Cohort and Case Control studies

Dr.Vanitha ,
Associate Professor,
Dept. of Community Medicine,
BMCH
 Objective
 Introduction
 Design of a cohort and case control study
 Steps in a cohort and case control study
 Types of cohort studies
 Applications
 Advantages
 Disadvantages / Limitations
 Epidemiological Studies

Observational Experimental
(Non-Interventional)
(Interventional)

Descriptive Analytical

Case - Control Cohort

 Cohort
Latin - ‘cohors, warriors’
Group of persons who share a common
characteristic or experience within a defined time
period and who are followed over time.
e.g.: Birth cohort Exposure cohort
Marriage cohort Survival cohort
 Prospective or Incidence study or Forward
looking or Follow up study

Done for evaluating the association between a

suspected cause (exposure) and subsequent risk of
developing the disease (outcome) under study.
 Design of a cohort study.
Disease
P Exposed
People
O ( Study
P with
U cohort ) No
out
L disease
A disease
T Disease
I Not exposed
O (Control
N
cohort ) No
disease
Direction of inquiry
Time
 Steps in a cohort study
1.Selection of study subjects
a) from the general population - relationship between
smoking and lung cancer - select participants from a general
population registry.
b) from special groups of population –
“select” groups - selecting individuals within a
specific age range or geographic area.
Exposure groups – Exposed to industrial smoke or not
exposed group
2.Obtaining data on Exposure
a) Review of Medical Records
b) Personal Interviews / Mailed Questionnaire
c) Medical Examination / Special Test
d) Measurement of exposure levels in the Environment
 Steps in a cohort study
3. Selection of comparison groups
a) Internal comparison (built in comparison) - Compares
subgroups within study (smokers vs. non-smokers in cancer
registry)
b) External comparison - (e.g., factory workers vs. national
registry).
c) Comparison with general population rates - Compares
study group to standardized population data (e.g.,
firefighters vs. India population mortality rates).
d) Multiple comparisons - (e.g., drug trial vs. placebo and
another drug).
 Steps in a cohort study
4. Follow up - Uniform and complete follow up of both
exposed and not exposed groups - be done.
 Complete ascertainment of outcome events in
exposed and non-exposed group is essential.
 Standardized diagnosis of outcome events in
exposed and non-exposed groups - be used.

Follow up – by
a. Periodic medical examination
b. Reviewing physician & hospital records
c. Routine surveillance of death records
d. Mailed questionnaires, telephone calls
& periodic home visits
5.Analysis: 1. Frame work of a cohort study
Disease
Cohort Total
+ (Yes) - (No)
Exposed to Risk
a b a+b
Factor
Not Exposed to Risk
c d c+d
Factor
II. Calculation of Incidence -
Incidence of disease among exposed group (study cohort)
IE = a /a+b
Incidence - among non-exposed group (control cohort)
INE = c / c+d
 Steps in a cohort study
5. Analysis - 1.E.g.
A cohort study on cigarette smoking and lung
cancer among 13,000 individuals.7000 were
smokers and 6000 were non-smokers .

Lung Cancer
Cigarette Total
Smoking + -

Yes 70 6930 7000

 Steps in a cohort study
Calculation of Incidence
IE = 70 / 7000 = 10 /1000 population = 0.01
INE = 6 / 6000 = 1/1000 population = 0.001
Estimation of Risk – a. Relative Risk / Risk Ratio (RR)
Incidence of disease
among exposed (IE )
RR = Incidence of disease among
non – exposed (I NE )
RR = 0.01/0.001 = 10
 Interpretation of Relative Risk (RR)
If RR = 1 - No association
If RR = >1 - indicates positive association
between exposure and disease (risk)
If RR = <1 - indicates negative association
and estimates protective effect / benefit
 The risk of lung cancer is 10 times more for smokers
compared to non-smokers.
 95 % CI of RR = 4.360 --- 22.935 and p = 0.001
 b . Attributable Risk percent (AR %)

IE - INE
AR% = x 100
IE
0.01 - 0.001 x 100 = 90 %
0.01
 90% of all lung cancers is due to smoking
 Types of cohort studies
1. Prospective (Concurrent) Cohort study
Time frame for a hypothetical
prospective study begun in 2009
200 Defined population
9

2019 Exposed Non-exposed

2029 Disease No No
disease Disease
disease
 Types of cohort studies
2. Retrospective (Historical) cohort study
Time frame for a hypothetical retrospective
cohort study begun in 2009
Defined population 1989

Exposed Non- exposed 1999

Disease No
Disease No 2009
disease
disease
 Types of cohort studies
3. Ambispective cohort study - Combination of
prospective and retrospective cohort design.

Data on exposure have been collected in the past

and are available from existing records

The investigator follow the cohort for subsequent

occurrence of disease.
 Applications of cohort studies
When good evidence suggest an association between
suspected risk factor and disease.

When interval between exposure and outcome is short

e.g. Rubella infection during pregnancy and

development of congenital malformations in offspring

Direct measurement of incidence

Advantages of cohort study
To calculate incidence
Several possible outcomes related to exposure
can be studied simultaneously
Relative risk can be directly estimated
Dose response ratios can be calculated
Bias can be minimized.

