Introduction to Epidemiology

Cohort Studies
 Design of a Cohort Study
 
The investigator selects a study population of
exposed and non-exposed individuals and
follow-up both groups to compare the
incidence of disease (or rate of death from
disease) in the two groups. (The design may
include more than 2 groups.)
 
 Design of a Prospective Study

Exposed Non-exposed
IDENTIFY:

Develop Do not develop Develop Do not develop

Disease disease Disease disease
 …….Design of a Prospective Study
Then: Follow to see whether Disease Disease Totals Incidence Rates of disease
Does Not
Develop

Exposed a b a+b a/(a+b) = Incidence in exposed

FIRST:
Select

Not c d c+d c/(c+d) = Incidence in

Exposed

non exposed
 Selection of Study Groups in
Epidemiological Studies

EXPERIMENTAL STUDY OBSERVATIONAL STUDY

POPULATION POPULATION

OTHER THAN RANDOM

RANDOM ALLOCATION ALLOCATION
(eg. Self-selection)

GROUP A GROUP B GROUP A GROUP B

 Selection of Study Populations
The essential ingredient in prospective studies is the
comparison of outcome (s) in an exposed group and
a non-exposed group (or a group with a certain
characteristic and a group without
that characteristic). Two basic ways to generate such
groups are shown on the next page.
 …….Selection of Study
Populations
Create a study population by selecting groups
for inclusion in the study on the basis of
whether or not they were exposed (e.g.
occupationally exposed cohorts and
comparison non-exposed groups) as seen in
the figure below.
…..Design of a Prospective Study
Or, Select a population on the basis of some factor
not related to exposure, such as community of
residence as seen in the figure below. Take histories
or entire population.. Use the result of the histories
and/or tests to separate the individuals into exposed
and non-exposed (or those who have and those who
do not have, certain biologic characteristics), (e.g.
Framingham Study).
 …..Design of a Prospective Study
.

Defined
IDENTIFY: Population
Then, by history
and/or tests separate NON-RANDOMIZED
into:

EXPOSED and NON-EXPOSED

Develop Do not Develop Develop Do not Develop

Disease Disease Disease Disease
…..Design of a Prospective Study
In certain cases, when approach (2) is
followed, a population is identified, but the
exposure may not take place for some time,
even many years after the population has
been defined.
 Types of Prospective Studies

1. Concurrent Prospective Study (Cohort, Longitudinal)

a. Exposure and non-exposure ascertained at present
b. Study groups followed for several years and incidence measured.

2. Non-Concurrent Prospective Study (Retrospective Cohort)

a. Exposure ascertained from objective records in the past.
b. Outcome ascertained at present.

3. A combination of Concurrent and Non-concurrent Prospective Study

a. Exposure ascertained from objective records in the past (as in
a non-current prospective study).
b. Follow-up and measurement of outcome continues into the
future
 TIME FRAMES FOR A HYPOTHETICAL CONCURRENT AND A HYPOTHETICAL
NON-CONCURRENT PROSPECTIVE STUDY CONDUCTED IN 1989

Defined NON-CONCURRENT
CONCURRENT
Population

2011 NON-RANDOMIZED 1993

Exposed Non-Exposed
2013 2003

2020 Disease No Disease Disease No Disease 2013

 Types of Potential Bias in Prospective Studies

 
1. Selection bias: exposed and non-exposed
 
2. Information bias : particularly in non-concurrent studies
 
3. Bias in assessment of outcome
 
4. Non-response bias
 
5. Analytic bias
 When is a Prospective Study Warranted?
• When there is good evidence of an association of the disease with a
certain exposure (from clinical observations and from retrospective
studies).
• When exposure is rare, but incidence of disease among exposed is high
(special exposure groups: Industry, irradiation).
• When the time between exposure and disease is short (e.g. infection
during pregnancy and congenital malformation in the offspring).
• When attrition of study population can be minimised.
• When ample funds are available
• When the investigator has a long life expectancy
 Sources of Information Data
The information about population, morbidity and mortality in
available or may be generated from various sources. Many
records are created which contain information relevant to
health of individuals and groups of people. Some records are
prepared for the general information of population, some for
legal purposes and others primarily for medical reasons, used
by the governmental agencies as a source of frequency of
diseases.

The principal sources of epidemiological data have been

outlined below :
 1. Census:

• The primary function of census is to provide demographic information, such as, total
population and its subgroups, it also provides social and economic characteristics of
people, occupation etc.
   
