[Type text]

Lecture Note of Epidemiology

DR. WAI YAN HTUN

 1

What is Epidemiology?

Epidemiology is the study of the distribution and determinants of health-related states or events in specified
populations, and the application of this study to the control of health problems.

AIMS OF EPIDEMIOLOGY

(a) To describe the distribution and magnitude of health and disease problems in human population.
(b) To identify the etiological (risk) factors in pathogenesis of disease &
(c) To provide data essential to
 Planning, Implementation & Evaluation of services
 Prevention, control & treatment of disease
 Setting up Prioritization (among health services)

USES OF EPIDEMIOLOGY

1) To study, historically, the rise and fall of disease in the population.

2) To investigate the modes of transmission of a new disease.

3) To determine the preventable causes of disease or injury.

4) To determine the natural history of disease.

5) To study the biologic spectrum of diseases.

6) To evaluate the individual risks & chances of getting diseases.

7) To define and refine syndromes.

8) To plan and evaluate community public health interventions.

9) To set the disease control priorities

10) To improve the diagnosis, treatment & prognosis of clinical disease.

11) To improve health services and research.

12) To provide expert testimony in courts of law

………………………………………………………………………………………………………………………...
 2

Natural history of disease process

Disease

– Not a static entity & it is dynamic process

– Does not appear without warning

– Disturbance in balance between interaction agent, host and environment

 Natural history of disease refers to the progress of a disease process in an individual over time, in the
absence of intervention. Without medical intervention, the process ends with recovery, disability, or death.
 Diseases process and other phenomena of interest in epidemiology are processes, not events.
 Every disease has a natural course of progression.
 Therefore, defining, observing and measuring health and disease require understanding of concept of “natural
history”
 It is the evolution of a pathophysiologic process.

NHDP is important for

1. planning preventive activities

2. Adjusting lead time & length bias for proper implementation of screening program
3. Forecasting prognosis
4. Evaluation of intervention

Stages of Natural History of Disease

The natural history of disease can be divided into two stages:

1. Pre-pathogenesis phase

2. Pathogenesis phase (Early Pathogenesis & Late Pathogenesis)

Pre- pathogenesis phase

 This is period of preliminary to the onset of disease in man.

 Agent, host & environment do not interact in this stage.
 High risk to develop disease
 So, we all are in the pre pathogenesis phase of many CD or NCD.

Pathogenesis phases - The imbalance between agent, host and environment.

Early pathogenesis phase

The disease has entered into the human host but clinical signs and symptoms are not demonstrable.
 3

Late pathogenesis phase

Sufficient anatomic and functional changes had occurred resulting in clear cut recognizable signs and symptoms.
The end result of the disease process may be complete recovery, disability or death.

Phases of the pathogenesis period

– Exposure – circumstance that leads to successful entry of disease agents. It may be single in individual cases
and common in outbreak.
– Incubation period (latent period) - The time interval between the onset of disease & first appearance of
clinical symptoms and signs.
– Usually communicable disease IP is shorter than non- communicable disease’s IP.
– During IP/LP, the individual is still healthy but disease process is in progress.
– If we can detect the disease IP/ LP, we can treat the individual very effectively. We try to detect the disease
during IP/LP by means of the tests called “Screening”.
– Period of illness – the period start eith the onset of disease (appearance of signs and symptoms) and ends
with complete disappearance of specific symptoms.
– Period of convalescence – the period starts with the end of specific signs and symptoms of the disease and
ends by the resumption of normal stage of health.

Stages of the Development of Infectious Processes

1. The Causative or Etiological Agent

2. The Reservoir of infection - (Human reservoir, Animal reservoir, Inanimate reservoir)
3. Mode of Escape from Reservoir (Portal of exist) - Respiratory tract, Intestinal tract,
4. Mode of Transmission - Direct transmission or Indirect transmission
5. Mode of Entry into New Host (Portal of entry)
6. Host Factor

Outcomes of natural history of disease

 Complete recovery
 Recover but in the stage of disease carrier. (typhoid, E coli)
 Complications & disability (+) - Temporary disability / Permanent disability
 Death
 4

………………………………………………………………………………………………………………………………………………………………………….

Dynamic of Disease Transmission

(Chain of Infectious Disease)

Six Components

1. Agent 4. Mode of transmission

2. Reservoir or source of infection 5. Mode of entry
3. Mode of exist 6. Susceptible host

Agent – microorganisms capable of producing infection.

Reservoir –the habitat in which an infectious agent lives, grows and multiplies it can be transmitted to a
susceptible host. It includes humans, animals, and the environment.

Human Many of the common infectious diseases have human reservoirs.

Reservoir Two types persons with symptomatic illness & carriers
A carrier is a person without apparent disease who is nonetheless capable of
transmitting the agent to others.
 Asymptomatic carriers, who never show symptoms during the time they are
infected
 Incubatory or convalescent carriers, who are capable of transmission before
or after they are clinically ill.
 A chronic carrier is one who continues to harbor an agent for a extended time
following the initial infection

Animal Infectious diseases that are transmissible under normal conditions from animals to
Reservoir humans are called zoonosis.
In general, these diseases are transmitted from animal to animal, with humans as
incidental hosts. Such diseases include brucellosis (cows and pigs), anthrax (sheep),
plague (rodents), trichinosis (swine), and rabies (bats, raccoons, dogs, and other
mammals).

Environmental Plants, soil, and water in the environment are also reservoirs for some infectious
Reservoir agents.

1. Portal of exit - The path by which an agent leaves the source host. The portal of exit usually corresponds
to the site at which the agent is localized.
2. Modes of transmission - After an agent exits its natural reservoir, it may be transmitted to a susceptible
host in numerous ways. These modes of transmission are classified as:

Direct Transmission
 5

– Immediate transfer of the agent from a reservoir to a susceptible host by direct contact or droplet
spread

Indirect Transmission

– An agent is carried from a reservoir to a susceptible host by suspended air particles or by animate
(vector) or inanimate (vehicle) intermediaries.
– In vector borne,
o Mechanical transmission the agent does not multiply or undergo physiologic changes in the
vector.
o Biologic transmission - When the agent undergoes changes within the vector, the vector is
serving as both an intermediate host and a mode of transmission.
– Vehicles that may indirectly transmit an agent include food, water, biologic products (blood), and
fomites (inanimate objects such as handkerchiefs, bedding, or surgical scalpels).
– Airborne - (droplet nuclei, dust).
o Dust includes infectious particles blown from the soil by the wind as well as material that has
settled on surfaces and become re suspended
 by air currents.
– Droplet nuclei are the residue of dried droplets. The nuclei are less than 5 μ (microns) in size and
may remain suspended in the air.
3. Mode of Entry
 The route by which infectious agent enter the host. It is usually correspond to mode of exist, but
some exception.
4. Susceptible Host
 Disease agent that enter the host is overcome by defensive and resistance mechanism of host.
 Susceptibility of a host depends on genetic factors, specified acquired immunity, and other
general factors.

Factors Influencing Disease Transmission

 6

Agent Host Environment

Infectivity Sex Housing

Pathogenicity Genotype Geography
Virulence Behaviour Occupational setting
Immunogenicity Nutritional status Air quality
Antigenic stability Health status Food
Survival

………………………………………………………………………………………………
 7

Level of Prevention
Prevention is the inhibition of the development of a disease before it occurs, interrupt or slow the progress of
disease.

The four level of preventions;

1. Primordial
2. Primary
3. secondary and
4. tertiary preventions.

Primordial Prevention

– Prevention of the emergence of living patterns that contribute to increased risk of disease (e.g. the
maintenance of low-fat diets in traditional societies).
– Special attention in the prevention of chronic diseases
– Efforts are directed towards discouraging children from adopting harmful lifestyles.
– The main intervention is through individual and mass education.

Primary Prevention

– It is defined as "action taken prior to the onset of disease, which removes the possibility that a disease will
occur", i.e., it includes prevention of disease by altering susceptibility or by reducing exposure for
susceptible persons.
– Primary prevention seeks to prevent the onset of specific diseases via risk reduction:
 by altering behaviors /exposures that can lead to disease (eg : cessation of smoking ) or
 by enhancing resistance to the effects of exposure to a disease agent (eg : Vaccination )

Approaches for the Primary Prevention of Chronic Diseases

– Population (Mass) Strategy - Directed at the whole population irrespective of individual risk levels. This
approach is directed towards socio-economic, behavioral and life style changes.
– High-risk Strategy - Aim to bring preventive care to individuals at special risk.

The three approaches to primary prevention are:

1. Removal of the noxious (infectious) agent

2. Preventing the contact between the agent and the host and

3. Strengthening the human host to increase his resistance to the noxious agent.
 8

*It should be directed towards the healthy person in community during the pre-pathogenesis period.*

Measures of Primary Prevention

(a) Health promotion

 Health education
 Environmental modification
 Nutritional interventions
 Lifestyles and behavioral changes
(b) Specific measures
 Specific health education concerning the disease
 Specific environmental measures;
- potable water
- sewage disposal
- food establishments as specific measures against gastro-intestinal disease
-pest control
-Sanitary housing and working place
 Food supplements against nutritional diseases
 Genetic counselling before marriage.
 Accident protective measures
 Active immunization against infectious disease
 Specific health legislation sanitary measure against introduction of specific disease into the
community or nation free from it.
 Radiation control as a carcinogen or mutagen against cancer and genetic disease.

Secondary Prevention

– It may be defined as " actions, which halts the progress of a disease at its incipient stage and prevents
complications".
– It is directed to finding the sick component of the community during the early pathogenesis stage.
– It has 2 main requirements
 Safe and accurate method of detection
 Effective method of intervention

Nature of measures:

1. Early accurate diagnosis

2. Prompt and adequate treatment

3. Institution of measures to limit spread of infectious diseases

 9

Types of measures

1. Early case finding, diagnosis and treatment.

2. Limiting spread of infectious diseases.

3. Reporting to health authorities.

4. Isolation, quarantine or surveillance of contacts.

5. Epidemiological investigation of cases.

TERTIARY PREVENTION

It is defined as “all measures available to reduce or limit impairment and disabilities, minimize suffering caused
by existing departures from good health and to promote patient’s adjustment to irremediable conditions”.

It is directed to the sick component of the community during the late pathogenesis.

Nature of measures: as in secondary prevention

Types of measures: as in secondary prevention, plus:

– Provision of special facilities for disabled or crippled conditions, e.g., physiotherapy, corrective
appliances (artificial limb), visual or hearing aids, special education, work education.
– Rehabilitation: Physical, mental, psychological and social. The aim is to bring the crippled individual
back to his family and society as productive and independent member.

………………………………………………………………………………………………………………………………………………………………………….
 10

Determinants of Health
(Factors Influencing Health)

Health is a state of complete physical, mental and social well- being and not merely an absence of disease or
infirmity. (1948)

– Health is multifactorial.

– Health lie both with the individual and externally in the society in which he or she lives.

– Disease depends on combination of genetic factors and environmental factors.

