievement of a high coverage for the main services and
programmes is the single most important managerial objective for
the DHMT. For example, the following coverage indicators are given
in part 4 of the district health profile in Section 13.11:
• Percentage of all births attended by a trained health worker.
• Percentage of 1-4-year-old children weighed regularly.
• Percentage of all cases of pulmonary tuberculosis who are receiv
ing treatment.
• Percentage of households with an adequate water supply.
• Percentage of mamed couples who are currently using a modem
method of family planning.
By estimating coverage, the DHMT is analysing the present situ
ation and obtaining a measure of how well it is meeting the opera
tional targets that were established the previous year. The next
step is to fix a target for the improvement to be achieved during the
following year. However, if the DHMT decides that the district
coverage of a particular service is reasonable, the objective may be
to maintain the same level of coverage for next year. Annual
plans may then give higher priority to achieving improvements in
other services.
I
4
i
Epidemiology and District Health Planning
13.9 Developing the district health plan
If the DHMT has used the available epidemiological information
and followed the above steps in health planning, it should by now
have completed the following planning processes:
• Analysed the present situation, including health status, in the
district.
• Developed the priorities for the next annual and medium-term
plans.
• Decided on which high-risk groups should receive priority.
• Made plans to improve access and coverage for the priority health
programmes.
• Decided on the objectives and indicators to evaluate progress.
Once the district health problems have been identified, the main
priorities established and the objectives and indicators chosen, the
next step is to develop the framework for the medium-term plan.
Only then should the more detailed annual plans be drawn up. The
annual plans are the means of achieving the medium-term objec
tives.
II
Once the processes have been completed, the DHMT should be in
a good position to plan all the health programme activities that are
needed to make the health plans work.
DECIDE ON PRIORITIES.
SET OBJECTIVES, SELECT
INDICATORS AND PLAN
ACTIVITIES
I.
I
1
u
It is important to stress that health planning is a complicated and
dynamic process. What is outlined here is a logical sequence, but
planning does not take place in a purely logical way. The DHMT is
advised to select a few important problems and not to attempt to be
comprehensive. If ail the problems are taken together they will
present an impossible task and district health planning may not even
be attempted.
The best approach, therefore, is to tackle only one or two priority
problems at a time. The following is a useful sequence for developmg
a plan of action, but for more guidance the DHMT will need to
consult their national health plan and the Ministry of Health's
guidelines for district planning.
• Choose a strategy. There may be several different ways to achieve
the objectives. For instance, plans to reduce maternal mortality
might include increasing the proportion of supervised deliveries at
health centres, training traditional birth attendants, improving
antenatal attendance rates and identifying high-risk mothers. As
planning proceeds, the practical constraints become clearer and
the DHMT may find that it has to revise its original strategy.
143
144
Manual of Epidem
• Undertake consultations on the proposed strategy. The commu
nity representatives and organizations, nongovernmental health
organizations, other sectors and local governments will all need to
comment and collaborate if PHC is to be successful. Local politi
cal parties and politicians can give powerful support.
• Identify all the necessary activities. Sort out all the activities and
tasks and determine the implications for staffing, facilities, sup
plies, transport and budgets. Consider also the implications for
community participation, intersectoral coordination and the col
laboration of all the other health care providers.
• Establish a timetable. Estimate the time required for all the ac
tivities and then fix starting and completion dates. These should
be realistic and, if possible, contained in the annual plan.
• Assign responsibility to staff. The DHMT will need to identify
individuals and organizations that will be responsible for carrying
out the different activities needed to make the strategy work.
• Allocate funds. Funds are always limited, so the DHMT needs to
estimate costs very carefully. Recurrent and capital costs are
usually estimated separately.
• Monitor and evaluate progress. It is not sufficient just to plan, it is
also necessary to find out whether the plan works. Can the rou
tine system give the necessary health information? Is epidemiol
ogical surveillance or a special survey required?
• Display and communicate the plan. It is most important that
everyone involved understands the health plan, including all the
activities and time schedules, and knows the names of the respon
sible staff. Maps, charts and graphs can be very useful for this dis
play. Opportunities should be taken to communicate the plans at
meetings. Local government should also be kept well informed.
The district health profile (see Section 13.11! is the most impor
tant means of demonstrating progress.
13.10 Evaluating progress
Implementation of the district plan for primary health care can be
evaluated in two main ways. The first is to assess what programme
activities have been achieved compared to what the DHMT proposed
in the district plan. The second is to see if the indicators of health
status have improved or if the frequency of disease or underlying
health problems has been reduced. Well planned activities have to be
implemented over several years to achieve an improvement in over
all health status indicators or a reduction in the frequency of many
diseases or underlying health problems. The indicators contained in
the district health profile are helpful for charting progress.
145
Epidemiology and District Health Planning
1
CHART PROGRESS USING
HEALTH INDICATORS
The DHMT should be able to evaluate its progress in implementing
health programme activities, but it is usually a much more difficult
problem to assess changes in health status. Although epidemiological
health information must be used in evaluation, it is also important
for the DHMT to combine this with its own knowledge and experi
ence. It is this combination of information with experience that will
enable the DHMT to make judgements and draw conclusions as to
how well PHC is being implemented in the district.
The main steps in undertaking an evaluation as part of district
health management are:
■■
• Select the necessary indicators for the health activities.
• State the objectives to be achieved in terms of the indicators.
• Collect the necessary epidemiological health information.
• Compare the results achieved with the targets.
HAVE THE OBJECTIVES
BEEN ACHIEVED?
• Judge the extent to which the targets have been met.
• Review the strategy and district health plans and make new
annual plans for next year.
Where a series of activities has to be performed over the twelve
months, a useful technique for monitoring progress is to use a cumu
lative graph. The total activity to be performed in one year is divided
by 12 to give the target for each month. Plotting the target values
and the cumulative graph by monthly totals shows the progress
being made each month.
i
Example: In a district with 200 000 people there are 4.0% under
one year old, which makes a total of 8000 infants. The DHMT wants
to improve on last year's average EPI coverage of 30% to meet the
national figure of 45% coverage with three doses of DPT vaccine.
Thus, the total number of infants to be fully immunized by the end
of the year is:
No. of infants to be immunized per year
8000
x
No. of infants to be immunized per month
3600
x
45
100
12
3600
300
The actual numbers of infants fully immunized with 3 doses of
DPT are as follows: January 310, February 300, March 280, April 240,
May 200, and June 170. Thus, the cumulative totals are as follows:
i
Monthly total
Cumulative total
January
310
310
Feoruary
300
610
March
Ami
280
240
890
1130
May
200
June
170
1330
1500
148
Manual of Epidemiolo
The cumulative graph is shown in Figure 13.4. The average for
January to June was 250 infants immunized per month. The immu
nization started off well but then started to fall short of the objective
of 300 infants per month. This suggests that the DHMT will have to
review the programme activities to find out why the programme is
not working well. The alternatives are to put in more effort and to
attempt to achieve a new monthly total of 350 per month for the rest
of the year, or to decide that the annual objective is too ambitious
and that it is only reahstic to immunize 250 infants per month,
which is equal to 3000 per year or an estimated coverage of about
38%. This would be an improvement on the previous year but still
falls short of the national average.
Figure 13.4. Cumulative graph showing number of infants fully immunized by
month compared to the objective of an average of300 per month
over one year
4000 n
_j
3000-
c
o
h-
2000 -
c
-J
ZD
s
ZD
CD
1000 -
0
J
F
M
A
M
J
; MONTHS
—I
J
A
S
0
N
D
13.11 Summary of district health profile
This
ims section presents a list of indicators for use in compiling
information for a district health profile to be used in health planning,
management and evaluation. The indicators have been arranged
as follows:
J
I
District Population
Health Resources
Total
Facilities
By age group
Personnel
Births
Finances
Fertility
Health Status
I
147
Epidemiology and District Health Planning
Health Programmes
Pregnancy and delivery
Nutrition
Child care
Morbidity
Environmental health
Mortality
Clinical care
This profile should be adapted to local circumstances. The list
contains only selected indicators and is not meant to be fully com
prehensive. It also needs to be emphasized that not all the possible
health programmes have been included and only four are given as
examples. The years 1 to 5 are shown to draw attention to the im
portance of annual and medium-term plans, for which targets can be
set and for which health information can be kept and updated as
time passes.
r
District Health Status and Health Planning Profile
I
1. DISTRICT POPULATION
A. Total estimated
B. Total by age group
0-11 months
1- 4 years
5-14 years
l
15-45 years
45+
C. Total births estimated
Crude birth rate
General fertility rate
D. Rate of growth
Year
Year
Year
Year
Year
1
2
3
4
5
148
Manual of Epidemiology for District Health Management
Z HEALTH STATUS
A. Nutritional status
Infants
% with low birth weight
Children
Weight forage
No. between 3rd and 50th centile
%
No. below 3rd centile
%
No. not making progress
(Less than 50%)
%
1-4 year olds
Upper-arm measurement 12.5-14 cm
No.
%
Upper arm circumference less than 12.5 cm
No.
%
B. Morbidity
Selected specific diseases by
number of new cases diagnosed
Diarrhoea
Jaundice
Malaria
Measles
Neonatal tetanus
Typhoid
Whooping cough
Diseases requiring prolonged therapy
Tuberculosis
No. of new cases expected
No. of new cases diagnosed
No. of cases completing therapy
No. of cases defaulting
No. of cases currently under therapy
Year
Year
Year
Year
Year
1
2
3
4
5
149
Epidemiology and District Health Planning
i
Leprosy
No. of new cases expected
No. of new cases diagnosed
No. of cases completing therapy
I
No. of cases defaulting
No. of cases currently under therapy
Epidemic diseases
Meningococcal meningitis
Cholera
Dengue
Trypanosomiasis
Yellow fever
C. Mortality
Total deaths
Total deaths estimated for district
Crude estimated mortality rate
Expectation of life
No. of deaths registered
Estimated percentage registered
No. of deaths certified
Estimated percentage certified
Estimated age-specific death rates
Infants
1- 4 years
5-14 years
15-44 years
45+ years
Maternal mortality
i
No. of deaths registered
No. of deaths expected
Disease-specific mortality
diarrhoea
malaria
malnutrition
measles
Year
Year
Year
Year
Year
1
2
3
4
5
150
Manual of Epidemiology for District Health Management
pneumonia:
0-4 years
5+ years
tuberculosis
3. HEALTH RESOURCES
A. Facilities
Health stations, posts, etc.