The census data are extremely useful as it not only provides basic framework for
planning, but also great source for developing denominator for calculating various
vital rates and indicators.
 
The first Census in India, was started in 1891.
 
Census is repeated every 10th year.

Types of Census : de-facto

de-jure
Changes and new developments in Census records.
 2. Registration of Vital Events:
• Vital events: Births, deaths, marriage.
• Use of vital events.
• Procedures of vital events registration.
• Lay reporting.
 3. Sample Registration System (SRS)

• Presently, it is a major source of information.

• Continuous enumeration of vital events/ diseases.
• Independent intermittent surveys.
 4. Records of notification of disease:
• Purpose of notification
• Notifiable disease
• Patterns of notification
- Internationally notifiable diseases
- Variations within the country
• Cross notification
• Limitations
 5. Hospital Records
Hospital records are the major source of morbidity
and mortality information, both frequency/quantum
and distribution.
• Interpretation of hospital based data.
• Disadvantages/limitations of hospital data
• Advantages
• Refining hospital based information
- Hospital Recording Rate.
- Defining catchment area
 6. Disease registers and records:
• Besides hospital records permanent registers
of various diseases may also be available for
further follow-up, analysis and outcomes.
Such morbidity registers are mostly available
for the diseases which are covered by various
National Health Programmes and in the
special studies.
 7. Other health surveys records:
• Several other records are maintained at
various health units such as Primary Health
Centres (PHC), Sub-centres, Mother & Child
Health Centres, School Health Records,
Immunisation Records, Private clinics etc.
These records may also provide useful
epidemiological information.
 8. Data from Epidemiological Surveillance
For particular diseases which are endemic,
regular Epidemiological Surveillance is
organised, e.g. Malaria, Tuberculosis, Leprosy,
and Childhood diseases such as Diphtheria,
Whooping Cough, Tetanus, Polio, Measles,
Tuberculosis.
 9. Records Linkage:
Record linkage is a process of bringing together of
information from different records concerning to a
person or a family. Such records could be :
- different vital records.
- records from different hospitals.
- records from different social agencies.
- records of occupations, hospitalisation etc.
 10. Special Population Groups:
Certain sub-groups of the population provide
opportunity for collection of data either not available
or less readily available for the general population.
These groups are :
- Armed Forces
- Social Security Schemes
- Insured Groups
- Other Occupational Groups
 11. Population/Epidemiological Survey
• Routine statistics from various sources as outlined
about, which often are not community based and
pose a great limitation for obtaining an appropriate
denominator.
- Types of survey
- Methods of Survey :
- sampling
- tools
- limitations
 12. National Sampling Surveys:
.
1) Measures of association based on ratios
- Cohort studies
Rate ratio
Risk ratio
Hazard ratio
Odds ratio (OR)
- Case control studies
OR of exposure and OR of disease
OR when the controls are a sample of the total population

- Prevalence ratio (or Prevalence OR) as an estimate of the risk

ratio
2) Measures of association based on absolute
differences: attributable risk
1. Measures of association based on ratios (relative measures of “effect”)
 
The relative measures of association often assume the names of the measures of
disease occurrence on which they are based.
 
Examples:
 
•  Risk ratio:
• Ratio of two incidence proportions.
•  Hazard ratio:
• Ratio of two hazards (cumulative incidences).
•  Rate ratio:
• Ratio of two rates (based on person-time).
•  Odds ratio (or relative odds):
• Ratio of two odds.
(Even though many of the concepts discussed in this lecture also apply to rate ratios
and hazard ratios, for simplification purposes, the discussion is based on the risk
ratio.)
IN GENERAL

• The OR is always further away from 1.0 than

the RR.
• The higher the incidence, the higher the
discrepancy.
 Relationship between RR and OR
….when probability of the event (q) is low:
q
1-q
=q
180
180

Or, in other words, (1-q) ˜1, and thus, the “build-in bias” term, 1-q- = 1.0
RR= 10000 = 6.00

And OR – RR. 1-q+

Severe Number Myocardial infarction  
   

Systolic
HTN 180
-------
Present Absent 180
RR= 10000 = 6.00 9820
  OR= ------- = 6.09
30
Yes 10000 180 9820 10000 30
-------
9970
No 10000 30 9970
 OR vs. RR: Advantages
• OR can be estimated from logistic regression
• OR can be estimated from a case-control study
because
……OR of exposure = Odds ratio of disease