Main factors influencing on health as the following……………

– Heredity (Human biology))

– Environment

– Life style

– Socio economic condition

– Health services

– Other factors

1. Heredity
– By genetic stand point HEALTH may be defined as that “state of individual which is based upon the
absence from the genetic constitution.
– Nature of GENES.
– Numbers of diseases are known to be of genetic origin, e.g., chromosomal anomalies, errors of
metabolism, mental retardation, some type of diabetes, etc.

2. Environment
– Hippocrates who first related disease to environment
– Pettenkofer in Germany revived the concept of disease environment association.
– Environment is classified as “INTERNAL” and “EXTERNAL”.
– So environment has direct impact on the physical, mental and social well-being of those living in
it.

– Environmental factors range from housing, water supply, psychosocial stress and family structure
through social and economic, organization of health and social welfare services in the
community.

 Internal environment
o It means “each and every component part of every tissue, organ and organ system and
their harmonious functioning within the system”.
o It is domain of internal medicine.
 11

o Some epidemiologists have used the term “microenvironment” depends on way of living
and lifestyle, e.g., eating habits, other personal habits (e.g., smoking or drinking), use of
drugs etc .

 External environment
o External or macro environment consists of those things to which man is external to the
individual human host”.
o Can be divided into physical, biological and psychosocial components, any or all of
which can affect the health of man and his susceptibility to illness.

Impacts of climate change

Major direct and indirect impacts on health

– Heat strokes

– Respiratory diseases

– Water borne diseases

– Vector borne diseases

– Malnutrition

– Injuries

– Psychological stress

3.Life Style

– Life style is the way people live, reflecting a whole range of social values, attitudes and activities.
– It is composed of cultural and behavioural patterns and life-long personal, habits (e.g., smoking,
alcoholism) that have developed through processes of socialization.
– Lifestyles are learnt through social interaction with parents, peer groups, friends and siblings and through
school and mass media.
– Health requires the promotion of healthy lifestyles.
– To achieve of optimum health demands adoption of healthy lifestyles such as
 Adequate nutrition
 Enough sleep
 Sufficient physical activity

4.Socio Economic condition

– Socioeconomic conditions have long been to influence human health.

– Mainly depends on levels of socio economic conditions among the community.

– Economic status: The per capita GNP is the most widely accepted measure of general economic
performance and it is the factor to reduce morbidity, increasing life expectancy and improving the quality
of life.
 12

– Education: A second major factor influencing health status is educating (especially female education).
World map of illiteracy closely concides with maps of poverty, malnutrition, ill health, high infant, child
mortality rates.

– Occupation; Unemployment usually shows a higher incidence of ill health and death. Loss of work may
cause loss of income & psychological and social damage.

5.Political System

– Health is also related to the country’s political system.

– The percentage of GNP spent on health is a quantitative indicator of political commitment.

– To achieve on each country’s GNP as at least 5% noticed by WHO for good health.

6. Health services

– The term health and family welfare services cover a wide spectrum of personal and community services
for treatment of disease, prevention of illness and promotion of health.

– The aim of health services is to improve, health status of population.

– E.g., Immunization of children

– Provision of safe water

– Care of pregnant women and children
o education of mothers.
o to reduce child morbidity & mortality
– To be effective health services must
o reach the social periphery
o equitably distributed,
o accessible at a cost the country and community can afford and
o socially acceptable

7.Other Factors

– Systems outside the formal health system.

– Non-governmental agencies (NGO).

– Other ministries

………………………………………………………………………………………………………………………………………………………………………….
 13

Measurement in Epidemiology

1. Measuring disease frequency or occurrence

2. Measuring association between exposures and outcomes

Measuring disease frequency or occurrence

– Number of cases
– Prevalence
– Incidence (Cumulative incidence)
– Odds
– Incidence rate (Incidence density)

Measuring association

– Absolute risk
– Relative risk or Risk Ratio
– Odds ratio
– Attributable risk

Exposure Outcome

Measuring disease frequency or occurrence

WHY?
– To examine the transmission of disease in human populations.
– For Health service planning.
– To determine burden of disease.

Epidemiological Outcomes
Continuous outcomes (e.g.Blood Pressure, Glucose level, Immune Level) Measured by Mean, Median
Categorical dichotomous outcomes (Disease/no disease Positive/Negative)
– Most often used

– Measure: Frequency or Proportion (Incidence and Prevalence use)

*Continuous can form categorical *.

1. Number of cases
•Mostly used during an epidemic (e.g. H1N1, JE)
•Not a proportion
 14

Proportion of disease =

Methods of defining cases

Incident case Prevalence case

Definition New cases occurring during a All existing cases counted on
period of time single survey or examination of a
group
How to measure Cohort Prevalence ( Cross sectional
study)
*Prevalence can’t measure in acute and severe infection*

*Cohort study on exposure and non-exposure group to define new cases*

*Cross sectional study on once time per object at the point / period of time*

Ratio __ Numerator is not part of denominator.

Proportion __Numerator is part of denominator/ express%
Rate comprises numerator, denominator, time specification and multiplier.

Definition of Cases

– Criteria for diagnosis (effect on Incidence & Prevalence)

– Definition of the time of disease occurrence
– Less precise when diagnosis relies on subjective assessment alone
*Depend on clinical examination and questionnaire*

Population at risk

Individuals must be able to develop the outcome of interest

– Disease free
– Have a specific organ to develop the disease
– No immunity for the particular disease
2) PREVALENCE
Prevalence is defined as the number of disease persons present in the population at a specific time divided by the
number of persons in the population at that time.
*Use Percentage in common disease and per 1000 in rare disease*

Prevalence per 1,000 = ×1000

Prevalence is used
– To measure of the burden of disease in the community
– By health planners
 15

– Measures the need for treatment and hospital beds

– Helps in planning health facilities and man power needs.
Point Prevalence

The number of persons with a disease or an attribute at a particular point in time. Only can get old cases.

Period Prevalence

The total number of persons known to have had the disease or attribute at any time during a specified period. Can
get old and new cases.

*Prevalent / prevalence remind with fraction.*

3) INCIDENCE (Cumulative incidence)

The incidence of a disease is defined as the number of new cases of a disease that occur during a specified period
of time in a population at risk for developing the disease.

Incidence per 1000 =

INCIDENCE AND PREVALENCE

Point prevalence = Incidence x duration

Useful in stable disease / not always true/ theoretically true

Characteristics of incidence and prevalence

Characteristics Incidence Prevalence

Numerator New cases occurring during a All cases counted on a single
period of time among a group survey or examination of a group
initially free of disease
Denominator All susceptible people present at All people examined, including
the beginning of the period cases and non cases
Time Duration of the period Single point

*Prevalence includes a mix of people with different duration of disease: not a measure of risk
*Incidence includes only new cases: use as a measure of risk

Decrease prevalence Increase prevalence

– Death or cure – Increase disease duration
– Decrease new cases ( incidence ) – Good diagnostic tool
– Immigration of healthy people – Immigration of disease
– Out migration of cases – Out migration of healthy person

4) Incidence Rate

– Incidence Density or Incidence Rate

– Not a proportion
 16

– Represent speed at which individual develop the disease

Incidence rate =

*Person time means disease free time.*

5) ODDS

– The ratio of the probability of the event of interest to that of the non-event
– Mostly used in Case-control study

Odds = =

*Odds is one type of ratio*

Important thinks
– Problems with numerators
– Problems with denominators
– If rate are not changing,
– Immigration equals outmigration

Prevalence = Incidence * Duration of disease

Incidence and proportion (percentage)

Relationship between prevalence and incidence

P = IR x D

– Prevalence depends on incidence rate and duration of disease (duration lasts from onset of disease to its
termination)
– If incidence is low but duration is long - prevalence is relatively high
– If incidence is high but duration is short - prevalence is relatively low
Conditions for equation to be true
– Steady state
– IR constant
– Distribution of durations constant
– Note that if the prevalence of disease is low (less than 10%), the equation simplifies to P = IR x D

………………………………………………………………………………………………………………………………………………………………………….
 17

Measuring Disease Association

Measure of association in Cross Sectional Study

Prevalence ratio =

D+ D-
E+ a b Prevalence of disease in exposed
E- c d Prevalence of disease in non- exposed

Measure of association in Cohort / experimental study

 Incidence (CI) can be known.

 Relative Effect (Risk/Rate Ratio or Relative Risk)
RR = =

 Absolute Effect (Risk Difference or Attributable Risk)

= incidence among exposed – incidence among non-exposed
= a / (a+b) – c / (c+d)

Then Follow to See

Incidence
Whether

Disease Disease Does Not Rates of

Develops Develop Totals Disease

Exposed a b a+b
a+b
First select { Not c
c d c+d
exposed c+d

Interpretation of Risk or Rate Ratio

RR > 1 greater in exposed group than in unexposed group. DIRECT relationship between
exposure and outcome

RR = 1 identical in both groups. NO RELATIONSHIP between exposure and outcome

RR < 1 less in exposed group than in unexposed group PROTECTIVE or INVERSE relationship between
exposure and outcome

Rate Ratio
 18

Person-time Incidence rate

D+ D-

E+ a P-t E+ a/ P-t E+

E- c P-t E- c/ P-t E-

Rate Ratio Incidence rate of the disease among exposed

Incidence rate of the disease among non-exposed
= a/ P-t E+
c/ P-t E-
Relationship between smoking and incidence rate of stroke in women

Person-year Incidence rate

Exposure Cases
exposure
(per 100000 p-y)

17.7
Never smoked 70 395,594

Smoker 139 280,141 49.6

Total 209 675,735 30.9

IRR = IRE+ / IRE- = (a/PyE+) / (c/PyE-)

(139/280,141) / (70/395,594) = 2.8

Interpretation: The rate of suffering from stroke among women who smoke was 2.8
times higher compared to non-smokers.

Risk Difference or Absolute Risk

D+ D- Incidence

E+ a b a+b a/a+b

E- c d c+d c/c+d

RD = Incidence among exposed – Incidence among non-exposed = a / a + b - c / c + d

Interpretation of Risk Difference

RD > 0 Extra cases are occurring due to the exposure; value give the size of access
risk
RD = 0 No difference in incidence between exposed group and unexposed group
 19

RD < 0 Number of cases is actually below the baseline value for the population
without exposed to the factor; factor is “Protective”

A prospective study of smoking and Coronary Heart Disease (CHD)

CHD No CHD Total Incidence

(Per 1000)

Healthy smokers 84 2916 3000 28.0

Healthy non-smokers 87 4913 5000 17.4

Risk Difference (Attributable risk for exposed) = 28.0 – 17.4 = 10.6

Interpretation: 10.6 cases of coronary heart disease per 1000 people are occurring due to
smoking and presumably could be prevented if smoking were eliminated.
Attributable Fraction (Exposed)
In the exposed persons, how much of the total risk of disease is actually due to
exposure?

Incidence in the exposed - incidence in the non-exposed

incidence in exposed

= 28.0 - 17.4 = 0.379

28.0

Interpretation: 37.9% of the morbidity from CHD among smokers is attributed to

smoking and presumably could be prevented if smoking were eliminated.