No. needed for district
No. operating
% of those needed
% of population with access
Hospitals
Total no. of hospital beds in district
Population to bed ratio
B. Personnel
Primary health care
Workers needed
Workers in post
No. of TBAs in district
No. of TBAs trained
% trained
Hospital care
No. of professional nurses needed
No. of nurses in post
%
No. of practical nurses needed
No. of practical nurses in post
%
No. of doctors needed
No. of doctors in post
%
No. of laboratory and technical personnel needed
No. in post
%
Year
Year
Year
Year
Year
1
2
3
4
5
151
Epidemiology and District Health Planning
ii
r
Year
1
Year
Year
Year
vear
3
4
5
2
C. Finances
Total health system expenditure in district
Per capita expenditures
Expenditure on primary care
Expenditure on hospital care
r
I
4. HEALTH PROGRAMMES
A. Management of pregnancy
Total no. of expected deliveries in district
No. of pregnant women who received
prenatal care
% receiving prenatal care
No. of deliveries attended by
trained TBAs or health personnel
No. of deliveries in health centres or hospitals
% attended by trained personnel
Estimated no. of couples using
birth spacing/family planning (FP) services
No. of couples accepting birth spacing
services for first time
No. of women aged 15-45
% of couples continuing FP
(compared to all women aged 15-45)
B. Child care
Infants
Immunization
No. of infants born in the year
No. receiving DPT vaccine
I
I
% receiving
1 dose
2 doses
3 doses
No. receiving polio vaccine
% receiving
1 dose
2 doses
3 doses
No. receiving measles vaccines
%
No. receiving BCG vaccine
M p--t too
06316
/c
%
C l!
DOC’
-------------- K—
15Z
Manual of Epidemiology for District Health Management
Year
Year
Year
Year
Year
1
2
3
4
5
Antimalarials
No. of infants receiving
antimalarials regularly
% of infants receiving antimalarials
Nutritional assessment
No. of infants weighed at least
once during year
% weighed at least once
No. of infants weighed at least six times
% weighed six times
1-4 year olds
Antimalarials
No. of 1-4 year olds
No. receiving antimalarials
% receiving antimalarials
Nutritional assessment
No. of 1-4 year olds assessed regularly
% assessed regularly
C. Environmental health
Water supply
Total no. of communities or households
No. with adequate water supply
% with adequate water supply
Excreta disposal
No. with adequate control
% with adequate control
I
Refuse disposal
No. with adequate disposal
% with adequate disposal
■
-asI
-------------_______________ J
I
l
153
Epidemiology and District Health Planning
Year
1
D. Clinical care
Clinic visits
No. of visits to village-level clinic
Population per clinic
Subcentre visits No./yr
No. referred from village
% of visits referred from village
Population in subcentre area
Health centre visits No./yr
No. referred from subcentre
% referred from village and subcentre
Total no. of outpatient clinic visits
No. of clinic visits per person per year
Hospital admissions
No. of hospital (non-maternity) inpatients
No. of hospital days
No. of hospital admissions for
preventable causes
No. of admissions
Infants
1- 4 years
5-14 years
15-44 years
45+ years
No. of deaths in hospital
Infants
1- 4 years
5-14 years
15-44 years
45+ years
No. of maternity admissions
No. referred because of risk factors
No. with unexpected complications
No. of maternal deaths in hospital
Year Year Year Year
4
3
5
2
154
Manual of Epidemiology for District Health Management
!
I 'V
■
-
■.
r
I
I
155
CHAPTER 14
A B C of Definitions and Terms
The definitions given in this chapter are valid as they are used in this publication but different
definitions may be used in other contexts. This chapter is largely based or/A dictionary of
epidemiology, edited by J.M. Last for the International Epidemiological Association and published
by Oxford University Press, 1983.
I,
I
i
Access. The proportion of a defined population that has a particular
facility within reasonable reach, which may be measured by
distance, time, costs or social and cultural factors.
Accuracy. The degree to which a measured value represents the true
value of the variable that is being measured. See Repeatability and
Validity.
Agent. A factor whose presence or deficiency is essential for the
occurrence of a disease, e.g. microorganisms, chemical substances,
vitamins and essential amino acids.
Age-sex pyramid. See Population pyramid.
Age-specific rate. A rate for a specified age group, with the numera
tor and denominator for the same age group.
I
1
e.g. 1 -4 year mortality
Number of deaths among 1-4 year old children
in area in one year
= --------------------------------------------Average total population aged 1-4 years
in same area in same year
x 1000
Airborne infection. A disease caused by an infectious agent capable
of being transmitted by particles or droplets suspended in the air,
e.g. measles and pertussis.
I
I
Arbovirus. A group of diverse animal viruses that are transmitted to
humans by blood-feeding arthropod vectors, such as mosquitos,
ticks, sandflies and midges. The term is an abbreviation of "ar
thropod-borne virus".
Association. Statistical dependence between two or more variables,
which are said to be associated if they occur together more fre
quently than would be expected by chance. Statistical tests enable
the degree of association to be calculated.
I
i
I
i
15B
Manual of Epidemiology for District Health Management
Attack rate. This rate usually refers to the incidence of new cases
during an epidemic. The secondary attack rate is based on the
number of new cases among contacts of a primary case that occur
within the accepted incubation period of the disease. The denomi
nator is the total number of exposed contacts during the same
period of time.
Average. See Mean, arithmetic
B
Bias. Any effect during the collection or interpretation of informa
tion that leads to a systematic error in one direction, e.g. errors
resulting from weighing scales under-recording a child's true
weight, or observer bias in the interpretation of rephes to ques
tions in a questionnaire.
Birth rate. A summary crude rate based on the number of live births
in a known population over a given period of time.
Birth rate
Number of births in area during one year_________
Average total population in area during the same year
x 1000
Birth weight. Infant's weight recorded at the time of birth. Lowbirth-weight babies weigh less than 2500 grams and the percent
age of such babies is commonly used as a general measure of
health status.
c
Carrier. A person or animal that has a specific infectious agent in
the absence of clinical disease and that serves as a potential
source for the further transmission of the infection.
Case. A person who is identified as having a particular characteristic
such as a disease, behaviour or condition. An epidemiological
definition of a case is not necessarily the same as the clinical defi
nition. Cases may be divided into possible, probable and definite,
depending on how well specific criteria are satisfied.
Case-control study. An analytical epidemiological study that com
pares cases of a particular condition with suitable control sub
jects, who do not have the condition, looking at the frequency of
associated factors in the two groups. Sometimes also called a
retrospective study. Often used to test hypotheses about etiology,
e.g. the link between lung cancer and cigarette smoking.
Case-fatality rate. The percentage of persons contracting a disease
who die from it. This rate is most commonly used for commnnicable diseases.
157
Definitions and Terms
Case-fatality rate
=
Nu-mber of deaths from a disease in a given period
---------------------------------------------------------- x
Number of cases of disease diagnosed in the same
period of time
100 S
Catchment area. The geographical area from which the people
attending a particular health facility come.
1
Census. The enumeration of an entire population, usually with
details being recorded on residence, age, sex, occupation, ethnic
group, marital status, birth history and relationship to head of
household. A de facto census only counts the people who are
actually present during the enumeration, whereas a de jure census
records all people by their normal place of residence at the time of
enumeration.
Chemoprophylaxis. The administration of drugs to prevent infec
tion from occurring or to prevent the infection from progressing
into disease.
Class. A group of observations made on a variable, considered to
gether for the convemence of analysis, e.g. haemoglobin values
may be classed by intervals of Ig/dl.
Clustering. The grouping of a series of cases in relation to time or
spacial area or both. The space-time clustering of cases in an epi
demic commonly indicates a point-source outbreak due to an in
fectious agent or toxic chemical.
Cluster sampling. A sampling method in which each unit selected is
composed of a group of persons rather than an individual, e.g.
villages and households.
Cohort. A well defined group of people who have had a common
experience or exposure, who are then followed up for the inci
dence of new diseases or events, as in a cohort or prospective
study. A group of people bom during a particular period or year is
called a birth cohort.
Communicable period. The time during which an infectious agent
may be transferred from an infected person to another susceptible
person, or from an animal to man or vice versa.
Confounding. A situation in which the effects of two variables are
difficult to separate from each other, e.g. level of family income
and availability of food as causes of malnutrition.
Contact. Exposure to a source of an infection. Transmission due to
direct contact may occur when skin or mucous membranes touch.,
as in body contact, kissing and sexual intercourse.
Contagious. Transmitted by contact or close proximity.
Control. Disease control programmes aim to lower the incidence of
new cases, or reduce the proportion of severe cases through treat-
I
I
si
158
Manual of Epidemiology for District Health Management
ment, to an acceptably low level, so that the disease is no longer
considered a major public health hazard.
Control group. Comparison group of people who do not have a
particular disease or condition or who have not been exposed to
the disease, intervention, procedure or other variable that is being
studied. Neighbourhood controls, which are commonly used for
convenience, are people who live in the same neighbourhood.
See also Case-control study.
Correlation. A measure of association that indicates the degree to
which two or more sets of observations fit a linear or straight-line
relationship. Correlation may be positive, when both variables in
crease together, or negative, when one increases as the other de
creases.
Coverage. A measure, usually expressed as a percentage, of people or
households who have actually received a particular service com
pared to all those who need it, e.g. percentage of households with
a reasonably safe water supply, percentage of infants immunized
with three doses of DPT vaccine.
e.g. obstetric coverage =
Number of deliveries
attended by a qualified health worker
Total expected number of deliveries in the same
populationdunng the same period of time
x W0%
Cross-sectional survey. A survey or study that examines people in a
defined population at one point in time. Cross-sectional surveys
usually supply prevalence data but repeated surveys can be used
to give an estimate of incidence.
D
Data processing. Conversion of raw data into a form suitable for
analysis with computers and statistical programmes.
Death rate. The proportion of a population who die from any cause
during a specified period of time. The rate can be made specific for
a particular cause, or group of causes, of death. The rate can also
be calculated for each sex and for any age group, thus providing
disease-, sex- and age-specific rates.
Crude death rate
Number of all deaths during one year
Average total population during same year
x WOO
Demand for health care. Willingness and/or ability to seek and use
services. Sometimes further divided into expressed demand or
actual use and potential demand or need.
Definitions and Terms
i
Demography. The study of populations, with reference to such
factors as size, age structure, density, fertility, mortality, growth
and social and economic variables.
Denominator. The lower portion of a fraction. In the calculation of a
rate, this represents the total population at risk.
Disease, subclinical. The condition in which a disease is only de
tectable by special tests and there are no apparent symptoms and
signs.
E
Endemic. The constant presence of a disease or infectious agent in a
given population or geographical area. Also used to refer to a
disease with a constant incidence of new cases in the area.
Epidemic. The occurrence in a community or region of cases of an
illness or other similar event clearly in excess of what is normally
expected. The characteristics of the illness, the area and the sea
son all have to be taken into account. To judge whether there is
an excess or not requires knowledge about the previous incidence
of the event in the same area.