Attributable Fraction (Population)

What proportion of the disease incidence in a total population can be attributed to a specific exposure?

Calculate incidence in total population

• The incidence among exposed

• The incidence among non-exposed

• The proportion of exposed people in total population

•
Incidence % non-smokers
in % smokers in Incidence in in
smokers population + non-smokers population

Assume the proportion of smokers in the population is 44%

 20

The incidence of CHD in total population =

28.0 17.4
= 22.1 per 1000
1000 0.44 + 1000 0.56 persons

Attributable Fraction (Population)

Incidence in total population- incidence in the non-exposed

Incidence in total population

22.1 - 17.4 = 0.213

22.1

Interpretation: 21.3% of the morbidity from CHD in this population is attributed to

smoking and up to 21.3% reduction in the incidence of CHD in the population could be
achieved by an effective smoking cessation program.

Risk Ratio (Relative Risk):

 How many times greater is the risk (or rate) of disease among the exposed relative to unexposed?
 Valuable in etiologic studies of disease

Risk Difference (Attributable Risk):

 How much disease can be prevented if the exposure is eliminated?

 Major applications in public health

E.g. Lung Cancer and CHD Mortality in Male British Physicians

Age-adjusted Death
Rates per 100,000 Attributabl
Relative Attributable e

Nonsmok Risk Risk fraction

Smokers ers

Lung Cancer 140 10 14.0 130 92.9

CHD 669 413 1.6 256 38.3

What disease is more likely caused by smoking?

If an effective smoking cessation program were available, would the preventive

impact be greater on mortality from lung cancer or from CHD?
 21

Measures of Association in Case – Control Study

First Select

Cases Controls
(With Disease) (Without Disease)

Then Measure Past Exposure

Were Exposed a b
Were not Exposed c d

Total a+c b+d

Proportions Exposed a b
a+c b+d

Can’t be estimate Incidence or Prevalence

But exposed / non-exposed proportion among disease / control can be estimated.

Disease Without
Disease

Exposed a b
Not Exposed c d

Incidence: Probability of getting diseased if you are exposed or not exposed

Incidence of disease among exp = a/ (a+b)

Incidence of disease among un-exp = c/ (c+d)

Odds: the ratio of the # of ways an event CAN occur relative to the # of ways an event CAN NOT occur

Odds of dis among exp = a/b Odds of exp among dis persons = a/c

Odds of dis among un-exp = c/d Odds of exp among non-dis persons = b/d

Cohort Study Case- Control Study

RR = RR can’t be estimated

OR = OR =

= ab / cd = ad /bc = ac / bd = ad / bc
 22

E.g. A Case-Control Study of Coronary Heart Disease and Cigarette Smoking

CHD Cases Controls

Smoke 112 176

Not smoke 88 224

Total 200 400

OR = ac / bd = 112 x 224 / 176 x 88

= 25088 / 15488 = 1.62 OR112 x

224 = 1.62 88 x 176

Interpretation: The odds of CHD among smokers is 1.62 times greater than that among non-smokers.

Smoking increases the odds of CHD by 1.62, as compared with not smoking

………………………………………………………………………………………………………………………………………………………………………
 23

Screening
Definition

– Screening is the search for unrecognized disease or defect by means of rapidly applied tests, examinations
or other procedures in apparently healthy individuals.
– E.g. Pap smear for cervical cancer, Fasting blood cholesterol for heart disease and Fasting blood sugar for
diabetes, etc.
– Public health field has recognized the importance of screening programs for the secondary prevention of
morbidity and mortality.
– A screening test is not intended to be a diagnostic test.
– Those who are found to have positive test results are referred to a physician for further diagnostic work-
up.

Screening test Diagnostic test

 Done on apparently healthy person  Done on those with indication or sick
 Applied to groups  Applied to single patient
 Based on one criterion or cut off point  Based on evaluation of no. of S/S and lab
 Less accurate findings
 Less expensive  More accurate
 No basis for treatment  More expensive
 The initiative come from the investigator or  Use as a basis for treatment
agency providing care  The initiative come from patient with
complaint

Screening time - Time between first possible detection and final critical point.

Lead time - Time lag between diagnosis by early detection and the usual time of diagnosis.

Types of screening Uses of screening

 Mass screening  Case detection
 High risk or selective screening  Control of diseases
 Multiphasic screening  Research purpose
 Educational opportunities

Aim Screening – predict likelihood of disease

 Screen for Risk Factors (apparently healthy person)
 Screen for Diseases Diagnostic – predict presence of disease (among
clinically relevant person)
Prognostic – predict the presence of disease
(among patient with disease)
 24

Types of Screening Uses of screening

– Mass screening – Case detection
– High risk selective screening – Control of disease
– Multiphasic screening – Research purposes
– Opportunistic screening – Educational opportunities

Mass screening

– Mass screening simply means the screening of a whole population or a sub-group, as for example ,all
adults.

– It is offered to all, irrespective of the particular risk individual.

High risk or selective screening

– Screening will be most productive if applied selectively to high risk groups.

– It is likely to result in the greatest yield of true cases and represents the most economical utilization of
screening measures.

Multiphasic screening

– The application of two or more screening tests in combination to a large number of people at one time
than to carry out separate screening tests for single disease.

Opportunistic screening

– Case finding, also referred to as opportunistic screening, is the utilization of screening tests for detection
of conditions unrelated to the patient’s chief complaint.

Criteria for screening

– Before a screening program is initiated, a decision must be made whether it is worthwhile, which requires
ethical, scientific and if possible financial justification.
– The criteria for screening are based on two considerations:

 DISEASE to be screened

 TEST to be applied.

(a) Disease

The disease to be screened should fulfils the following criteria before it is considered suitable for
screening:

 important health problem

 25

 recognizable latent or early asymptomatic stage;

 the natural history of the condition should be adequately understood
 detect the disease prior to the onset of signs and symptoms;
 facilities should be available for confirmation of the diagnosis;
 there is an effective treatment;
 agreed-on policy concerning whom to treat as patients
 good evidence that early detection and treatment reduces morbidity and mortality;
 the expected benefits (e.g., the number of lives saved) of early detection exceed the risks and
costs.

(b) Screening test

 Acceptability
 Repeatability / Reliability
 Validity (accuracy)
 Others such as yield, simplicity, rapidity, ease of administration and cost

Acceptability

– Since a high rate of cooperation is necessary, it is important that the test should be acceptable to the
people at whom it is aimed.

Repeatability / Reliability

– That is the test must give consistent results when repeated more than once on the same individual or
material, under the same conditions.

– The repeatability of the test depends upon three major factors, namely observer variation, biological (or
subject) variation and errors relating to technical methods.

– Main Measures of Reliability

• Binary or categorical measures e.g. disease/no disease,

• Overall percent agreement

• Kappa coefficient

• Continuous measures e.g. rating scale, BP (Intra class correlation coefficient (ICC)

Reliability

– Reliability is analogous to precision

– Reliability is how well an observer classifies the same individual under different circumstances.
 26

Reliability includes:

– assessments of the same observer at different times - INTRA-OBSERVER RELIABILITY

– assessments of different observers at the same time - INTER-OBSERVER RELIABILITY
– Reliability assumes that all tests or observers are equal
– One way is to look simply at percent agreement.
– Percent agreement is the proportion of all diagnoses classified the same way by two observers.

USE OF THE KAPPA STATISTIC TO ASSESS RELIABILITY

Kappa is a widely used test of inter or intra-observer agreement (or reliability) which corrects for chance
agreement.

KAPPA VARIES FROM + 1 to - 1

+ 1 means that the two observers are perfectly reliable. They classify everyone exactly the same way

0 means there is no relationship at all between the two observer’s classifications, above the agreement that would
be expected by chance.

- 1 means the two observers classify exactly the opposite of each other. If one observer says yes, the other
always says no.

GUIDE TO USE OF KAPPAS IN EPIDEMIOLOGY AND MEDICINE

Kappa > .80 is considered excellent

Kappa .60 - .80 is considered good

Kappa .40 - .60 is considered fair

Kappa < .40 is considered poor

1st WAY TO CALCULATE KAPPA

1. Calculate observed agreement (cells in which the observers agree/total cells). In both table 1 and table 2 it is
95%

2. Calculate expected agreement (chance agreement) based on the marginal totals.

MD 1
Yes No
Yes 1 3 4 (n1)
MD 2 No 2 94 96 (n2)
3 (n3) 97 (n4) 100

Overall agreement (OA) = = 95 %

 27

Expected agreement (EA) = (3 \ 100 x 4 \ 100) + (97 / 100 x 96 / 100) = 93%

Kappa =

Kappa = = 0.28%

A 2nd WAY TO CALCULATE THE KAPPA STATISTIC

( )

( ) ( )
= = 0.26 #
( ) ( )

Observer variation

– Intra-observer variation or within observer variation. This is a variation between repeated observations
by the same observer on the same subject or material at the same time & minimized by taking the average
of several replicate measurements at the same time.
– Inter-observer variation. This is a variation between different observers on the same subject or material,
also known as between observer variation.
– Observer errors can be minimized by –

 Standardization of procedures for obtaining measurements and classifications

 Intensive trainings of all the observers

 Making use of two or more observers for independent assessment, etc.

 It is probable that these errors can never be eliminated absolutely.

 28

Biological (subject) variation

– There is a biological variability associated with many physiological variables such as blood pressure
,blood sugar ,serum cholesterol, etc.

– Biological variation may be checked by repeat measurements over time.

Errors relating to technical methods

– Lastly, repeatability may be affected by variations inherent in the methods, e.g. defective instruments,
erroneous calibration, faulty reagents; or the test itself might be inappropriate or unreliable.

Quality of tests

• Measures of validity / accuracy - Sensitivity, Specificity

• Measures of efficacy - Predictive value

• Measures of reliability (reproducibility)

– Overall agreement
– Kappa
– Intra class correlation coefficient

Validity (accuracy)

– The term validity refers to what extent the test accurately measures which it purports to measure.

– In other words, validity expresses the ability of a test to separate or distinguish those who have the disease
form those who do not.

– Validity has two components : sensitivity and specificity.

– Validity is analogous to accuracy

– Validity is how well a given test reflects another test of known greater accuracy

– Validity assumes that there is a gold standard to which a test or observer should be compared.
(a) Sensitivity is the probability that a diseased person (case) in the population tested will be identified as
diseased by the test.

Sensitivity = a/(a+c) *100

(b) Specificity is the probability that a person without the disease (non-case) will be correctly identified as
non- diseased by the test.

Specificity = d/(b+d) *100

 29

Sensitivity is preferred when Specificity is preferred when

1. A dangerous but treatable disease must not be 1. “Rule in” (confirm) disease which
missed. misclassification of normal as “disease” is
2. “Rule Out” diseases that is dangerous and harmful.
contagious. 2. Further diagnostic process to confirm diagnosis
3. After positive by a screening test, is not practical or unaffordable.
confirmatory test is affordable

Understanding Predictive Values

– Clinician’s perspective: If a test result is positive, how likely is it that this individual has the disease?