I
Epidemiology. *The study of the distribution and determinants of
health and disease in populations and its apphcation to the pre
vention and control of health problems and diseases.
Eradication. The extermination of an infectious agent, thus halting
transmission of infection e.g. smallpox has been eradicated
throughout the world and malaria has been eradicated from cer
tain areas.
Evaluation. A process that attempts to determine as systematically
and objectively as possible the relevance, effectiveness and impact
of activities in the tight of their objectives. Evaluation is often
carried out separately for inputs, processes, outcomes and impact.
Expectation of life. The average number of years an individual is
expected to live if current mortality trends continue. Life expec
tancy at birth is the average number of years a newborn baby can
be expected to live under existing conditions. As many deaths in
developing countries occur during infancy and childhood, the
average life expectancy in these countries is much lower than in
developed countries.
F
False negative. A false result in a screening test, leading to the
classification of a person, who is actually positive, as negative or
normal.
159
180
Manual of Epidemiology for District Health Management
False positive. A positive test result in a subject who is actually
negative, i.e. a healthy person is wrongly said to have a particular
disease or attribute.
Fertility rate. See General fertility rate.
Fetal death rate. Also called stillbirth rate. The number of fetal
deaths in one year expressed as a proportion of all births (live plus
stillbirths) in the same year.
No. of fetal deaths in one year
Fetal death rate
No. of fetal deaths plus live births in same year
x
1000
G
General fertility rate. Similar to the crude birth rate, except that the
denominator is restricted to women of childbearing age, i.e. 15-44
years.
General fertility rate
=
No. of live births in area in one year
Average female population aged 15-44 years for
same area and year
x
1000
Growth rate of populations. Also known as the natural rate of population increase. In the absence of the effects of migration, it is
calculated as the crude birth rate minus the crude death rate.
H
Health. A state of complete physical, mental and social wellbeing
and not merely the absence of disease or infirmity.
Health indicator. A measure that reflects, or indicates, the state
of health of persons in a defined population, e.g. the infant
mortality rate.
Health information system. A combination of health statistics from
various sources, used to derive information about health status,
health care, provision and use of services, and impact on health.
Herd immunity. The resistance of a group or community to inva
sion and spread of an infectious agent, due to the resistance to
infection in a high proportion of individual members of the group.
The herd immunity results from the lowered probability of the
disease agent being transmitted from an infected person to a
susceptible one when a high proportion of individuals are
not susceptible.
Holoendemic. Describes a disease that is virtually universal in
the population, with symptoms in childhood, leading to a state of
equilibrium and a lower incidence of symptoms in adults, e.g.
malaria in some communities, especially in Africa.
Definitions and Terms
181
Host. A person or animal that is infected under natural conditions.
A number of microorganisms and parasites may have several
different hosts.
Household interview survey. The collection of information from a
representative sample of households by trained interviewers. It is
usually a cross-sectional survey to collect information about indi
vidual members and on common features, e.g. water supply.
Hyperendemic. A disease that is constantly present at a high inci
dence (or prevalence) and that affects all age groups.
I
Incidence. The number of new cases or events or attendances occur
ring in a defined population within a given period of time, com
monly one year.
I
Incidence rate. A measure of the rate at which new cases or events
occur in a defined community.
Incidence rate
No. of new cases or events
diagnosed in population in one year
Average total population at risk in same area in one year
x WOO
Incubation period. The time interval between invasion of a suscep
tible host by an infectious agent and the appearance of the first
symptom or sign of the disease.
Infant mortality rate. A measure of the rate at which deaths occur
in children less than one year old.
Infant mortality rate
No. of deaths in children
less than 1 year old in one year
No. of live births in same year
x
1000
M
Maternal mortality rate. A measure of a woman's risk of dying from
causes associated with pregnancy. A maternal death is the death
of a woman while pregnant or within 42 days of the termination
of pregnancy, irrespective of the duration and the site of preg
nancy, from any cause related to or aggravated by the pregnancy
or its management but not from accidental or incidental causes.
Some countries have extended the period of 42 days to up to
one year.
Maternal deaths are subdivided into (a) direct obstetric deaths
and (b) indirect obstetric deaths resulting from pre-existing di seas,e
or disease that developed during pregnancy and which was not
due to direct obstetric causes, but which was aggravated by
i
*
1B2
Manual of Epidemiology for District Health Management
physiological effects of pregnancy. The death of a pregnant
woman from an incidental cause (e.g. motor car accident) is not
classified as a maternal death.
Maternal mortality rate
=
No. of maternal deaths
in given area during one year
No. of live births in the population
in the same area during the same year
x
10 000
Mean, arithmetic. This is also commonly called the average. It is
calculated by adding together all the individual values in a group
of measurements and dividing by the number of values in the
group.
Measurement scale. The complete range of possible values for a
measurement. Scales can be divided into five main types:
dichotomous - two mutually exclusive groups, such as positive
and negative.
nominal - quahtative categories, such as for religions.
ordinal - ordered qualitative categories, such as social classes
I to V.
interval - scale with equal distances for each interval but no
particular starting or zero point, such as date of birth.
ratio - interval scale with a zero starting point, such as weight,
blood pressure, income.
Median. The central value in a range of measurements that divides
the set into two equal parts.
Mode. The most frequently occurring value in a set of observations.
Monitoring. The continuous measurement and observation of the
performance of a service or programme to see that it is proceeding
according to the proposed plans and objectives. If monitoring
reveals that there are problems, management decisions will have
to be taken to alter or improve the service or programme so that it
comes back on track.
Morbidity. Any departure from a state of wellbeing. Morbidity can
be expressed in terms of people who are ill and/or as episodes of
illness.
N
Neonatal mortality rate. The number of deaths in infants under 28
days of age in a given period, usually one year, per 1000 Eve births
in the same period.
I
183
Definitions and Terms
Non-respondents. Members of a study sample or population who do
not take part, respond or participate, for whatever reason, in the
study. Respondents may differ from non-respondents and a high
non-response rate may be an important source of bias.
Notifiable disease. A disease that, by statutory requirements, must
be reported to the public health authority.
Numerator. The upper portion of a fraction. In calculating a rate, all
people included in the numerator should have been derived from
the denominator. However, this is not true for the numerator in a
ratio.
o
Observational study. Study, survey or investigation that is made by
observing subjects and where no interventions, or at least no
additional ones, are implemented at the same time.
Observer error. Variation or error in measurements due to failure
of the observer to measure or identify the phenomenon accu
rately. Variation can be due to such faults as the observer missing
an observation, poor technique, incorrect reading or recording,
and misinterpretation of answers to questions. Observer error is
particularly important if it is non-random and biased.
Output. The immediate results that come from health care or pro
gramme activities expressed as units of service, such as number of
outpatient visits or persons immunized.
p
P or probability value. The letter P followed by <, the symbol for
less than, and a number (usually 0.05, 0.01 or 0.001) is a state
ment of the probability that the association or observation could
have occurred by chance. The number 0.05 means the observation
would be expected to occur by chance 1 in 20 times; similarly,
0.01 means 1 in 100. An association is commonly accepted as
statistically significant if P is <0.05.
Pandemic. An epidemic occurring over a very wide area.
Pathogenesis. The mechanism by which an etiological agent pro
duces disease.
Perinatal mortality rate. The officially accepted definition is as
follows:
Perinatal mortality rate
=
Late fetal deaths (23 weeks or more gestation)
plus first-week postnatal deaths
Fetal deaths plus total live births
in same population over same period
x
1C00
184
Manual of Epidemiology for District Health Management
However, the definition accepted in many countries that do not
have good vital statistical records leaves fetal deaths out of the de
nominator. Perinatal mortality is a useful indicator of the quality
of antenatal and obstetric care and is usually given as a rate per
1000 births per year.
Population. The total number of inhabitants of a given area or coun
try. In sampling, the population may refer to the units from which
the sample is drawn, not necessarily the total population of
people. The term population is also commonly used to refer to
particular subgroups, such as priority or high-risk groups.
Population pyramid. A graphical representation of the age and sex
composition of a population. A pyramid with a broad base, sloping
sides and narrow apex is typical of many developing countries.
This shape is due to high fertility and high mortality at younger
ages.
Postneonatal mortality rate. The number of in rant deaths between
28 days and one year of age in a given year per 1000 live births in
that year. In developing countries this rate largely reflects deaths
due to infectious diseases and malnutrition.
Predictive value. The probability that a person with a positive (or
negative! result in a screening or diagnostic test is in fact a true
positive (or true negative) . These are called the positive and
negative predictive values of the test. The predictive value de
pends on the sensitivity and specificity of the test and on the
prevalence of the condition being screened. See Validity.
Prevalence. The number of cases or events or conditions in a given
population at a particular point in time.
Prevalence rate. The total number of cases or events or conditions at
a particular point in time divided by the total population at risk at
the same point in time. Prevalence rates are most commonly used
for diseases or events that have a long average duration.
Prevalence study or survey. See Cross-sectional study.
Prevention. Measures aimed at promoting and maintaining health,
by such interventions as improving nutritional status, immunization, suitable water supplies and excreta disposal (primary preven
tion). Secondary prevention comprises measures aimed at ensur
ing the early detection of diseases and infections, whereas tertiary
prevention is concerned with reducing symptomatic illness and
disability.
Definitions and Terms
R
Random. Describes a happening or event due to chance and not
determined by other factors.
Randomization. The separation or allocation of individuals to two
or more groups at random. Randomization should form two or
more groups with variables randomly allocated between the
groups.
Randomized controlled trial. An experiment using people randomly
allocated to treatment or intervention groups and a control group.
The results are assessed by looking for any significant difference
between these groups. Such trials are the most rigorous and scien
tific way of testing the effectiveness of new interventions.
Random sample. A sample derived by random selection of sample
units. Each individual unit, such as village, household or person,
should have an equal chance of being included in the sample.
Relative risk. The ratio of the risk of death or disease in an exposed
population to the risk in the unexposed population.
Repeatability. The ability of a test to produce results that are identi
cal or closely similar each time it is conducted. Precision is an
other term that is often used. See Accuracy and Validity.
Representative sample. A sample that resembles the original popula
tion or reference population in every way. To ensure this, all
chosen samples should be compared with the original population,
particularly for important variables such as age and sex.
Reservoir of infection. The natural habitat of an infectious agent,
which may be a person, animal, arthropod, plant, soil, etc. It is
where the agent normally lives and multiplies.
r
Response rate. The number of interviews or examinations
completed divided by the total due to have been earned out,
expressed as a percentage. A high non-response rate can be an
important source of bias.
Retrospective study. See Case-control study.
Risk. The probability that an event will occur, e.g. that an individ
ual will become ill or die within a stated period of time or age.
The term is usually used with reference to unfavourable events.
I
Risk factor. The term is used in at least two different ways: (1) an
attribute, variable or exposure that is associated with an increased
probability of a specified event, such as the occurrence of a dis
ease. Such preceding factors are not necessarily causal (also called
risk markers); (2) an attribute, variable or exposure that actually
increases the occurrence of a specified event, and is therefore
believed to be causal (also described as a determinant).
185
IBB
Manual of Epidemiology for District Health Management
s
Sample. A selected subset of a population. A sample may be random
or non-random and it may be representative or non-representative. In an epsem (equal probability of selection method) sample
all the population units have an equal chance of being selected. A
simple random sample is an epsem sample.
Screening. This is the presumptive identification of unrecognized
disease or behaviour by using tests, examinations, questionnaires
and other procedures. Screening sorts people into positives and
negatives or normals. People who are positive will probably re
quire further investigation. It is important to examine the results
for the proportion of false positives and false negatives. See also
Sensitivity and Specificity.
Sensitivity. The proportion of true positives correctly identified by a
screening test. See Predictive value and Specificity.
Seroepidemiology. The use of serological investigations, particularly
antibody levels, to detect infections and transmission patterns.
Socioeconomic status. A descriptive classification of a person's
position in society, using such criteria as income, educational
level, occupation and dwelling place. Attitudes towards health
and health status are often closely linked to socioeconomic
status.
A classification similar to the one used by the Registrar-General
of the United Kingdom is as follows:
Social group
Occupation
I
professional
II
intermediate
I1IN
non-manual skilled
HIM
manual skilled
IV
partly skilled
V
unskilled
This classification may be applicable in an industrialized society
but is less useful in many developing countries.
Specificity. The proportion of true negatives correctly identified by a
screening test. See also Predictive value and Sensitivity.
Sporadic. A disease or event that occurs infrequently and irregularly.
A term usually applied to certain communicable diseases.
Spot map. A map showing the geographical distribution of people
with a particular characteristic, commonly used in the investiga
tion and control of an epidemic.
Standard deviation. A measure of the dispersion or variation of a set
of quantitative observations or measurements on either side of
the mean or average.
Definitions and Terms
Standardization. Application of statistical techniques to standardize
two or more populations for differences that may exist between
them, particularly in the age-sex structure, to enable valid com
parisons to be made.
Statistical significance. See P or probability value.
Stillbirth rate. See Fetal death rate.
Total fertility rate. An estimate of the total number of children a
thousand women would bear if they went on having children at
the present age-specific fertility rates. It provides an answer to the
question: How many children does a woman have on average
during her lifetime?
Transmission of infection. The spread of an infectious agent, either
through the environment or from person to person. The main
mechanisms of transmission are: direct contact, placental, fomirebome, vector-bome and air-borne.
I
Trend. A long-term general movement or change in frequency,
usually either upwards or downwards. A downward trend in a dis
ease or unhealthy behaviour means that it is becoming less fre
quent.
Under-reporting. Failure to identify or count all cases or events,
leading to a numerator that is smaller than the true one. This
leads to estimates of frequency that are lower than the true value.
I
V
Validity. The degree to which a measurement actually measures or
detects what it is supposed to measure. This concept is particu
larly important in screening procedures. See Accuracy and Re
peatability.
Variable. Any characteristic or attribute that can be measured.
Virulence. The degree of pathogenicity, or ability to produce dis
ease, of an infectious agent.
Vital statistics. Systematically tabulated information about births,
marriages, divorces and deaths, based on registration of these vital
events.
z
Zoonosis. An infectious or communicable disease that can be transmitted from vertebrate animals to human beings.
107
168
i
«
Manual of Epidemiology for District Health Management
I
169
appendix 1
Ethical guidelines fop epidemiological
investigations
These guidelines were developed by the Scientific Working Group on Epidemiology of the UNDP/
World Bank/WHO Special Programme on Research and Training in Tropical Diseases, World
Health Organization, Geneva.
Background
These guidelines focus specifically upon ethical aspects that have
posed particular problems in epidemiological studies of communities
in developing countries. They are intended as a supplement to the
Declaration of Helsinki4 and the WHO/CIOMS guidelines on re
search involving human subjects,6 and should not be viewed in
isolation. They represent provisional guidelines for consideration,
not rules for execution.
I
I
r
The international declaration concerning ethics as set forth in the
Declaration of Helsinki, adopted by the 18th World Medical Assem
bly in 1964 and revised by the 29th World Medical Assembly in
1975, covers well most of the important ethical considerations
concerned with clinical research in human subjects. The WHO/
CIOMS proposed international guidelines for biomedical research
involving human subjects were framed with special reference to
developing countries and focus on problems of informed consent
from certain disadvantaged subjects, such as children, the mentally
ill or prisoners, and broadly ethical problems of developing countries
and community-based research. Further, the guidelines discuss
ethical review procedures at institutional and national levels and
touch upon the need to inform communities in which research will
be undertaken and upon the need for compensation from accidental
personal injury.
The ethical concerns of controlled clinical trials, which are well
outlined in these documents, include some that are also important
to epidemiological studies. In particular the ethical issues surround
ing three matters—informed consent, use of control or comparison
(e.g. placebo) procedures, and trials of drugs or other materials that
have not been approved for use in humans in the country of origin—
all may be of great importance in epidemiological studies in develop
ing countries, but are not specifically reviewed here.
•
WHO Chronicle, 30: 360-362 (1976).
b
Proposed international guidelines for biomedical research involving human
subjects. Geneva, CIOMS, 1982.
170
Manual of Epidemiology for District Health Management
Level of review
Proposals for research activities should undergo several levels of
procedural review, as described below.
Individual
The fundamental level of review is the individual person,- every
participant in a study must provide informed consent to take part in
the research. The many complex issues that informed consent may
involve are generally well covered in the above-mentioned docu
ments. Illiteracy and differing cultural concepts of health and disease
do not alter the basic principles of informed consent. Written con
sent may be a legal requirement in some countries, but from the
ethical viewpoint it is neither necessary nor sufficient. Written
consent per se is not the requirement; the critical need is that the
person consenting has an understanding of the procedures, including
the benefits and hazards. Informed consent should not be considered
as an inhibiting factor in research: indeed, a properly informed sub
ject may be a much more useful participant.
Community
The second level of review, which may be of particular impor
tance for some types of epidemiological research, is a review by the
community. It is not simply a matter of consent; it is a matter of
understanding. For a research programme involving a community,
the community and the researchers should be working in partnership
since the purpose of the research should be mutually rewarding. A
well informed community will not only be more cooperative, but
also more useful in its collaboration.
National level
The third level is that of the responsible institutional or national
review body and is well covered in the WHO/CIOMS guidelines.
International
The final level of review for externally sponsored projects (i.e.,
those funded by an external national or international agency) should
ensure that the research protocol conforms to the requirements of
the Declaration of Helsinki and that the approaches to the three
prior levels are documented.
Detailed procedures for obtaining individual consent or commu
nity understanding will vary considerably and must be left to the
responsible institution and/or national review bodies and to the
principal investigator.
Obtaining approval at all levels and, in particular, obtaining the
informed consent of individuals in no way reduces the responsibility
of the investigator to these individuals and their communities A
checklist of suggested areas for review at each level is given at the
end of this appendix.
I
Ethical Guidelines for Epidemiological Investigations
Issues of particular concern to epidemiological studies in
developing countries
Health care expectations of the community
Epidemiological studies made in a community or other defined
population in developing countries may involve important consid
erations that are not encountered in clinical research. In many areas,
the only medical care available to the population is that of tradi
tional practitioners, and modem health care with the infrastructure
required for its delivery may be minimal or absent. The conduct of a
health-related research programme brings with it the quite reason
able expectations on the part of the people that some kind of health
care will be provided. One-shot, "bleed and fly" safari-type studies
have often produced unfulfilled expectations in the people, leading to
disappointment and reduced cooperation for future studies. Although
one-shot studies have had their usefulness - and may not raise ethi
cal issues when directed at solving a specific health problem such as
an epidemic - they are generally unacceptable today. Such studies,
• usually carried out by outsiders, in addition to raising expectations
among the people, make no long-term contribution to the health
infrastructure or the research capabilities of the country. Almost
always the same information can be better obtained through direct
cooperation with institutions in the host country.
I
*
Longitudinal community-based studies inevitably raise the expec
tations of the people for improved health care. There are no clear
answers to the vexing questions raised by these expectations. The
first step is that a reasonably full understanding must be achieved
between the community and the researchers before, during and after
the research. Frequently, during a longitudinal, community-based
study, the health infrastructure must be strengthened in order to
obtain information about the health of the people. When this is the
case, every attempt should be made to strengthen it on a continuing
basis by such means as training, improved record-keeping, develop
ment of sampling procedures for future use. In general, it is under
stood that the actual provision of health care is the responsibility of
the national health services and community: therefore, the re
searcher must work closely with both and resolve these issues as
fully as possible.
Involvement of local personnel
Related to this is the need to strengthen the research infrastruc
ture on a continuing basis whenever possible. At the least, this
means as much involvement as possible of local personnel in the
early stages of design of the protocol, its implementation, the collec
tion of data, and the analysis and feedback of the information to the
community and the government.
171
172
Manual of Epidemiology for District Health Management
Control or placebo groups
Tn establishing the effectiveness of virtually any intervention
procedure (vaccine trial, mass drug distribution, etc.) on a commu
nity basis, the need for control or placebo groups may create special
problems. The basic ethical issues are the same as those in clinical
trials, but the procedures used may be different. Important principles
are (1) that the control group receive the best currently established
form of intervention, if one exists, and (2) that if the new interven
tion procedure is demonstrated to be better (e.g. higher benefit/cost
ratio) then it should be provided to the control population as soon as
possible.
All the people concerned should be fully informed about the
proposed research. They should be provided with a clear explanation
of the investigation, the reasons for undertaking it and the possible
implications. Any coercion is unacceptable.
Sometimes there may be practical problems in determining who
truly represents the community; occasionally it is necessary to wor
through dual channels where there may be both legal and traditional
authorities. Generally full discussions with community leaders before, during and after the completion of the research - are both
necessary and very helpful. Again to be emphasized is that the ulti
mate objective of research is to improve the health of those in the
community; hence a cooperative partnership arrangement to ensure
hill understanding on the part of both the researchers and the com-
Use of past medical records
Epidemiological studies carried out from a hospital or clinical
base rarely raise ethical issues that are different from those of any
other clinical research investigation. Although there has been some
concern in the United Kingdom and the USA concerning the use of
past medical records without specific individual consent of both the
doctor and the patient as a potential 11 invasion of privacy', it is no
longer considered an issue provided that there is no identification of
individual patients and that complete confidentiality is assured. As
long as these principles are adhered to, the issue seems unlikely to
become a problem in developing countries.