– Predictive value varies with the prevalence of the disease in the screened population

– Bayes’ theorem: As the prevalence of a disease increases, the positive predictive value of the test
increases (PPV) and its negative predictive value (NPV) decreases.

– Predictive value of a positive test

 The proportion of patients who test positive who actually have the disease.
 Predictive value of a positive test = a/(a + b) x 100
– Predictive value of a negative test
 The proportion of patients who test negative who actually free of the disease.
 Predictive value of a negative test = d/(c + d) x 100

How to increase predictive value

Positive predictive value Negative predictive value

Select high specificity test Select high sensitivity test

Repeat the test
Two-stage screening: parallel or serial

Yield
– The amount of previously unrecognized disease that is diagnosed as a result of the screening effort.
– Also known as PPV.
– It depends upon –
 sensitivity and specificity of the test,
 prevalence of the disease
 the participation of the individuals in the detection programme.
Percentage of false negatives = c/(a + c) x 100
– Percentage of false positive = b/(b + d) x 100
 30

– Likelihood Ratio = (If the test is +, how much more likely the patient is to have the disease

than not have it.)

– Efficiency of the test = (a + d)/(a + b + c + d)x100

Simplicity – The test should be easy to learn and perform.

Safety – The screening test should not carry potential harm to screenees.

Rapidity – The test should not take long to administer, and the results should be available soon.

Cost – Cost–benefit ratio is an important criterion to consider in the evaluation of screening programs. The lower
the cost of a screening test, the more likely it is that the overall program will be cost beneficial.

Consequences of a False Positive Consequences of a False Negative

– Follow-up tests – False sense of security

– Cost
– Might neglect future screening tests
– Potential harm

– Anxiety
 31

Hypothetical Two-Stage Screening (same Sensitivity & Specificity)

 32

As prevalence increases, positive predictive value increases.

As specificity increases, positive predictive value increases.

33
 34

Bias

– Any systematic error in the design, conduct or analysis of a study that results in a mistaken estimate of an
exposure’s effect on the risk of the disease.

– Bias in screening is any systematic error that affects the evaluation of screening test performance

Bias in Screening

1) Lead time bias

2) Length bias

3) Over diagnosis bias

4) Selection bias

Lead-time

– Lead-time is the advantage gained by screening,i.e., the period between diagnosis by early detection and
diagnosis by other means.

– The problem of an illusion of better survival only because of earlier detection is called lead time bias.

– Screening advances the diagnosis of cancer and leads to longer survival, but no benefit in mortality
reduction.

– If screening identifies early disease before it presents clinically, it may appear to improve survival by
increasing the interval between diagnosis and death.

– In this way, individuals may appear to live longer.

 35

Length bias

– Less rapidly progressive cases are likely to be detected by screening.

– Less aggressive forms of disease usually have better survival.

– Screening detects less aggressive cancers with long preclinical phases (and better prognoses).

– A screening program is more likely to detect a larger proportion of cases of a slowly progressive, less
aggressive condition.

– It is also more likely to miss cases, which progress quickly and have a less favourable prognosis.

– Slow-growing, less aggressive conditions will invariably be present in the population for longer periods of
time, and a screening programme is more likely to be successful in detecting them.

– Length-time bias might lead to a false conclusion that screening has lengthened the lives of those who were
found positive

Over diagnosis bias

– Benign or indolent cancers are often detected.

– Cancers diagnosed have malignant potential but not likely to cause death.

– Lesions identified by screening which would not have progressed to clinical disease in the absence of
screening.
 36

Selection bias

– People who choose to participate in screening programmes often differ from those who do not.

– Randomized study design minimizes effect

– “Healthy volunteer effect”

 In some screening programs, people who are at high risk may be more likely to attend (E.g;
women with a family history of breast cancer).

 In some screening programs, individuals at lower risk are more likely to attend (E.g. women at
low risk of cervix cancer are more likely to accept an invitation for a smear test).

Ethical Issues

– Screening is sometimes done for diseases for which effective treatment is not available.

– For example, we are yet without a cure for infection with the human immunodeficiency virus (HIV).

– Screening is nonetheless important to prevent spread of the disease from infected to uninfected
individuals and to improve the prognosis of those who may be affected by initiating appropriate
treatments.

– For those diseases for which effective treatments are available, it is important to consider the capacity of
the medical community to handle the increased number of individuals requiring definitive diagnoses.

– Implementation about informed consent (Information about risks and benefits of tests and treatments).

Disadvantages of Screening

– Cost and use of medical resources on a majority of people who do not need treatment

– Adverse effects of screening procedure

– Those caused by a false positive screening result. So, stress and anxiety caused by prolonging
knowledge of an illness without any improvement in outcome

– A false sense of security caused by false negatives, which may delay final diagnosis

Barriers to Screening

– Patient barriers

 Social & cultural norms

 Psychological factors (fear and anxiety)

 Access to the health care system and insurance status

 Behavioral factors

 Perceptions of personal risk for disease

 Self-efficacy
 37

– Physician barriers

 Lack of time and competing priorities

 No reimbursement for counseling on preventive behaviors

 Mobile populations—documentation and follow up difficult

 Lack of professional consensus on benefits of some screening tests

 Organizational or systems problems

Principles of Good Screening Programs

Screen for health problems that –

– Are important to the individual and community

– Have an acceptable form of treatment

– Has a natural history that is adequately understood

– Has a recognizable latent or early symptomatic stage

– Has a suitable screening test

– Is economically beneficial

……………………………………………………………………………………………
 38

Epidemiological methods
Choice of Study Design

The choice of study design is mainly determined by:

– Existing knowledge about the problem

– Objectives of the study

– Available resources

The fundamental choice is between quantitative research approach and qualitative research approach.

1. Quantitative research (numerical information & to quantify the problems, causes and solution)
2. Qualitative research (textual in-depth information on feeling, ideas, motives and interactions among
people
3. Mixed-method research (both quantitative and qualitative)
– To quantify the magnitude and distribution of a health problem Descriptive
– To compare groups to elicit the causes/risk factors Analytical
– To assess the efficacy of drugs, treatments, interventions Experimental

Exploratory Studies

Descriptive studies
Case reports and case series
Cross-sectional study
Ecological (Correlational) studies)
Surveillence
Longitudinal study- time span study

Analytical studies
Case-control studies
Cohort studies
Ecological studies

Experimental studies
Clinical trials
Field trials
Community trial
Randomized controlled trials
Programme trials
Quasi-experimental studies
Before- after study
 39

Exploratory studies
– Small scale studies of relatively short duration
– Carried out when little is known about a situation or a problem

Descriptive Studies
Simply describes the occurrence or distribution of disease/s or event/s in a single subject or population group.

Uses of descriptive study

1. Provide data about the magnitude of the disease, type of disease problems in the community in terms of
morbidity and mortality.
2. Provide clue to disease aetiology and help in formulation of aetiological hypothesis (Causal association
between a factor and a disease)
3. Provide background data for planning, organization and evaluation of health services.
4. Contribute to further research by describing variations in disease occurrence by time, place and person.

Advantages of Descriptive Studies

– Available data (often)

– Rapid and low cost

Disadvantages of Descriptive Studies

– Lack of temporality between exposure and outcome

– Difficult to use to assess causation (outcome may have affected “exposure”)

Case reports and case series

– Describe clinical and other characteristics of a well-defined group of patients without comparison group
(e.g., patients with a certain disease)

– Detailed report of a patient (case report) or a series of patients (case series)

– An objective report of a clinical characteristic or outcome from a single subject or event.

– Can’t be used to test for a valid statistical association.

– Can’t assess disease burden.

Cross-sectional studies
– Describes the health problems by time, place and person by making all measurements on a
representative sample (or) all members of a population on a single occasion .
– May be descriptive or analytic.
– All data on exposure/s and outcome and all other variables of interest are collected simultaneously
– Likened to taking a snapshot picture at one time
– Also known as Prevalence Studies as they are used to study prevalence of certain factors.
 40

– A cross-sectional study conducted the total population is called a census.

– Measure of effect - Prevalence ratio, prevalence difference, OR
– Strengths
 can study multiple exposure and outcome
 rapid and cheap
 useful for rare disease
 useful for disease burden
– Weaknesses
 Subject to sampling and recall biases
 Can’t estimate disease incidence, only prevalence

…………………………………………………………………………………………………………………………………………………………………..

Ecological (Correlational) studies

– In ecological studies, the unit of observation for both exposure and outcome is the group, and therefore
analysis is also by group.

– This approach may be useful in the group level but would be difficult to measure at the individual level.

– Conclusions of ecological studies may not apply to individuals; thus caution is needed to avoid the
ecological fallacy.

– Ecological fallacy: is an error of logic that occurs when inferences are made reguarding individuals based
on aggregate data from the population to which the individuals belongs.

– Two main types – geographical studies & time series studies

– Useful in –

 International comparison

 Study of group level effect (e.g. Legislation)

 Hypothesis formulation

– Rapid and low cost.

– Ecological study needs not necessarily be cross sectional; they can also be longitudinal.

………………………………………………………………………………………………………………………
 41

Analytical studies

Cross sectional comparative studies

– Usually, exposure is estimated simultaneously with the disease, and different exposure subpopulations are
compared with respect to their disease prevalence. Such study is sometimes called cross-sectional
comparative study

– We simultaneously collect information on both the outcome of interest and exposure to potential risk
factor(s).

– We then compare the prevalence in the exposed and non-exposed groups

– Cross-sectional studies are especially useful for defining the health needs of a population at a particular
point in time, and for investigating common exposures and common outcomes.

– However, as they are based on prevalent (existing) cases rather than incident (new) cases, they are of
limited value for investigating etiological relationships.

– Measure of effect - Prevalence ratio, prevalence difference, OR

Diseased Non – diseased

Exposed A B A+B
Non – exposed C D C+D
A+C B+D

( )
Prevalence ratio (PR) = ( )
=

( )
Prevalence odds ratio (POR) = ( )
= =

Advantages

– Fairly quick & easy to perform

– Useful for determining the prevalence of disease in a defined population as well as risk factors
– Useful for determining the prevalence of risk factors in a defined population
– Useful for measuring current health status & planning of health services

Disadvantages

– Unable to determine temporal relationship of a presumed cause and effect.

– Selects for “longer-lasting” diseases

– Not suitable for studying severe/rapidly fatal diseases (Recall bias

– Does not give information about incidence or relative risk

………………………………………………………………………………………………………………………
 42

Cohort study
– Cohort means as a group of people who share a common characteristics or experience within a defined time
period.

– One of the main observational study designs.

– Also called follow-up, incidence, panel, prospective

– Answers the question: What is the effect of this exposure?

– Applicable in situations for which adequate numbers of exposed subjects can be found and studied and
outcome can be assessed

– Primary method of studying unusual or new exposures

– Aims to simulate Intervention studies

Indication for Cohort study

– Good evidence of an association between exposure and disease.