Anonymity of the community
The anonymity of the community, in general, should be protected
with the same degree of concern as that given to the protection of
the individual. If, as often happens, sensitive data emerge in the
course of the studies of a community, it will be the responsibility of
the investigator to use the utmost discretion in relating these data.
However, sometimes the location and the circumstances are impor
tant to an understanding of the research that has taken place. Very
often no purpose is served in not revealing the identity of the com
munity. Indeed sometimes a community may take pride in being
11
Ethical Guidelines for Epidemiological Investigations
associated with the research. The best approach is to have continu
ing communication with the community as suggested above in the
community review procedures (level 2 review).
Behavioural research
It has sometimes been forgotten that informed consent may apply
to observation of personal behaviour just as with any other research
method affecting human subjects. The use of hidden observers and
secret observation procedures (including photography) is rarely
justifiable in behavioural research. When observers are required, as
in research into human contact with water, their existence should be
made known, but they can, of course, be stationed inconspicuously
without being hidden. Experience has shown that such observers are
eventually accepted—and even ignored.
Environmental effect
In wide-scale application of measures to control vectors or inter
mediate hosts of disease organisms, the effect on the community and
the environment of the methods used must be anticipated and care
fully monitored.
Ethics in training
Training at all levels should include consideration of the ethical
issues that will be involved in the trainees' future duties.
Checklist
The following is a checklist of suggested areas for review at
each level:
• Parent institution; funding organization
- scientific merit of the study
- consistency with ethical guidelines
- monitoring of results
i
- implementation of recommendations.
• National review committee
- scientific merit of the study
- consistency with ethical guideline
- use of controls
- confidentiality of records
- anonymity of subjects
- use of non-intervening observers
- cost to the community under study
- means of communication to the community of the nature of
the research
173
Manual of Epidemiology for District Health Management
174
- nature of informed consent
- acceptance of the research by the study community
- potential conflict and competition between proposals
- termination of controlled trials
- end points in the research process
- implementation of recommendations
- communication of results to the com mumty.
• LocaTcommunity
- nature and necessity for research
- possible harm and possible benefits to the community
- communication of issues to the people
- obtaining of consensus and the non-desirability of coercion
- monitoring of results
- implementation of recommendations
- end points.
• Individual
- informed consent is required from ail individuals.
Acknowledgements
These guidelines were originally drafted by Professor H. M. Gilles,
Dean, Liverpool School of Hygiene and Tropical Medicine, with
assistance from Professor F. Dunn, University of Califorma,
Berkeley. A number of useful suggestions from members of the
Scientific Working Group on Epidemiology and the secretariat of the
UNDP/World Bank/WHO Special Programme for Research and
Training in Tropical Diseases have been incorporated into the pres
ent version. Special thanks are due to Professor I. Riley, formerly
Professor of Community Health, University of Papua New Guinea,
for his comprehensive review and for the checklist of suggested areas
for review and to Dr P. Rosenfield, formerly Secretary, Scientific
Working Group on Social and Economic Research, for her compila
tion and analysis of the many useful suggestions submitted.
»
.
7 .\pi.
W'
■ I
175
appendixZ
Estimating sample size for a
prevalence studya
The size of sample needed for a prevalence study depends upon
the accuracy required and the prevalence of the condition itself. For
instance, leprosy may have a prevalence of around 1 per 100, or 10
per 1000 people. In a sample of 100 people, therefore, only 1 case
would be expected and there is a reasonable chance that no cases at
all would be observed. Thus, such a small sample is unlikely to give
an accurate assessment of the leprosy prevalence rate. Even in a
sample of 1000 people we would expect only 10 leprosy cases. For a
more common condition, such as schistosomiasis with a prevalence
of 30%, say, a sample of 100-200 people would give a reasonably
accurate prevalence rate and examining as many as 1000 people
*
* would probably not be necessary.
If the sample is selected correctly, the larger the sample, the
closer the estimate of prevalence in the sample is likely to be to the
true prevalence in the whole community from which the sample is
drawn. The smaller the sample, however, the smaller will be the
time and resources required. Also supervision and quality control
may be easier with a small sample, which will ensure the accuracy
and repeatability of the information collected. Thus in a prevalence
study, the sample size required is the smallest one that will give an
estimate of prevalence with the desired degree of accuracy. The table
on page 176 shows examples of minimum sample sizes for various
levels of expected prevalence and specified margins of sampling error
in the estimated prevalence.
To use this table, first select the appropriate column in the table
according to how close to 50% the prevalence is expected to be. (If
the figure is higher than 50%, use 100 minus the figure.) Then select
the appropriate row in the table according to the amount of sampling
error that can be tolerated in the estimated rate.
For example, if it is suspected that the prevalence of schistosomia
sis is somewhere between 20% and 40% in the population and that a
survey should have a good chance of estimating the prevalence to
within 5% of the true value, it is necessary to examine a random
sample of at least 369 people. When the survey is completed, if the
sample shows a prevalence of 32.5% the range for the prevalence in
the population (from which the sample was randomly drawn) is
32.5% plus or minus 5%, i.e. between 27.5% and 37.5%.
•
For a more detailed discussion of sample size, see: Lwanga, S.K. & Lemeshow, S.,
Sample size determination in health studies: a practical manual. Geneva,
World Health Organization (in press).
Manual of Epidemiology for District Health Management
17B
Minimum sample size for a prevalence survey according to expected
prevalence rate
Maximum expected prevalence rate (%)2
Margin of sampling
error tolerated1
1%
2.5%
0.5%
1 522
381
3 746
937
235
1%
2%
5%
10%
5%
10%
7 300 13 830
1 825 3 458
865
457
139
73
35
15%
20%
30%
40%
50%
6 147
1 537
246
62
28
8 068
9 220
2 305
369
93
41
9 604
2 401
385
97
43
2017
323
81
36
1
This represents the 95% confidence interval. For example, if the true prevalence
was 10% and we took a sample of size 139 we would be 95% certain that lhe^^
prevalence measured in the sample would be between 5% and 15% (i.e. 10 ± 5 /ol.
In general, do not accept a sampling error of greater than 5%.
2
Select the highest rate that the prevalence is likely to be. If the rate is expected to
be higher than 50%, use 100 minus the expected rate.
With a low-prevalence condition, such as leprosy, with a preva
lence rate of 1-2%, it is likely that only a margin of error of 0.5% to
1% would be acceptable. Referring to the above table, a 1 /o error on
an estimated prevalence of 2.5% would require a sample size of at
least 937 people. In other words, for a district leprosv survey to give
a reasonably accurate estimate of the prevalence rate of leprosy, the
sample size will need to be 1000 or more people.
To work out the required sample size for values not shown in the
table, use the following equation:
—----wnere
where
(E/1.96)2
n is the minimum sample size required
n
=
p is the maximum expected prevalence rate ( o o \I
q = 100 - p
E is the margin of sampling error tolerated [oo-.
e.g. if p = 40; q = 60; E = 5:
'V
n
(40 x 60)
(5/1.96)2
368.8 or 369 people
177
A p p e n d i x 3 Using random numbers
In order to draw a random sample for a study population, the table
of random numbers shown overleaf can be used. For example, if the
reference population is living in 130 listed villages and it is intended
to draw a random sample of 10 villages, then proceed as follows:
• Since the-reference population is living in 130 villages we require
three-figure numbers. Select any 3 adjacent single number col
umns between 1 and 40 (i.e. columns 11, 12 and 131
• Run down these 3 columns and pick out the first 10 numbers
between 1 and 130, i.e. first numbers are 48. 81, "2. etc.
• At the end of columns 11, 12 and 13 start again at the top or the
next three, i.e. 14, 15, and 16.
• Proceed like this until 10 numbers have been selected. These
numbers correspond to the numbers of the selected villages in the
village listing.
• The same procedure can be followed for randomly selecting, ror
example. 10 or 20 people from each village. However, remember
that you may want to define this sample more precisely, and only
select children aged 0-4 years or all women aged 15-44 years or
adults aged 15 years or more.
I
178
Manual of Epidemiology for District Health Management
RANDOM SAMPLING NUMBERS
1
10
18 10 49 89 75
50 89 75 71 55
11 15 50 84 49
70 25 51 01 81
62 86 38 01 20
95 19 70 36 92
85 61 50 19 61
83 55 66 76 74
90 51 34 31 18
99 56 78 99 98
27 24 80 09 77
34 63 66 89 97
28 98 45 23 35
06 96 34 21 67
19 62 94 14 54
44 36 96 82 39
76 96 59 93 98
31 61 97 08 88
42 95 12 75 72
95 42 30 03 62
48 55 12 87 21
46 18 81 87 56
66 47 43 88 02
61 91 88 50 00
85 74 04 57 53
89 09 53 94 07
54 87 27 50 35
49 13 89 98 96
97 37 11 88 77
99 70 37 54 02
65 67 36 23 39
53 69 94 34 45
54 08 33 44 54
>95 54 39 60 78
88 79 66 20 03
S 68 82 57 41 23
55 73 62 41 71
17 50 60 03 20
11 64 11 75 35
78 32 11 34 33
11
20
57 96 23 76 80
27 63 29 98 47
34 67 34 36 82
16 19 30 09 68
04 82 62 77 31
85 05 39 25 78
87 14 59 61 75
68 47 68 66 86
74 55 41 42 81
77 87 25 77 60
14 13 96 19 16
29 99 91 27 17
60 68 32 66 37
08 12 58 74 35
83 15 22 30 16
55 96 96 89 04
79 41 35 91 77
35 43 85 84 51
33 23 70 66 71
83 35 78 07 35
41 86 33 99 44
81 03 74 48 49
61 25 59 10 35
19 31 08 80 39
44 43 44 61 57
92 21 54 01 70
73 27 60 10 55
21 02 44 94 30
45 16 03 17 01
40 71 13 59 37
07 20 59 36 85
46 09 .52 84 40
42 81 46 46 42
27 35 07 35 53
48 81 94 46 07
57 52 47 09 83
45 35 51 28 64
35 64 36 90 97
76 49 67 96 84
55 30 20 68 10
21
30
93 00 28 92 31
38 94 60 09 62
53 90 49 23 88
02 21 05 62 33
49 63 64 70 99
84 34 14 28 76
53 44 19 12 00
49 47 63 51 43
70 15 36 55 16
34 13 82 02 11
22 48 88 26 25
14 56 41 05 32
43 44 27 92 07
91 64 68 15 01
92 99 79 27 67
43 89 96 59 17
66 88 50 31 77
94 85 55 05 33
76 89 28 45 92
67 85 83 57 36
83 14 01 42 54
28 37 85 93 69
09 65 92 36 93
14 03 80 46 41
29 24 36 38 79
31 91 39 51 03
13 21 24 10 55
50 70 71 02 16
00 67 28 09 39
84 38 47 11 31
47 17 51 32 75
82 80 75 72 79
01 44 13 13 97
93 29 83 01 86
91 39 12 45 51
11 27 88 40 16
82 46 10 85 71
29 78 17 83 29
11 75 73 34 90
68 96 94 82 04
31
40
44 33 49 42 80
61 42 86 50 58
06 89 27 08 16
45 95 87 67 47
39 66 55 18 11
20 20 17 79 94
65 02 00 70 99
87 42 58 36 04
10 88 62 68 72
32 31 43 48 10
42 67 93 74 00
90 14 45 30 61
91 64 22 32 72
36 52 07 00 39
13 22 25 43 19
10 84 24 12 44
06 24 08 19 51
86 42 20 51 41
12 21 41 92 53
96 97 62 67 06
59 31 64 10 04
84 92 33 52 70
47 04 89 17 03
78 82 03 69 52
49 25 39 73 02
94 83 98 31 15
84 78 88 46 83
35 31 13 14 45
28 39 11 36 82
48 92 28 96 37
07 74 63 68 01
43 97 07 96 15
35 11 85 48 41
52 11 41 68 50
68 94 53 77 83
22 64 86 22 18
21 57 92 10 58
08 99 20 47 79
97 74 85 88 37
94 10 52 73 51
17B
I
A p p e n d i x 4 Organizing an epidemiological survey
Before a survey or investigation is ready for implementation much
preparatory work needs to be done. The administrative and organiza
tional details must be dealt with before the fieldwork is started.
The procedures involved will be discussed under the following
headings:
• Identifying the procedures required.
• Implementing the investigation.
I
• Supervising the fieldwork.
Identifying the procedures required
Although preliminary plans for the investigation will already have
been made, it will be necessary to translate these into the specific
procedures required to implement the survey. A useful checklist of
the procedures involved can be obtained by answering a few ques
tions.
Who will carry out the investigation}
An investigation usually involves a large number of people, often
for considerable periods of time. The staff concerned should there
fore be well organized so that the investigation can be conducted
efficiently and in the shonest time possible.
To organize the staff, job specifications for the posts must first be
prepared. This eliminates the risk of confusion concerning the roles
of different staff and also ensures that all necessary tasks will be
carried out.
Next, activities should be grouped and coordinated so that they
can be performed most efficiently. This can only be done if proper
authority or hierarchical relationships are established. These rela
tionships provide for a group of activities to be placed under the
supervision of one person—the "manager" or "supervisor". Instruc
tions and other vital information can then be rapidly channelled
through these supervisors to the most peripheral staff. A clear line of
authority helps to harmonize working relationships within each
group. Otherwise there will be a lot of time wasted, frustrations, a
lack of coordination and duplicated efforts.
--
An effective way of clearly illustrating the line of authority is to
draw up an organizational chart. This is a diagrammatic representation of the entire team showing the main lines of authority and
180
Manual of Epidemiology for District Health Management
indicating how the different functions are linked together. A short
post description for each post will clarify this even further. It should
state the basic functions, major duties and scope of authority of a
particular post.
A clear distinction should be made between the functions and
authority delegated to line and staff personnel. A leader of line per
sonnel (e.g. pesticide spray team leader) has a supervisory role and is
directly responsible for the accomphshment of tasks assigned to that
group. Staff personnel (e.g. evaluation team), on the other hand, have
an advisory role with a basic function of assisting and advising line,
personnel. They should remain in an advisory capacity and not
assume an executive role, e.g. instructing the line staff to implement
their recommendations,- if this happens, resentment may develop.
Care must be taken to ensure that the staff personnel do not undermine the authority of the line personnel.
It is also important to delegate authority so as to permit staff to
carry out their allocated duties. This will not only permit better
coordination of the entire investigation but also help to train and
develop promising junior staff. If insufficient authority is delegated,
the subordinates will be hampered in carrying out their duties. If too
much authority is delegated, the superior's position becomes redun
dant and control of the staff more difficult. A balance is essential for
the efficient functioning of the investigation team.
A few other basic management principles should be followed to
minimize problems. These include the concept of holding the superi
ors responsible for all organizational activities of subordinates. This
follows the principle of granting authority with responsibility: those
given authority over other staff should also be held responsible for
their work performance. To ensure that there is no confusion, subor
dinates should be responsible to a single superior. Problems arise
when they become accountable to more than one superior, as then
they will have to divide their time and may find it hard to identify
themselves with a particular job. Finally, it is also important to
ensure that once authority has been delegated, it should be exer
cised. Decisions that have been delegated should be made by the
person concerned. They should not be referred to the superior, nor
should they be overruled unless there is a compelling reason to do so.
What is to be done}
A detailed job description should be carefully worked out for all
staff involved. This enables the staff member concerned to know
what is expected of him/her. The job description should also specify
the work objectives in terms of quantity (e.g. thick and thin blood
films to be collected from 50-80 persons per day), and quality (e.g.
with a constantly low discrepancy rate never rising above 10%
following re-examination of either positive or negative slides).
Organizing an Epidemiological Survey
Where will the investigation take place?
Consideration should be given to both the physical location of the
investigation and the population among whom the investigation is
to take place.
Although the general area may have already been selected and the
basis for selecting samples stated, the actual task of selecting the
exact location of the investigation has still to be completed. It
should, of course, be based upon the scientific considerations that
have been predetermined. In practice, several constraints may make
the predetermined selections impractical. A village, for example,
selected purely by random sample, may be virtually inaccessible to
the survey team, perhaps reachable only after a 30 km trek on foot.
The periodic migratory pattern of the population may also rule out a
particular area. The accessibility of an area must, therefore, form an
important consideration in the selection of a community for investi
gation.
Geographical accessibility, however, does not always mean that
the villagers can be contacted. The survey team may be able to get to
them but they must also respond to its approaches. The acceptability
of the investigation to the local community is thus equally impor
tant. Some investigations may be unacceptable because of local
beliefs. Collection and removal of faecal samples, for example, is
taboo in certain communities, as it is believed that the faeces can be
used to harm the donor. Examination of the specimen may be
permitted in situ under strict supervision, but the specimen cannot
be taken away.
I'
A thorough understanding of local customs and culture is there
fore necessary. This helps not only to overcome existing problems,
but also to prevent problems from arising in the future through sheer
ignorance and insensitivity to local customs. Such an understanding
can only be obtained through preliminary visits to the area. Some
people may feel that these visits are superfluous as they assume that
they already have a good understanding of the area and its people.
Unfortunately this is not always correct. Cultural practices may vary
markedly even in a relatively small area. Information obtained
through written or verbal reports can never convey the same depth
of understanding of the area as a personal visit. Time spent in the
area can also be used to build up a close rapport with the village
leaders. Meticulous attention to such apparently minor details may
be crucial to the quality of the investigation.
When is the investigation to be done?
I .
II
I
An investigation should be implemented according to specified
target dates and completed within the scheduled time.
181
182
Manual of Epidemiology for District Health Management
Before carrying out the investigation, check on the local condi
tions to ensure that there is no special local factor that would pro
duce problems for the study. For example, the time of day could be
unportant in many Melanesian villages which may be almost de
serted at certain periods.
The climate and season are also important features. Field surveys
during the rainy season may run into transportation and communi
cation problems. Flooded rivers, impassable mud tracks and inade
quate bridges are only some of the many possible problems. Mainte
nance and storage of equipment also becomes difficult in a situation
where there is continuous heavy rain. The rain might also affect the
vectors and animal reservoirs of disease, producing situations that
are unrepresentative of the area for the most part of the year. This
should have been considered during the planning stage, but it should
be checked again before the study is started.
In rural, particularly agricultural, regions the different seasons
also affect the activity and mobility of the population. Surveys car
ried out at planting or harvesting periods would elicit poor response
rates as most of the people have to work in the fields. Therefore
surveys may need to be carried out in the early morning and late
afternoon and evening. Market days on the other hand could be
either an advantage or disadvantage, depending upon the type of
survey being made. The movements of the population in an area
where markets are held regularly should be taken into consideration
when planning visits for village-based studies.
Numerous other examples can be quoted. All point towards the
importance of having a thorough knowledge of local customs, festi
vals, culture and other social factors.
How is the investigation to be done!
The methods by which data are to be collected (e.g. questionnaire)
and recorded must be finalized. The number, content and design of
all forms should also be finalized.
Training of ail participating personnel is essential. Regular class
room and field sessions should be held. Free discussion and com
ments should be encouraged. Indeed many potential field problems
may come to light during such sessions, especially when experienced
field staff are present. At the end of the training session, a practical
test could ensure that those who complete the training have attained
the required level of skill. Such a test could include, for example, the
identification of positive malana slides from a pre-prepared set and
the recording of a set of information. By such techniques the quality
of the field investigators can be raised considerably, so that a fair
degree of reliability of results can be achieved.
The manner in which questionnaires are administered can be
crucial. Personnel can gain practice in using the questionnaire on
Organizing an Epidemiological Survey
183
■
each other and on their own famihes. Only after a period of such
classroom training should they go into the community to carry out a
trial run. This very useful exercise can extend from one day, when
testing out a series of methods, to a longer period if further decisions
are to be made on organization of, say, data record forms or supply of
specimens to the laboratory. Locations other than the intended study
village should be selected for this purpose.
I
II
I
I I
Training on the use of equipment should also be carried out.
Calibration and standardization of all instruments will reduce meas
urement errors, e.g. in weighing machines.
The content of such training sessions should extend beyond
technical skills. Such sessions should also be used to motivate the
workers to a high level of interest and efficiency. This can be done
by familiarizing them with the goals, objectives and components of
the investigation. In particular, the importance of the study and the
benefits that it will bring to the community should be stressed. In
this way the field worker will feel that he or she is an essential team
member of an important and useful study. In addition this feeling
will be transmitted to the community through the workers in con
tact with them. If the workers are poorly motivated and show a lack
of interest in their work, the community's opinion of the investiga
tion will be adversely affected. Do not forget that as long as the team
is in contact with the community, it is being constantly watched
and assessed. If members show a lack of either interest or expertise,
this will quickly be noted and a corresponding drop in the coopera
tion and participation of the community may be expected.
Relevant and ongoing training to reinforce both the quality of
work and the motivation of the staff should form part of the pro
gramme. Any questionnaire that will be used in the survey should
also be used in training.
Operational manuals can be very useful for the field workers,
particularly for those working in remote areas with no immediate
access to supervisory staff. These could also form the basis for train
ing the staff and sufficient quantities should be prepared before the
staff are sent out.