– When exposure is rare and incidence of disease is high among exposed.
– When attrition of population can be minimized.

Time perspective

– Prospective or Concurrent Cohort study

– Retrospective or Non - concurrent Cohort study

– Combination of retrospective & prospective.

To assemble Cohort study

– Cohort must be disease free

– Both of group in cohort are equally susceptible disease.

– Both the group should be comparable in respect of all the possible variables

– The diagnostic criteria of disease must be defined before hand.

Advantages Disadvantages
– Temporal relationship – Expensive to carry out
– Evaluation of multiple effects from a single – Long follow-up; Attrition as consequence
exposure – Normally requires a large number of subjects
– RARE exposures – Not suitable for rare diseases
– Direct calculation of disease rates – Records may inadequate for ascertainment
– Minimize the selection bias (Prospective) (Retrospective)
– Use for disease with long latency (Retrospective)
 43

Measurement

– Risk

– Risk ratio = Relative risk = RR

– Incidence rate (IR) = Incidence density (ID)

– Incidence rate ratio ( IRR) or Incidence density ratio (IDR)

– Risk difference (RD) or Attributable Risk (AR)

Retrospective Cohort Design

– Uses previously collected data on a well-defined cohort

– Common approach for disease or treatment registries since meticulous record-keeping is required

– All follow-up took place in the past

– Subject to many of the same biases of other retrospective designs

– Allows estimation of “prospective-like” measures

Elements of Cohort Study

– Selection of study subjects

– Obtaining data on exposure

– Selection of comparison groups

– Follow-up

– Analysis (Incidence among exposed & non-exposed/ Estimation of risk)

Selection of study subjects

– From general population and population that can be readily studied.

Obtaining data on exposure

– From cohort members

– Review of records
– Medical examination or special tests
– Environmental surveys

Selection of comparison groups

a. Internal comparison – single cohort enter the study and be classified into several comparison groups
according to degree of exposure to risk.
b. External comparison – putting up external control group. The study & control cohort should be similar in
demographic and other important variables.
c. Comparison with general population rates
 44

If no one is available, the mortality experience of the exposed group is compared with the mortality
experience of the general population in the same geographic area as the exposed people.

Follow up

– Periodic ME
– Reviewing records
– Surveillance
– Mailed questionnaire, phone calls, periodic home visit.

Analysis

– IR of outcomes among exposed & non exposed.

– Estimation of risk.

………………………………………………………………………………………………………………………...

Case control study

– It is also called retrospective study or comparison study.

– It is usually as a first step to explore aetiological hypothesis.

– This study compares a group of individuals with the disease (Cases) and a group of individuals without
the disease (Controls).

– By the data from a case-control study,we cannot estimate the prevalence of the disease

4 basic steps in conducting in case-control studies

1. Selection of cases & control (D/s + ----D/s -)

2. Matching (similar age, sex, occupation)

3. Measurement of exposure (interview, questionnaires,)

4. Analysis & interpretation (exposure rate/ OR)

Selection of Cases

– Establish criteria for diagnosis or the study disease

– When diagnosis relies on subjective assessment, case definition will be less precise

– Cases should be representative of all of disease people in the community (target population) in term of
risk factors or other characteristics
– Sources (Hospital, physicians’ practices, Clinical patients, Registries of patients with certain disease)
 45

– Studying mild forms of a disease results in largest possible case group but may include non-cases as cases
(misclassification)

Controls

– Should be similar to the cases in all respects other than the disease in question

– Should be representative of all persons without the disease in the population from which the cases are
selected

– Should have the potential to become cases

Selection of Controls

1. Hospital control – selected from the same hospital as the case

2. Relatives – taken up from relative (spouse & siblings). Siblings are not suitable where genetic conditions
are under study.
3. Neighbourhood – controls living in the same locality as cases
4. General population – obtained from defined geographic area

Use of Multiple Controls

1. Controls of Same Type

2. Controls of Different Types

Control of the Same Type

a. One control (e.g. hospital control) may not represent the rate of exposure that is “expected” in a
population of non-diseased persons
b. Two controls or three controls for each case
c. Used to increase the power of the study

Multiple Controls of Different Types

– Exposure of the hospital controls- not represent the rate of exposure- in a population of non-diseased
persons
– The controls may be a highly selected-non-diseased individuals and have a different exposure experience
– In using multiple controls of different types, the investigator should ideally decide which comparison will
be considered the “gold standard of truth”
– E.g. hospitalized patients smoke more than people living in the community.We do not know what the
prevalence level of smoking in hospitalized controls represents or how to interpret a comparison of these
rates with those of the cases. To address this problem, we may choose to use an additional control group,
such as neighborhood controls
 46

Matching

– The process of selecting the controls so that they are similar to the cases in certain characteristics, such as age,
race, sex, socioeconomic status, and occupation which are known to influence the outcome of disease and
which if not adequately matched for comparability could distort or confound the result.

– Confounding is a special form of bias that occurs when the association between an exposure is in fact the
result of another variable.

– Control of confounding by Design stage (restriction, matching, and randomization) & Analysis stage
(stratification or multivariate)

– Matching may be of two types – group matching & individual matching

Group matching (or frequency matching)

(1) selecting the controls with the same characteristic

(2) If cases are married, controls will be also married

Individual matching (or matched pairs)

– A control is selected who is similar to the case

– E.g. If the first case enrolled in our study is a 45-year-old white woman, we will seek a 45-year-old white
female control

Practical Problems with Matching

To match according to too many characteristics, difficult or impossible to identify an appropriate control

E.g. If the case is a 48-year-old black woman who is married, has four children, lives in zip code 21209, and
works in a photo-processing plant, it may prove difficult or impossible to find a control

Conceptual Problems with Matching

– More important problem

– Unplanned matching may inadvertently occur in case-control studies
E.g. If we match the cases (breast cancer) and the controls (no breast cancer) for marital status, we can no
longer study whether or not marital status is a risk factor for breast cancer. Why not? Because in matching
according to marital status. If 35% of the cases are married, we create a control group in which 35% are also
married
– Planned or an inadvertent manner--Overmatching

Measurement of exposure

– Information about exposure should be obtained in precisely the same manner both for cases and controls.
– Can be obtained by interviews, questionnaires, past records.
 47

– Important too rule out bias or systematic errors.

Analysis

– To find out exposure rates among cases and controls to suspected factor.

– The relative risk (RR) cannot be estimated directly, instead of RR, it is estimated indirectly using Cross
product ration or Odds Ratio (OR) which is closely related to Relative Risk.

– To measure of the strength of the association between risk factor and

outcome.

If exposure is associated with disease, a/a + c ˃ b/b + d

Advantages of Case-Control Study

– Quick and inexpensive

– Well-suited to the evaluation of outcomes with long latent periods

– Optimal for the evaluation of rare diseases

– Can examine multiple etiologic factors for a single disease

– Less time & Fewer subjects

– Can simultaneously examine a large number of potential exposures.

Disadvantages

– Cannot directly compute incidence rates of disease

– Temporal relationship between exposure and disease may be difficult to establish
– Data on past exposures status may be unavailable or prone to recall bias
– Bias

 Selection Bias (Selection of case & Selection of control)

 Information Bias (Limitations in recall & Recall bias)

Case-Control Studies based in a Defined Cohort

– A population is identified and followed over time

– At the time the population is identified, baseline data are obtained from records or interviews, from blood
or urine tests, and in other ways
 48

– The population is then followed for a period of years

– A small percentage of study participants manifest the disease, whereas most do not.

Nested Case-Control Studies

– A case–control study is enclosed(‘nested’) within a cohort study. both the cases and the controls are taken
from within the population of a cohort study.
– Nested case–control studies can be useful when it would be too expensive or otherwise unfeasible to
perform
– laboratory tests on the entire cohort.
– Advantages
 Avoid selection bias by drawing cases and controls from the same cohort
 Are cost-effective
 Can avoid recall bias by using data collected before the onset of disease
Example
– Figure - A shows the starting point as a defined cohort of individuals
– Some of them develop the disease in question but most do not over a 5-year period
– During this time, 5 cases develop—1 case after 1 year, 1 after 2 years, 2 after 4 years, and 1 after 5 years

– Figures - B–I show the time sequence in which the cases develop after the start of observations
 49

– At the time each case or cases develop, the same number of controls is selected
– The solid arrows denote the appearance of cases of the disease and the dotted arrows denote the
selection of controls who are at risk of developing the disease.
– Figure - B shows case #1 developing after 1 year and Figure - C shows control #1 being selected at
that time
– Figure - D shows case #2 developing after 2 years and Figure - E shows control #2 being selected at
that time
– Figure – F shows cases #3 and #4 developing after 4 years and Figure - G shows controls #3 and #4
being selected at that time
– Finally, Figure - H shows the final case (#5) developing after 5 years and
– Figure - I shows control #5 being selected at this point

– Figure - I is also a summary of the design and the final study populations used in the nested case-control
study
 50

– At the end of 5 years, 5 cases have appeared and at the times the cases appeared a total of 5 controls were
selected for study
– In this way, the cases and controls are matched on calendar time and length of follow-up
– Because a control is selected each time a case develops
– A control who is selected early in the study could later develop the disease and become a case in the same
study

Case cohort Design

– The second type of cohort-based case-control study is the case-cohort design

– Cases develop at the same times that were seen in the nested case-control design, but the controls are
randomly chosen from the defined cohort with which the study began

– This subset of the full cohort is called the sub-cohort

– An advantage of this design is that because controls are not individually matched to each case, it is possible to
study different diseases (different sets of cases) in the same case-cohort study using the same cohort for
controls

– In this design, in contrast to the nested case-control design, cases and controls are not matched on calendar
time and length of follow-up; instead, exposure is characterized for the sub-cohort

– This difference in study design needs to be taken into account in analyzing the study results
 51

……………………………………………………………………………………………..
 52

Experimental Studies
– Similar in approach to cohort studies excepting under the direct control of the investigator

Aims

– To provide the scientific proof of aetiological or risk factor

– To provide a method of measuring the effectiveness and efficiency of helath services for the prevention ,
control and treatment of disease

– Investigator intervenes the study group by deliberate application or withdrawal of the suspected cause.

– Two types – Randomized control trial (RCT) & Non-Randomized Experiments or Quasi-Experiments

– Measure of effect
 Relative: Rate/risk/odds ratio
 Absolute: Rate/risk/odds difference
 Other: Vaccine efficacy, difference in mean/median
– Strength
 The evidence generated can be of extremely high quality.
 If the sample is large enough, then the validity can be guaranteed.
 Blinding minimizes observation bias
– Weakness
o Ethical issues
o Potential to be extremely high cost.
o Potential biases from
 Loss to follow-up of many subjects (more likely with long follow-up periods).
 Unequal follow-up (in terms of accuracy or completeness) between the two groups.
 Placebo effect (response to any therapy regardless of physiological effect).
 The true effect of intervention is the percentage effect in the treatment group minus the
percentage effect due to placebo.