As discussed earlier, achieving a sympathetic relationship with
the community is vital. Often enthusiasm will be kindled by the
novelty of the project. As it wears off, the problems of maintaining
or even improving upon the established links have to be faced. This
can be tackled in a variety of ways. The basic philosophy underlying
all methods should be to provide something needed by the commu
nity. In areas where no medical care is provided, the introduction of
effective treatment services is one sure way of attracting and sustain
ing community participation. The key here is to ensure that the
services provided are indeed effective and provide some benefits
(preferably including some that are obvious to the community) to the
184
Manual of Epidemiolo
people participating in the scheme. If untoward reactions are blamed
on the services provided (whether justified or not) then the opposite
reaction, i.e. rejection of the investigation, may develop in the com
munity. If additional services are provided, however, provision
should be made for their continuation after the study is completed.
Expectations must not be unduly raised. (See Appendix 1 for other
ethical considerations concerning epidemiological studies in com
munities.)
Implementing the investigation
While the investigation is being planned and the study population
identified, the local government authorities should also be informed
Whatever clearance is required from such authorities should be
obtained. At times, the study areas selected may be experiencing
internal security problems. In such circumstances, further clearance
from higher authorities may have to be acquired.
When the necessary approval has been obtained, some publicity
about the investigation may have to be arranged. This may take the
form of visits to the leaders of the community to elicit support for
the project, as well as to encourage them to spread the news of the
project by word of mouth during community meetings or during
prayer times (especially in Muslim areas). Announcements on local
radio station broadcasts regarding the forthcoming survey can also
be helpful. In rural areas, where literacy is low, newspapers will
probably not be widely read.
Finally a visit should be made to the study population area before
the start of the study. This visit should be conducted in the company
of the community leaders, so that proper introductions can be made.
This visit should have the objectives of reassuring the study popula
tion of the confidentiality of the information collected and also of
motivating them to turn up for any necessary medical examinations
or to be at home when field workers call. The visit can also serve to
identify any problems that might produce a high non-response rate
among the selected sample. Distribution of containers (e.g. for stool
examination) can, moreover, be made during this visit.
Management of crowds
Large crowds may be expected, particularly during the early
phases of the programme and in rural areas, where such an event
could be talked about for months afterwards. Efficient crowd control
is thus necessary. Spectators should be politely, but firmly, separated
from respondents. Use of flow charts would expedite the handling of
the respondents. Minimizing the waiting time and the total rime
spent by the respondents at the examination centre can go a long
way towards ensuring the success of the investigation.
i i
Organizing an Epidemiological Survey
I!
Collection of data
Basically, three types of data are collected in any investigation:
• Data obtained from the respondent or family, usually by question
naires.
• Data obtained from examination of the respondent.
• Data from specimens collected from respondents, e.g. blood.
No matter how the data are obtained, they should be entered on a
standard recording form which can subsequently be processed to
produce the information required.
I
I
I
1
I
I
A unique identification number should be given to each respon
dent in the study. The same number should be used for the same
respondent throughout. Avoid issuing different numbers for the
same person as this will produce many problems in subsequently
identifying the person and in data processing and tabulation.
Communication with the respondents is another vital aspect that
can easily be overlooked. Always ensure that there are sufficient
people present who can translate the local dialects for the team.
' Providing sufficient privacy and reassuring the respondent that all
data obtained will be kept confidential will help to ensure a better
and fuller response to questions.
Another important factor to keep in mind is that when any female
patient is being examined, particularly a young woman, a female
chaperone must be present throughout. This helps the respondent
and the community to have confidence in the investigating team.
When any investigative procedure is planned (e.g. taking blood’,
the procedure and the investigations planned should be carefully
explained. Careful attention should also be paid to the labelling and
storage of specimens. The donor should be clearly identified on the
specimen container in waterproof ink, and an effective glue should
be used to ensure that the labels do not drop off during either the
storage or processing stages. These rather obvious and simple precau
tions are often overlooked, resulting in problems that take consider
able effort to overcome.
Supervising the fieldwork
Constant supervision and evaluation of the work done are an
essential part of any investigation, as they help to ensure a good
standard of work.
Check on methods used to collect data
I
The first thing to ensure is that what is being done is what was
planned. Sometimes, despite all the repeated planning and training
sessions, very basic facts and instructions are either distorted or
185
188
Manual of Epidemiolo
omitted. The field investigators may then introduce procedures that
are neither acceptable nor uniform. What appears to be very simple
and clear to the supervisors may actually be most confusing to the
field investigator who may have little or no formal education. There
fore check all basic data collected to ensure that they are the data
required and are being obtained in the manner prescribed. This is
most important during the early stages of the investigations when
problems are most likely to occur; at this stage corrective action can
minimize the effect of any error introduced.
Check on how the work is being done
Once it has been established that the methods used are those
recommended, check on how much (quantity) data are being col
lected and how well (quality).
In order to do this, standards must first be estabhshed for all tasks
performed. These standards should reflect the nature and type of
activity performed. They should be objective, accurate and expressed
in terms that can measure the performance of the staff
Supervision is necessary at various levels in the field, laboratory
and elsewhere. It may be direct when a supervisor works as part of a
group, usually made up of a number of teams, and accepts the re
sponsibility for maintaining the standard of work. For example, the
supervisor may issue a coloured flag to each of the teams, which is
left at the entrance to the house or compound in order to locate
them easily, thus facilitating a spot check visit by the supervisor or
other senior personnel.
Supervision is also necessary in an indirect way, for example,
during the review process of examining record forms filled in during
the field activities. It is often found that a field worker is filling in
the form incorrectly. A common error is to record a child's weight.as
90 kg instead of 09 kg. Re-exammation by one microscopist of a
regular percentage of blood films examined by another is a form of
indirect supervision or quality control.
Continuous supervision and evaluation will not only give an
indication of how work is being carried out, but will also stimulate
the field workers to produce better results, as they will know that
their work is being constantly evaluated. Sometimes only limited
supervision by senior team members is possible in the field, but one
field worker may stand out in a group as having the respect of his or
her fellow workers, the population and local officials, and can be
given the post of field supervisor.
Check on how work may be improved
Finally, never assume that the procedures and techninne-s that
have been planned will produce the desired results. All stages of even
the most carefully planned investigations may need to be improved
Organizing an Epidemiological Survey
upon, particularly once the fieldwork has started. However, before
any changes are made, the DHMT should consider whether reorien
tation or retraining is necessary.The DHMT should allow for the
possibility that changes may be needed and should arrange for them
to be included in the relevant manual of working instructions.
Pilot trials can also help to solve logistic and administrative
problems. For example, the amount of transport required, the tech
niques of specimen preservation and storage to be adopted, the ade
quacy of working and living conditions, and the setting of a realistic
time schedule can all be evaluated and improved upon by using
information gathered from a pilot study.
All data should be checked as they are collected and reviewed
each day. Simple tabulations should be done every day. If possible
the data obtained should be analysed and problems identified. Ad
justments to existing plans should be introduced whenever neces
sary, to overcome any problems found. Staff meetings can help in
identifying these solutions. Sometimes it may help to repeat the trial
runs in order to produce valid information. However, such runs need
. not stretch over a long period, but can be scheduled to last one or
two days only.
l
I
187
188
■
I
I
Manual of Epidemiology for District Health Management
189
A p p e n d i x 5 Screening and diagnostic tests
Validity
Screening and diagnostic tests may be based on standardized
interviews, physical examinations or laboratory tests, or on more
sophisticated measurements such as radiography, electrocardiogra
phy, slitlamp examinations of the eye, sonography, and histopathol
ogy. In the selection of the test and the criteria to be applied, the
epidemiologist has to consider the validity and the predictive value
of the different methods.
The validity of the test refers to the extent to which the test is
capable of correctly diagnosing the presence or absence of the disease
concerned. These two aspects are referred to respectively as the
' sensitivity and the specificity of the test. For example, a test is said
to have a sensitivity of 90% if it gives a positive result in 90% of
persons who actually have the disease. On the other hand, a test is
said to have a specificity of 90% if it gives a negative result in 90%
of persons who actually do not have the disease. Examples that
illustrate sensitivity and specificity are shown below.
The test under consideration is always compared to the "true"
situation, as shown in the following table:
Test result
t
True disease
positive
negative
total
present
a
b
a + b
absent
c
d
c + d
a +c
b + d
a + b + c - d
Total
a
Sensitivity
a + b
False negatives
b
True prevalence of disease
a+b
or condition
a+b+c+d
Specificity
False positives
when a + b + c + disa representative sample of the population
I
d___
C -r d
c
190
Manual of Epidemio
Positive predictive value of test
a
a +c
Negative predictive value of test
=
d___
b+d
Sensitivity and specificity are ratios comparing test results to the
"true" disease situation. However, tests are actually used the other
way around when they are needed to predict which individuals have
the disease or condition being investigated - hence the importance of
the positive and negative "predictive values"
Predictive value
Tests are generally judged on the basis of their sensitivity and
specificity. Such evaluations are essential, but they may not provide
all the information that a user of a test may need to mahe decisions
concerning the best strategy for the particular circumstances under
consideration. The predictive value of a test, which depends upon
the prevalence of the disease or condition, as well as the test's sensi
tivity and specificity, is the most important measure for determining
its usefulness under field conditions.
The equation below shows the relationship between sensitivity,
specificity and positive predictive value:
PV( + )
PxS,
(P x S,) + ((1 - P) x (1 - S2))
Where: P is the prevalence of the disease or condition
St is the sensitivity and S2 the specificity of the test.
The corresponding equation for the likelihood that a test-negative
person actually does not have the disease is:
(1-P) x S2
PV(-)
((1-P) x S2) + (P x (1 - S,))
These equations can be used to calculate predictive values for
different combinations of test sensitivity and specificity and disease
prevalence.
Predictive value and different prevalence rates
Consider, for example, a diagnostic test that has a sensitivity of
95% and a specificity of 95%. If this test is used in a population
where the prevalence of disease is 20%, the predictive value of a
positive result is 83%, calculated by the above equation thus:
PV[+)
0.2 x 0.95
—-------------------------------(0.2 x 0.95) + (0.8 x 0.05)
=
0.83 or 83%
!
I
When this same test is used in a population where the prevalence
of disease is only 1%, the predictive value of a positive result is only
16%, as obtained by the same equation thus:
0.01 x 0.95___________
PV(+)
0.16 or 16%
(0.01 x 0.95)+ (0.99x0.05)
’ I.
I
This means that of all the positives found by the screening test
only 16%, or about 1 in 6, are true positives.
Predictive value and disease control programmes
This example shows that even for a test of fairly high sensitivity
and specificity, the predictive value of a positive result will fall
dramatically from 83% to 16% when the prevalence of disease
falls from 20% to 1%, a situation which can occur when a control
programme has been successfully implemented.
If this test is applied to a sample of 2000 persons of known disease
status on both occasions, the expected distribution of test results in
relation to actual disease can be shown in the form of 4-fold tables,
as shown below:
Before control the prevalence rate is 20%, so that 400 out of 2000
have the disease.
Test result
positive
negative
total
yes
380
20
400
no
80
1520
1600
460
1540
2000
i
Disease present
1
total
I
Test sensitivity = 95% (380/400)
Test specificity = 95% (1520/1600)
•Positive predictive value is 380/460 = 83%
• Note that the positive predictive value can be calculated directly from the figures
in these tables only when the numbers of persons with and without disease
reflect the true prevalence in the population.
1
192
Manual of Epidemiology for District Health Management
After a successful control programme the prevalence rate drops to
1%, so that 20 out of 2000 have the disease.
Test result
Disease present
positive
negative
total
yes
19
1
20
no
99
1881
1980
total
118
1882
2000
Test sensitivity = 95% (19/20)
Test specificity = 95% (1881/1980)
•Positive predictive value is 19/118 = 16%
Thus, before the control programme, nearly everyone who had a
positive test actually had the disease or condition; when the preva
lence was reduced to 1 % only about 1 in 6 of those with a positive
test actually had the disease or condition. Before the control pro
gramme there were fewer than 20% false positive results whereas
after the programme more than 80% of the positive test results were
false,- yet there was no change in the test sensitivity or specificity.
It may also be noted that when the true prevalence rate was 20%,
the test yielded 460 positive results, giving an apparent prevalence
rate (observed) of 23%. When the true prevalence rate fell to 1%, the
test yielded 188 positive results, giving an apparent prevalence rate
(observed) of 5.9%, which is nearly six times higher than the true
rate.
Recognition of changes in predictive values of test results which
are due to changes in prevalence is also important for selection of
intervention procedures, both for the individual patient (e.g. selec
tive chemotherapy for schistosomiasis) and for community pro
grammes (e.g. mass chemotherapy for lymphatic filariasis or contin
ued zonal spraying for onchocerciasis). What are the relative costs,
risks and benefits of the particular decision either to treat an individ
ual or community showing a false positive result or not to treat if
the result is a false negative? It may be useful to change the diagnos
tic approach when the prevalence drops, e.g. perhaps to a multistage
programme with a sensitive screening test followed by a more spe
cific (and probably more costly) test. Note that to increase the
predictive value of a positive test an increase in the specificity will
be more effective than a similar increase in sensitivity.
I
193
■
«i
i
Age standardization
appendix 8
.1
This is an example showing that wrong conclusions can be drawn
if crude, or unstandardized, infection prevalence rates are used.
When there are differences in age structure between two popula
tions, there are two methods of standardization available which are
easy.W use and which are explained below:
•I
• Age-specific rates.
• Direct age standardization.
*
Example: Two villages, endemic for schistosomiasis, each have a
population of 500 people who were examined for the presence of
Schistosoma mansoni eggs in faecal specimens. The statement was
made that the two populations were similar with respect to their
total population, age range and male to female sex ratio. The overall
prevalence rates for the total population in villages A and B are
shown in the following table:
Comparison of crude prevalence rate of Schistosoma mansoni infection in
Villages A and B
I
Village B
Village A
Age
Standard
No.
Na
%
No.
No.
%
population
examined
positive
positive
examined
positive
positive
(A + B)
5
40
120
55
21
10
10
40
80
70
30
20
50
30+
50
100
150
80
70
50
150
5
20
40
56
36
30
10
40
80
70
30
20
100
150
200
160
190
200
TOTAL
500
252
50.4
5G0
187
37.4
1000
(years/
0-4
5-9
10-14
15-19
20-29
50
50
80
120
50.4-37.4___________
Difference between the two rates
i
Standard error of difference
4.2, P<0.01
As the difference between the rates for the two villages (50.4%
and 37.4% respectively) was statistically different, it was concluded
that there was a true difference in the endemicity of schistosomiasis.
That is, village A was more heavily infected than village B.
194
Manual of Epidemiology for District Health Management
Using age-specific rates
While it is true that the total prevalence rates (expressed as
percentages) for the two villages were significantly different, all the
age-specific rates were exactly the same for both villages (see third
column for each village). As the investigators looked only at the total
prevalence rate and not at the age-specific rates, their interpretation
was biased. They did not control for differences in the age composi
tion of the two populations. The peak infection rate among teenagers
and the fact that there were more people in village A in the 10-14-year
age group led to a higher-rate for the total percentage.
Using direct age standardization
There are four main steps:
1. Calculate incidence or prevalence rates for each age group,
e.g. for 0-4 year olds in village A:
Prevalence rate
=
x 100
10%
50
2. Form a new standard population for each age group by adding
populations A + B see the last column in the table.
3. Age-specific rates for A and B are multiplied by the standard
population to give the expected total number of infected people in
that age group, e.g. the number for village A 10-14 year olds is
80 x 200
160
100
4. If this process is completed for each age group and for each village
it will be found that, for each village, given the standard age
structure, there would be an expected 439 cases.
Prevalence rate
Total expected cases
x 100
Total standard population
439 x 100
1000
=
43.9%
Thus, age standardization shows that there is no difference be
tween villages A and B and therefore the original conclusion was
incorrect.
Conclusion
The above example shows how use of age-specific rates and direct
age standardization can prevent incorrect conclusions from being
drawn. In this example, an apparent difference between villages A
and B was found not to exist when these methods were used. How
ever, the reverse can also happen: an apparently insignificant differ
ence can become statistically significant after standardization, but
this effect is less common than the one demonstrated above.
195
-I
L
I
Index
Access 139-141, 155
Accuracy 28-31, 79, 155
Age 29
Age-specific rates 15, 23, 24-25, 155, 193-194
Age standardization 193-194
Age-sex pyramid 21, 22, 155
Analysis of data 54-56, 99-111
Ascariasis 50
Association 123, 155
Average 110
Bar charts 119-121
Bias 83, 156
Birth rate 23-24, 156
Birth weight, low 19
Bivariate distribution 122-123
Budgeting 88, 144
Case
definition 15-16, 49-50
detection 53, 62
Case-control study 65-66, 157
Case fatality rate 41-42, 156
Census 21, 27, 39, 157
Certification of death 42-43
Child mortality rate 25
Cholera 48, 50
Cluster sample 77
Coding 96-98
Cold chain 90
Communicating information 56, 125-129
Confidence interval 175, 176
Confidentiality 172
Consent, informed 85-86
Controls 172
Correlation 110, 123
Coverage 139-140, 141-142
Cross-tabulation 107-109
I
I
Data processing see Analysis of data
Death, causes of 39
Death rate see Mortality rate
Declaration of Helsinki 169
Denominator population see Population
at risk
Diagnosis, community 5-7
Diagnostic criteria 50
Diphtheria 49
Diseases, important 34-35, 36
District 1-2
District health management team 2-3, 7-8
Dummy tables 75-76
Epidemic 59-69
confirmation 60-62
control 67-68
definition 59-60
description 62-65
point-source 63
propagated 63-64
reporting on 68
Epidemiology
definition 9
descriptive 11
Ethical issues 84-86, 169-174
Evaluation 144-146
Expected cases 16, 41, 141-142
Facilities 50-51
False negauve 159, 189
False positive 160, 189
Fenility rate 24
Fieldwork 89-90, 185-187
Figures 113-124
Filanasis 50
Food-bome disease 64
Frequency
cumulative 116-118
measurement 12-13
Geographical distribution 124
Graphs 115-118
Growth rate, population 26-27
Guinea worm infection 50
Hand-tallying 100-101
Health care, provision 19, 139-140
Health facilities 50-51
Health status 33-34, 147-153
High-risk groups 138-139
Histogram 118-119
Immunizaaon coverage 77, 139-140
Incidence 12, 13-17, 161
Incubation period 63, 64, 161
Indicators 17-20, 135-136
Infant mortality rate 24-25
Information sources 4-5, 27-28, 35-36, 50-53
Information systems 45-47
International Health Regulations 48
Interviews 79-81
Leishmaniasis 48, 50
Leprosy 48, 50
188
Manual of Epidemiology for District Health Management
Malaria 48, 50
Malnutrition 19, 139
Maps 124
Marital status 30
Maternal mortality rate 25, 161
Relapsing fever 48
Repeatability 83, 165
Reporting 49-51, 56-57
Reports 126-129
Response rate 78-79, 127, 165
Mean, arithmetic 110, 118, 162
Measles 49, 50
Migration 27
Monitoring 145, 162
Morbidity
indicators 19
patterns 37
Mortality
indicators 19
patterns 38-39
Mortality rate 24, 25
child 25
disease-specific 41-42
infant 24-25
maternal 25, 161
neonatal 24, 162
perinatal 163-164
Sample 166
cluster 77
random 76
systematic 77
Sample size 78, 175-1.76
Sampling 76-77
Scatter diagrams 122-123
Schistosomiasis 49, 50
Screening tests 189-192
Seasonality 40-41, 74, 182
Sensitivity 84, 189-190, 192
Severity of disease 34-35
Sex 30
Sex-specific rates 15
Source of infection 62-64, 65-66
Specificity 84, 189-190, 192
Specimens 90
Standard deviation 110, 118, 166
Standardization 111, 193-194
Statistics, summanring 110
Supervision 186
Surveillance 47-49, 168
in the community 52
Surveys 53, 71-86
checklist 91, 92
cross-sectional 74
'longitudinal 74
nutritional 90
objectives 75
organization 87-92, 179-187
uses 71-73
Non-respondents 78-79, 163
Normal distribution 118
Notification of disease 36
Nutritional status 19
Objectives, survey 88, 127
Observer error 79, 163
Onchocerciasis 50
Pie chans 121-122
Pilot-testing 81, 88-89, 187
Plague 48
Planning 7-8, 131-135, 143-144
Point-source epidemic 63
Poliomyelitis 47, 48, 50
Population 21-23, 32
at risk 10, 14
density 23
growth 26-27
pyramid see Age-sex pyramid
reference 76
study 76
Predictive value 84, 190-192
Prevalence 12-17, 19, 164, 175-176
Prevention 67-68
Priority chan 137-138
Profile, health 135, 146-153
Questionnaires 79-81, 93
Random numbers 177-178
Range 110
Rates 13-15, 16-17
demographic 23-25
of population growth 26-27
Record forms 93, 98
Records 35-36
Registration of death 38-39, 42-43, 51
Tables 75-76, 113-114
Tabulation 102-109
Tally sheet 100
Tapeworm infection 50
Targets 133-134
Tetanus 49
Timetable 144
Trachoma 50
Transmission of infections 62-65, 167
Transpon 90
Trypanosomiasis, 48
Tuberculosis 49, 50
Two-by-two table 107-109
*
Validity 83-84, 189
Variables 81-82
Whooping cough 49
Yellow fever 48
I
Position: 4650 (1 views)