Randomized control trial (RCT)

Basic steps for RCT

1. Drawing up a protocol
2. Selecting reference and experimental population
3. Randomization
 53

4. Manipulation or intervention
5. Follow up
6. Assessment of outcome

Drawing up a protocol

Specifies the amis, objectives, questions,criteria for selection, sample size, procedures & treatment applied.

Selecting reference and experimental population

Reference or target population – population to which the finding of the trials

Experimental or study population – derived from the reference population that really participate in study.

– They must give informed consent

– They should be representative of the population
– They should be qualified or eligible for the trial.

Randomization

– An experiment in which subjects are randomly allocated into groups, usually called study and control
groups, to receive or not to receive an experimental preventive or therapeutic procedure or intervention.
– Randomization is an attempt to eliminate bias and allow for comparability and it is the heart of control
trial.

Manipulation or Intervention

It creates an independent variables (e.g. drug, vaccine) whose effect is then determined by measurement of the

final outcomes which constitutes the dependent variable (e.g. incidence of disease, survival time, recovery

period)

Follow up

It may be mentioned that some losses to follow- up are inevitable due to factors, such as death, migration and
loss of interest. This is known as attrition.

Assessment

– Positive results: that is, benefits of the experimental measure such as reduced incidence or severity of the
disease, cost to the health service or other appropriate outcome in the study and control groups.
– Negative results: that is, severity and frequency of side- effects and complications, if any, including
death. Adverse effects may be missed if they are not sought.
– Bias may arise from three sources :
 54

– First, there may be bias on the part of the participants, who may subjectively feel better or report
improvement if they knew they were receiving a new form of treatment. This is known as "subject
variation".
– Secondly there may be observer bias, that is the investigator measuring the outcome of a therapeutic trial
may be influenced if he knows beforehand the particular procedure or therapy to which the patient has
been subjected. This is known as "observer bias."
– Thirdly, there may be bias in evaluation - that is, the investigator may subconsciously give a favourable
report of the outcome of the trial.
– In order to reduce these problems, a technique known as "blinding" is adopted
– SINGLE BLIND TRIAL: the participant is not aware whether he belongs to the study or control group
– DOUBLE BLIND TRIAL: Neither the doctor nor the participant is aware of the group allocation and the
treatment received
– TRIPLE BLIND TRIAL: The participant, the investigator and the person analyzing the data are all
"blind". Ideally, of course, triple blinding should be used; but the double blinding is the most frequently
used method when a blind trial is conducted.
Some of the study designs of controlled trials

1. Concurrent parallel study designs

2. Cross over type of study designs

Types of RCTs
Clinical trials
Participants are assigned to an experimental treatment and followed for event of interest

Clinical trials may…

 55

– be randomized or non-randomized

– include a control group or have no control group

– compare current treatment to an historical control

– employ parallel or cross-over design

– employ blinding of investigator and/or participant

– The randomized, double-blind, placebo-controlled, parallel design is considered to be the best to

determine efficacy

Preventive trials
– Implies trials of primary preventive measures the trials of vaccines and chemo- prophylactic drugs
– Since preventive trials involve larger number of subjects and sometimes a longer time span to obtain
results there may be greater number of practical problems in their organization and execution.
Risk factor trials
– A type of preventive trial is the trial of risk factors in which the investigator intervenes to interrupt the
usual sequence in the development of disease for those individuals who have "risk factor" for developing
the disease The concept of "risk factor" gave a new dimension to epidemiological research.

Non - randomized trials

Sometimes the researcher cannot decide who will be exposed or who will not be exposed, in other words, he

cannot do matching of experimental and control groups. This sort of experiments are also called Non-

Randomized Experiments or Quasi-Experiments

 56

Natural Experiments

All are surveys and study the result of naturally occurring changes or differences

• Famine effect of Starvation

• Grading radiation effect of Hiroshima

Before and After study

• Before and after comparison studies without control => comparing the incidence of disease before the
after introduction of a preventive measure.

• Before and after comparison studies with control => a similar community is chosen as control.

Meta-analysis
– Pools results across multiple studies

– A review article with quantitative summary

– Typically combines results of several experimental studies

o Useful for combining small studies

o Studies should have same or similar treatments

o Pools results to get single measure of effect

– Beware: meta-analyses combining experimental and observational designs

– Dependent upon articles reporting sufficient data (N, effect measure, variance)

Rating evidence

1. Systematic reviews ( meta-analysis)

2. Randomized trials
3. Non-randomized trials
4. Cohort study
5. Case-control study
6. Cross-sectional study
7. Ecological study
 57

Sampling
Population
– is a set of all units in which we are interested.
– Typically, there are too many units in a population.
Sample
– is a subset of the population.
– Represent the whole population
– A representative sample has all the important characteristics of the population from which it is drawn
– Identifying the sample based on

 Research problem and focus

 Objectives

 Accessibility/feasibility

 Pre-test

Sampling is used for –

– Cost - saves time, labor and money.

– Quality of data - More time and effort can be spent on getting reliable data from each sampled individual
– More timely reporting
– Fairly’ precise
– More accurate result

Should not be used in –

– When size of the population is small (less than 50 members )

– When there is no interest in describing a population.
– May reduce validity and reliability of the results (during an epidemic, information is required on all cases)

Sampling Terminology

Sampling Units- may be an individual, a household, a school

Survey Population – the entire population of sampling units from which we draw our sample

Target population – the population to which the findings of the survey are to be extrapolated

Sampling frame –

– the list of all the sampling units in the survey population

– The sample is a subset of sampling frame
– It may be divided into subgroups by age, sex, social class.
– The frame may consist of areas or clusters
 58

Sampling interval – no of units in sampling frame / no of units in the sample [e.g. 500/100]

Sampling fraction or ratio – no of units in the sample: no of units in the sampling frame [100:500 or 1:5]

Classification of Sampling

Sampling
Probability simple random sampling
Methods
systematic sampling
stratified sampling
cluster sampling
two stage sampling
multistage sampling

convinence sampling
Non - Probability
snow ball sampling

Quota sampling

Self selection sampling

Purposive / judgemental

Probability Vs Non - Probability

Probability Non - Probability
– Random sampling – Non – random sampling
– Equal opportunity to be selected – Not known which individual to be selected
– Objective – Subjective
– Eliminate bias – Possible bias
– Conclusive – Exploratory

Probability Sampling

– Involves random selection procedures

– Selected on the basis of chance
– All units of the study population should have an equal chance of being included in the sample
– Probability sampling requires that a listing of all study units exists or can be compiled. This listing is called
the sampling frame.
1. Simple Random Sampling

– Simplest form of probability sampling

– Make a numbered list of all units

– Decide on the size of sample

 59

– Select require number of sampling units

– E. g. A lottery method or Random table, Scientific calculator or Computer.

– Advantages

 minimal knowledge of population needed

 Easy to analyze data

– Disadvantages
 Low frequency of use
 Do not use researcher expertise
 Larger risk of random error
2. Systematic Sampling
– Individuals are chosen at regular intervals
– Every (sampling interval) from sampling frame
– The first sample is start with random
– Calculate the interval size by k =

– Advantages
 Moderate cost & moderate usage
 Simple to draw a sample
 Easy to verify
– Disadvantage
 Periodic ordering required
3. Stratified Sampling
– Population is first divided into subgroups or strata according to one or more characteristics e.g.
sex and age groups
– Random or systematic sampling is then performed independently in each stratum
– Only possible when the proportion is known
– The proportion of each stratum in the sample should be the same as in the population
– Advantages
 Assure representation of all groups in sample population
 Characteristic of each stratum can be estimate and comparison made
– Disadvantages
 Require accurate information on proportion of each stratum
 Stratified lists costly to prepare
4. Cluster (Area) Sampling
– The population is divided into subgroup usually geographically.
– A simple random is taken from each clusters
– Effective under the following conditions:
 60

 A good sampling frame is not available or costly, while a frame listing clusters is easily obtained
 The cost of obtaining observations increases as the distance separating the elements increases
– Advantages
 Can estimate characteristics of both clusters and population
– Disadvantages
 Each stage in cluster sampling introduces sampling errors – the more stages there are, the more
errors there to be.
5. Two stage sampling
– Population is divided into primary sampling units (e.g. villages, classes of school children,
households etc.)
– A sample of primary sampling units – selected by simple random, stratified or systematic sampling
– Individuals are then chosen from each of these primary units using any method of sampling
6. Multistage sampling
– Carried out in stages
– Using smaller and smaller units at each stages
– Advantages
 More accurate and more effective
– Disadvantages
 Costly
 Each stage in cluster sampling introduces sampling errors – the more stages there are, the
more errors there to be.

Non – probability sampling

– any sampling procedure in which probability of selection is unknown or zero

– It will not be possible to tell whether or not the sample truly represents the reference population

– It is nearly impossible to avoid bias

Reasons for using Non-probability sampling

– Survey of Hard-to-Identify Groups

– Survey of specific groups
– Survey of Pilot Situation

Convenience sampling

– Use of a group of individual units that is readily available

– Sample is opportunistic and voluntary
– Often used during preliminary research efforts to get an estimate without incurring the cost or time
required to select a random sample
 61

Snowball Sampling

– Used when population listing is unavailable

– Relies on previously identified members of a group to identify other members

– Used with outlaws or unpopular people. Example. Survey of CSW, Drug users

– Advantages
 Low cost
 Useful in specific circumstance and for locating rare population.
– Disadvantages
 Not independent
 Projecting data beyond sample not justified.

Quota Sampling

– A sample is selected based on the proportions of subgroups needed to represent the proportions in the
population
– The sample is not a random sample
– First identify the stratums and their proportions as they are represented in the population
– Then convenience or judgment sampling is used to select the required number of subjects from each
stratum.

Self - selection sampling

– It occurs when you allow each case usually individuals to identify their desire to take part in the research.
– Advantages
 More accurate
 Useful in specific circumstance to serve the purpose.
– Disadvantages
 More costly due to advertizing
 Mass are left

Judgment sampling

– an extension of convenience sampling

– When using this method, the researcher must be confident that the chosen sample is truly representative
of the entire population.
– Advantages
 Assurance of quality response
 62

 Meet the specific objective

– Disadvantages
 Bias selection of sample may occur
 Time consuming process

Errors
– When we take a sample, our result will not exactly equal the correct result for the whole population due to
errors.
– Two types of errors
 Non – sampling error (systematic errors or bias)
 Sampling errors (random error or chance)

Sampling errors (random error or chance)

– It is the difference between population parameter and sampling statistics.

– The errors cannot be avoided or totally eliminated.
– It can only be reducing by increasing “n” and by using the appropriate sampling methods.
– e.g. Type 1 errors (α) & Type II errors (β)

Non – sampling error (systematic errors or bias)

– Arising from errors in design and implementation.

– Associated with the data collection and processing procedures

Sources of bias in sampling

– Improper sampling procedures

– Non – response
– Study volunteers only
– Sampling of registered patients
– Missing cases of short duration
– Seasonal bias
– Tarmac bias i.e. selection of study areas due to easy accessibility.

Reducing the bias by –

– Pre – test the data collection

– Follow up
– Replace with additional people
 63

Characteristics of good sampling design

– Truly representative
– Small sampling errors
– Bias are controlled
– Sample size is adequately large
– Economically viable

………………………………………………………………………………………………
 64

Military Epidemiology
– Militaries maintain public health programmes to monitor, prevent and treat infections that could reduce
the operational effectiveness of their resources.
– The military is involved in public health because diseases do not respect a uniform.
– To advance mission objectives or national goals, military forces may extend their public health
capabilities to civilian populations not adequately served by civilian public health programmes.
– In humanitarian emergencies, well equipped militaries may us their logistical, communication,
organizational & epidemiological surveillance for populations vulnerable to epidemics.

Types of epidemiologists

1. Clinical epidemiologist
2. Public health epidemiologist
3. Population epidemiologist
4. Military epidemiologist

Civil VS Military Population

Civil Military
– Various age – Younger age
– Both genders – Male predominance
– Background diseases – Healthy admission
– Expected environment – Extraordinary field
– General risks – Occupational hazards
– Health care access – Access to care
– Industrialization – Specific lifestyle
– Treatment oriented – Prevention oriented

Unique aspect of military

– Population – Young & healthy

– Occupation – Outdoor & traumatic injury
– Environment - Crowded condition

Adverse conditions in the field

1. Necessarily crowded - greater opportunities for spread of communicable diseases.

 65

2. Camp period is short. Sanitary methods and appliance have to be improvised, repeated moves make
careless of sanitary rules.

3. No special sanitary departments - carry out own sanitary measures.

4. Fatigue, exposure, scarcity or imperfectly cooked food and unaccustomed climates lower the soldier's
vitality and resistance.

5. Insects- more prevalent, especially when sanitary precautions are relaxed. 6. Water supplies- always
liable to pollution, purification methods must be rapid, so imperfect.

The military epidemiology’s major fields

– Infectious disease (Diarrhoea, Hepatitis A, STD, Meningitis, Influenza, Measles & Tropical diseases)
– Vaccination
 Routine vaccination
 Occupational pre exposure prophylaxis
 Operational pre exposure prophylaxis
 Case specific post exposure prophylaxis
 Research and development
– Disaster medicine
 Natural and man - made disaster
 Data collection
 Medical assistance
 Logistics assistance
– Injury prevention (Vehicle, training, heavy machinery, firearm accidents)
– Occupational exposure (Aciustic trauma, Stress, Radiation, Chemical, Lasers)
– Food hygiene
 Integral member of food hygiene team
 Coordinate food hygiene team
 Conduct data analysis
 Evidence based recommendation

Functions of military epidemiologists

1. Prevention
 Implement vaccination plans
 Devise health education strategies
 Devise health promotion strategies
 Special attention to deployed personnel
2. Surveillance
 Notifiable disease
 66

 Compute disease incidence

 Follow temporal trend of disease

3. Intervention
 Receive notification of suspected outbreak
 Collect initial data
 Determine need for onsite investigation
 Travel to site epidemic
 Investigate suspected outbreak
 Initiate epidemic control
 Identify patient contacts
 Institute chemoprophylaxis
4. Consultation
 Occupational medicine
 Accident prevention team
 Travel medicine
5. Education
 Core curriculum
 Physician education
 Military publication

Advantages in military epidemiology

It has –

– Good report rate

– Captured population
– Easy enforcement
– Clear denominator
– In house services

………………………………………………………………………………………………
 67

Surveillance

– Surveillance is the ongoing, systematic collection, analysis, and interpretation of health data essential to the
planning, implementation, and evaluation of public health practice, closely integrated with the timely
feedback of these data to those who need to know. (CDC)
– Surveillance is information for action
– Surveillance can –
 Estimate the magnitude of a problem
 Determine geographic distribution of illness
 Characterizing disease pattern (trend)
 Detect epidemics/outbreaks
 Generate hypotheses, stimulate research & Further investigation
 Evaluate whether control measures work
 Monitor changes in infectious agents & Disease control program
 Detect changes in health practices
 Serve as an early warning system, identify public health emergencies
 Guide public health policy and strategies
 Setting priorities
 Evaluation of programs and control measures

Elements of surveillance system

 68

Types of surveillance

– Passive surveillance

– Active surveillance

– Sentinel surveillance

– Integrated surveillance

– Others

o Rumour surveillance

o Syndromic surveillance

Passive Surveillance

– Source is the routine reporting of health data (Notifiable diseases, Health registry & Hospital data)
– Provider - initiated
– Useful source of health information (Baseline data, Monitor trend & Monitor impact)
– Advantages
 Low cost
 Good for monitoring large numbers of typical health events
– Limitation
 Under-reporting due to Asymptomatic illness, Access, Inadequate facilities & Logistic issues

Active surveillance

– Active case findings

– Health Department-initiated
– Sero - surveillance (testing blood markers)
– Health surveys (as needed or regularly)
– Advantage – Complete and better quality & Good for detecting small numbers of unusual health events
– Limitation – Resource intensive & More expensive,

Sentinel surveillance

– Undertaken in selected institutions or group for health data

– Monitor disease trend & Detect outbreaks

– Unable to detect outside sites/groups

– Not useful for rare condition

 69

Integrated Disease Surveillance

– Decentralizing and integrating surveillance mechanisms

– Undertake surveillance for limited number of health conditions and risk factors

Syndromic surveillance

– Allows us to identify groups of signs and symptoms that precede diagnosis and signal a sufficient
probability of a case or an outbreak that warrants a further public health response

– Illness syndromes (Fever, Respiratory, Gastrointestinal)

– Medical prescription

– Absenteeism

– Use- early identification

– Receive information by automated system

– Example: EBOLA VIRUS, JE(AES)

Rumour Surveillance

– Unofficial sources of Information (Blogs, Hearsay, Media, Social media

– Uses in alert authorities & early detection of outbreak

Planning a Surveillance System

– Establish objectives

– Develop case definitions

– Determine data source or data collection mechanism

– Field test methods

– Develop and test analytic approach

– Develop dissemination mechanism

– Assure use of analysis and interpretation

– Criteria for selecting and prioritizing health problems:

Public health importance of the problem:

– incidence, prevalence

– severity, sequele, disabilities

– mortality caused by the problem

– socioeconomic impact
 70

– communicability

– potential for an outbreak

– public perception and concern and

– international requirements.

Criteria for selecting and prioritizing health problems

– Ability to prevent, control, or treat the health problem:

o preventability and

o Availability of control measures and treatment.

– Capacity of health system to implement control measures for the health problem:

o Immediate response

o availability of resources

Establish priorities based on:

– Frequency (incidence, prevalence, mortality)

– Severity (case-fatality, hospitalization rate, disability rate, years of potential life lost)

– Cost (direct and indirect)

– Preventability

– Communicability

– Public interest

– Usefulness for public health action

– Others

Characteristics of good surveillance system

– Clearly define objectives

– Simplicity

– Flexibility

– Quality

– Acceptability

– Sensitivity and PPV

– Validity

– Representativeness
 71

– Timeliness

– Stability

– Evaluation (meeting objectives and public health functions)

A high quality public health surveillance system

– Involves and encourages the community to report all cases of diseases and other health problems

– Uses both active and passive surveillance for effective disease control and prevention

– Collects only useful data, using a simple data collection method

– Uses laboratory services to confirm clinical diagnosis of disease

– Reports data to the higher level when required and without delay

– Quickly takes the right actions to improve services or programmes after data are reported.

Effective surveillance system has following functions

– Detection and notification of health events

– Collection and consolidation of pertinent data

– Investigation and confirmation (epidemiological, clinical and/or laboratory) of cases or outbreaks

– Routine analysis and creation of reports

– Feedback of information to those providing the data

– Feed-forward (i.e. the forwarding of data to more central levels)

– Reporting data to the next administrative level

Prerequisite for effective public health surveillance

– Formulation of standard case definition of the health event and should keep in view the objectives and
logistics of the surveillance system

– Sets out appropriate method for data collection, analysis, interpretation and feedback of information

– Allocates resources efficiently and effectively

– Ensuring regularity of the reports

– Action on the reports

Ethical and Legal Issues Relating to Surveillance

 72

– Professional obligations

– Protecting confidentiality and privacy

– Informed consent

– Maintaining public trust

– Right of Access

Challenges

– Failure to report on time

– Incomplete and late reporting

– Inadequate data analysis

– Failure to use available information to check trends

– Poor feedback to health workers and communities

– Duplication of efforts

– Under utilization of surveillance information in decision making

Major sources of disease surveillance

1. Mortality report
2. Morbidity report
3. Epidemic report
4. Laboratory report
5. Report of individual case investigation
6. Reports of epidemic investigation
7. Report of special surveys (AFP, Goiter)
8. Information on animal reservoirs and vectors
9. Demographic data
10. Environmental data

Disease under national surveillance in Myanmar

4 . Principal epidemic disease – Cholera, DHF, Plague, AIDS

6 . EPI diseases – TB, Polio, DPT, Measles
5 . GI disease – Diarrhoea, Dysentry, VH, Typhoid, Food poisoning
6 . Others – Meningitis, encephalitis, ARI, Malaria, Rabies, Snake bite, Tetanus

………………………………………………………………………………………………
 73

Causal inference
A cause is something that makes a difference. Insofar as epidemiology is a science...[that] aims to discover the
cause of health states, the search includes all determinants of health outcomes. These may be both active agents...
and static conditions such as the attributes of persons and places. (Mervyn Susser)

As the purpose of epidemiology to Identify factors that cause the distribution of disease

Goal of epidemiology: learn causes of diseases and factors that could prevent or delay disease development

Causal inference: a process of determining causal and preventive factors.

– Causal inference is not a simple (or quick) process

– No single study is sufficient in establishing causal inference

– Requires critical judgment and interpretation

– Can one “prove” causal associations?

Theories of causal inference

– Deductive reasoning

– Inductivism (In Epidemiology we use inductive reasoning)

– Bayesianism

Deduction: reasoned argument proceeding from the general to the particular.

Induction: any method of logical analysis that proceeds from the particular to the general. Conceptually bright
ideas, breakthroughs and ordinary statistical inference belong to the realm of induction.

Types of causal relationships

1. Rothman’s sufficient-component cause model

2. Counterfactual model (potential-outcome)
3. Causal diagrams

Rothman’s sufficient-component cause model

– Necessary and sufficient (E.g., rabies, HIV exposure in AIDS)

– Necessary but not sufficient

 Multiple factors acting in a specific temporal sequence

 E.g., multistage carcinogenesis

– Sufficient but not necessary

 E.g., both ionizing radiation and benzene exposure cause leukemia independently
 74

– Neither sufficient nor necessary

 Many different pathways of getting the same disease

– A component cause is any one of a set of conditions which are necessary for the completion of a
sufficient cause

– A sufficient cause is a set of minimal conditions or events that inevitably produce disease

– A necessary cause is a component cause that is a member of every sufficient cause

Counterfactual model (potential-outcome)

– Ideal comparison to obtain a measure of effect would be of study subjects with themselves in both an
exposed and an unexposed state

– One of the two conditions in the definitions of the effect measures must be contrary to fact – exposures or
treatment vs. a reference condition

Effect measures vs. measures of association

Effect is:

 The endpoint of the causal mechanism

 Change in a population characteristic that is caused by the factor being at one level versus another

Effect measures:

 Can never achieve counterfactual ideal

 Logically impossible to observe the population under both conditions

Measures of association

 Compares what happens in two distinct populations

 Constructed to equal the effect measure of interest

 Absolute: differences in occurrence measures (rate or risk difference)

 Relative: ratios of occurrence measures (rate or risk ratio, relative risk, odds ratio)
 75

Comparison of absolute and relative effect measures

Causal diagrams

Provide a unified framework for evaluating design and analysis strategies for any causal question under any set of
causal assumptions.

Practice of causal inference

1. Is the observed association valid? Is it true?

– Association appears causal but is due to:

 Bias or systematic error (misclassification of E or D)

 Confounding (other variable causes the D and this variable correlates with E)

 Chance or random error (just this once)

2. Did the exposure actually cause the disease?

– Use causal guidelines to decide if association is truly causal

– The most important is temporality

Bradford Hill Criteria for Causal Inference Criteria Causal Inference

1. Strength of the association

2. Consistency

3. Specificity

4. Temporality

5. Biological gradient

6. Plausibility

7. Coherence

8. Experiment

9. Analogy
 76

Strength of the association

Strong associations are less likely to be caused by chance or bias

A strong association means a very high or very low relative risk

The strength of the association is measured by the relative risk or odds ratio

CAVEAT

Environmental associations with very low relative risks

Consistency

Replication of findings in different populations under different circumstances, in different times, with different
study designs

CAVEAT

– Lack of consistency does not rule out a causal association, because some effects are produced by their
causes only under unusual circumstances

– Publication bias

– Contradictory findings across different studies are not unusual in studies of weak effects

Specificity of the association

– Specific exposure associated with only one disease

– Effect has one cause, not multiple causes

CAVEATS

– Many exposures are linked to multiple diseases

– Many diseases have multiple causes

Temporality

– Exposure must precede disease (cause must precede effect)

– It is often easier to establish a temporal relationship in a prospective cohort study than in a case-control
study or a retrospective cohort study

– The temporal relationship of exposure and disease is important not only for clarifying the order in which
the two occur but also in regard to the length of the interval between exposure and disease

Biologic gradient (dose-response relationship)

– Presence of a dose-response or exposure-response curve with an expected shape

– Changes in exposure are related to trend in risk of disease

 77

– Strong evidence for causal relation suggesting biologic relation

– As the dose of exposure increases, the risk of disease also increases

– If a dose-response relationship is present, it is strong evidence for a causal relationship

– In some cases in which a threshold may exist, no disease may develop up to a certain level of exposure (a
threshold); above this level, disease may develop

CAVEAT

– Thresholds, i.e., no disease past a certain level of exposure

Plausibility

– The proposed mechanism should be biologically (etiologically) plausible

– Reference to a “coherent” body of knowledge

CAVEAT

– New diseases and new causes

– Theoretical plausibility

Coherence with established “facts”

A cause-and-effect interpretation for an association does not conflict with what is known of the natural history
and biology of disease

Implications:

If a relation is causal, would expect observed findings to be consistent with other data

Hypothesized causal relations need to be consistent with epidemiologic and biologic knowledge

CAVEATS

– Data may not be available yet to directly support proposed mechanism

– Science must be prepared to reinterpret existing understanding of disease process in the face of new
evidence

Experiment

– refers to ‘cessation of exposure’, i.e., elimination of putative harmful exposure results in the decrease of
the frequency of disease

– If a factor is a cause of a disease, we would expect the risk of the disease to decline when exposure to the
factor is reduced or eliminated

– A reduction in incidence being related to cessation of exposure, which adds to the strength of the causal
inference regarding the exposure

– When cessation data are available, they provide helpful supporting evidence for a causal association
 78

– E.g. Emphysema is not reversed with cessation of smoking, but its progression is reduced

CAVEATS

– If the pathogenic process has already started, removal of cause does not reduce disease risk

– Reduction in disease frequency might not be for etiologic reason hypothesized

Analogy

– Similar exposures can cause similar effects, e.g., medications and infectious agents may cause other birth
defects
– In judging whether a reported association is causal, the extent to which the investigators have taken other
possible explanations into account and the extent to which they have ruled out such explanations are
important considerations

CAVEAT

– Limited by the current knowledge

……………………………………………………………………………………….......
 79

Confounding and Interaction

Variability and Errors

Random error (sampling error)

Variation in summary results obtained from random samples taken within a population that results from
differences in the measures of a characteristics or disease model for various samples. Random error can be
reduced by taking large sample size or repeated measure

Systematic error (bias) – Error in a study that leads to a distortion of the results

Different types of Bias

Selection bias

Error due to systematic differences in characteristics between those who are selected for study and those who are
not.

Information (misclassification) bias

A flaw in measuring exposure or outcome data that results in different quality (accuracy) of information between
comparison groups.

Confounding

The distortion of the association between an exposure and disease outcome by an extraneous, third variable called
a confounder.

 A confounding variable must be a risk factor for the disease.

 A confounding variable must be associated with the exposure under study (in the population from
which the case derive).
 A confounding variable must not be an intermediate step in the causal path between the exposure
and the disease.

The distortion introduced by a confounding factor can lead to OVERESTIMATION or UNDERESTIMATION of

an effect depending on the direction of the association that the confounding factor has with the exposure and
disease variables.

Correction of confounding

Study design

 Selection / restriction
 Matching
 Randomization

Analysis

 Stratification (classification)
 Standardization (adjustment)
 Advanced statistical methods (multivariate analysis, logistic regression, etc.)
 80

Control of confounding using Stratified analysis

1. Direct pooling (Woolf's method): Gold standard, but requires large dataset

2. Maximum-likelihood method: robust predictor of direct pooling that is valid for sparse datasets using
complex equations that require a computer

3. Mantel-Haenszel method: easy to compute and provides valid estimate of maximum-likelihood measure
when relative effect values are near the null value (e.g. OR <3)

Stratified analysis

– Stratified analysis: to divide our data into several sub-groups (strata) defined by the potential
confounder.
– SA allows us
 to look at variation of the effect across strata (effect modification) comparing stratum-specific
RR estimates
 If no EM present, to compute a weighted average of the RR, the adjusted RRMH , that removes
the confounding effect
– Confounding is present when a third variable distorts the estimation of E-D relationship (either
inducing an apparent relation when there is in fact none, or enhancing/ masking a true relation)

Pros and cons of stratification

Advantages

– Mantel-Haenszel summary statistics is easy to calculate and interpret

– Gives a hands-on feeling for the data

Disadvantages

– Requires categorical confounders or continuous confounders that have been divided into intervals
– Cumbersome if more than a single confounder

(To control for more than one and/or continuous confounders, a multivariate technique such as logistic
regression is preferable)

Effect Modification

– Effect modification or interaction is present when a third variable modifies the true relationship between
E-D. The true effect measure of interest varies across levels of this third variable.
– A variation in the magnitude of a measure of exposure effect across levels of another, a third variable
– Synonyms – Effect measure modification / Statistical interaction / Heterogeneity of effect
– The effect can be greater than would be expected (positive interaction, synergism) or less than would be
expected (negative interaction, antagonism)
 81

Models of Effect modification

– Assessment of homogeneity/heterogeneity of effects (2 references)

 Detection of additive interaction (AR model)
 Detection of multiplicative interaction (RR model)
– Comparison between observed and expected joint effects of Risk factor and third variable (1 reference)
 Detection of additive interaction (AR model)
 Detection of multiplicative interaction (RR model)

Steps for control of confounding and the evaluation of effect modification through stratified analysis

1. Stratified by levels of the potential confounding factor

2. Compute stratum-specific unconfounded RR estimate
3. Evaluate similarity of the stratum-specific estimates by either eyeballing or performing test of statistical
significance.
4. If effect is thought to be uniform (i.e, if confounding is present), calculate a pooled unconfounded
summary estimate using RRMH
5. Perform hypothesis testing on the unconfounded estimate, using MH chi-squared and compute confidence
interval
6. If effect is thought not to be uniform (i.e, if effect modification is present)
a. Report stratum-specific estimates, results of hypothesis testing, and confidence intervals for each
estimate
b. If desired, calculate a summary unconfounded estimate using a standardized formula

Difference between confounding and effect modification

Confounding Effect modification

– Confounding produces a biased measurement – Effect modification represents a real difference

of an effect between an exposure factor of in the disease producing effect of an exposure
interest and a disease that results from the factor as the level of another (third) factor is
action of an extraneous factor. varied.
– Confounding produces a distorted measures of – Effect modification is not a bias. It represents
a biological effect. a real biologic modification of the effect of one
– Belong to study variable as the level or presence of a third
– Weighted RR different from crude RR factor is changed.
– Distortion of effect – Belong to nature
– Criteria confusion in data – Different effect in different strata
– Prevent in protocol – Simple
– Control in analysis – Useful
– Increased knowledge of biological mechanism
– Allows targeting of PH actions
 82

Example of confounding RR of smokers contracting bronchitis

RR=(202/1259)/(55/1389)=4.05

Among high AP group, there are 1048/1307=80%

smokers

Among low AP group, there are 211/1341=16%

smokers

Smoking is differentially distributed among high

and low AP groups

RRcrude =(178/1307)/(79/1341)=2.3

RRsmokers=(168/1048)/(34/211)=0.99  1.0

RRnonsmokers=(10/259)/(45/1130)=0.97 1.0

RRcrude=2.3 and smoking specific RR=4.0 are

above the RRsmokers  1.0, RRnonsmokers  1.0.
Therefore, crude RR is confounded by smoking
status
 83

Example of effect modification

RRcrude =(70/100)/(40/100)=1.75

RRoverweight =(30/50)/(20/50)=1.5

RRnormal=(40/50)/(20/50)=2.0

Because RRoverweight is less than RRcrude

and RRnormal is more than RRcrude, weight
is an effect modifier and not a confounder.
The study results must be presented
stratified by weight, and can not be pooled
or adjusted for weight.

ORcrude =(190x157)/(266x176)=0.64

ORmale=(141x112)/(208x144)=0.53

ORfemale=(49x45)/(58x32)=1.19

Because ORmale is different from ORfemale and

they are not similar to ORcrude, gender is an
effect modifier. The study results must be
presented stratified by gender, and can not be
pooled or adjusted for gender.
 84

Example of Stratified analysis

ORcrude=2.2

OR<40 =2.8, OR>40 =2.8

Because OR<40 is similar to OR>40 and

they are different from ORcrude, age is a
confounder. The study results must be
presented pooled or adjusted for age.

Age adjusted ORMH

=(21*59/123)+(18*95/208)=18.29=2.8

(17*26/123)+(7*88/208) 6.56