Disadvantages -
Involves a large number of people
Long time for completion
Loss to Follow up (Attrition)
Administrative/ logistic problem
 Case Control studies
 Begins with(cases) who have disease & comparison
group (controls) who do not have disease
 Previous exposures of both groups are investigated
 If previous exposure is more common among the
cases, then it is evidence that exposure caused the
disease
 Case control studies enhance the understanding
of cause – effect relationship in many chronic
disorders 19
 Case Control study Design

Exposure present Cases

(disease)

Exposure absent
Study
population
Exposure present
Controls
(no disease)
Exposure absent

Inquiry

Study begins here

 Course of Case Control Studies

 Selection of cases
 Selection of controls
 Information on Exposure
 Analysis
 Selection of cases
 Case definition (diagnostic criteria)

-precise definition for cases

- Objective evidence preferred ( HPE in Cancer)
 Criteria for study population (sampling frame –
Source)
Defined population (Primary study base)
Convenient source of eligible cases ( secondary study
base)
 Selection of cases
 Defined population (Primary study base)

Sources of cases -
Geographically defined population
Membership of a health care plan
Occupational group
Cancer registry
Children in school
 Convenient source of eligible cases ( secondary study base)

Hospital or General practice

 Selection of cases
 Select all consecutive eligible cases diagnosed in a given
period or select a random sample representing all eligible
cases identified
 Ideally only incident cases /recent cases - be enrolled
 However in some circumstances - include prevalent cases
- If not possible to find exact date of onset of disease
- Prevalent cases diagnosed closest to the initiation of
the study to be included
 Cases should be chosen independently of exposure
 Selection of controls
At - time of selection control - be free of disease under study

Controls should come from study base as that of cases

Controls should represent disease free people in the study

base ( sampling frame ) rather than all diseases free people.

Make sure that controls do not have the disease under

study by using appropriate methods / objective methods
 Selection of controls

Sources of controls -

 Definable population – Population controls

 Hospital controls
 Special groups controls
 Selection of controls
Population control -
If cases are selected from a definable population,
controls can be selected from the same population
 Geographically defined population
 Membership of a health care plan
 Occupational group
 Cancer registry
 Children in school
 Selection of controls
Hospital control –

1. If cases are from hospital source, controls are chosen from

people who if they develop the disease under study would
have received care at these hospitals.

eg- cases are patients diagnosed with lung cancer at a

specific hospital. Controls are selected from patients admitted
to the same hospital for non-cancer conditions (e.g., fractures
or appendicitis) who would have been diagnosed at this
hospital if they had lung cancer.
 Selection of controls
Hospital control –

2. When cases are chosen from a narrow range of

providers of health care, controls are often chosen from
patients with other illness treated by the same health care
provider.

3. Controls may also be chosen from Individuals who are

tested for the presence of the disease under study and are
found not to have it (e.g. CHD)
 Selection of controls
Hospital control –
4. A. Cases – be taken to omit controls with conditions
known to be related to the exposure under study

eg- compare smoking prevalence between bladder cancer

cases and controls. If controls with COPD / lung cancer
were included, their higher smoking rates could make
smoking appear less strongly associated with bladder
cancer, leading to biased results. Excluding such controls
helps accurately estimate the odds ratio for smoking as a
risk factor for bladder cancer..
So, To minimize selection bias relating to having chosen ill
controls an attempt can be made to omit potential controls
with conditions known to be related to the exposure
 Selection of controls

Special groups control -

 Friends
 Neighbors
 Spouse
 Relatives

Controls must be selected independently of their exposure

 Selection of controls
Matching -
Controls may be selected in such a way that
certain risk factors (confounders) are distributed
equally among cases & controls.
Group matching (frequency matching)
Pair matching (Individual matching)
Can we have more than one control group? - yes –
if any specific need only
 Case Control Studies
Sample size –
Probability of disease in exposed
Probability of disease in not exposed
Anticipated odds ratio

Control to case ratio - Optimal ratio 1:1

Not efficient beyond 4:1 ratio
 Information on Exposure
Assess prior exposure status and relevant
characteristics of the cases and controls
Care must be taken not to include exposures that
took place after the illness began
 Presence of exposure
 Intensity of exposure
 Duration of exposure
 Information on Exposure

 Self Reported
 Interview (Questionnaires)
 Records (Medical / Occupational)
 Biological markers
 Clinical examination
 Environmental data
 Analysis
Compare prior exposure experiences of cases and
controls to estimate the association between the
exposure (Risk factor) and the disease
The strength of association is estimated by Odds
Ratio (OR)
OR = 1 (No association)
OR > 1 (Positive association)
OR < 1 (Negative association)
 CHD + CHD - ve
ve
C 152 110
Periodontitis
+ve
C 36 70
Periodontitis
-ve
152 X 70
Total Odds ratio = 188 180
---------- = 2.7
36 X 110

95% CI = 1.6 to 4.4 and P value = < 0.001

 Case Control Studies

Bias (Systematic error)

 Selection bias
 Measurement bias
 Confounding bias
 Case Control Studies
Advantages -
 Relatively less time consuming & less expensive
 Evaluation of diseases with long latent periods
 Optimal for evaluation of rare diseases
 Can examine multiple risk factors for a single
disease

Other uses -
 Evaluation of interventions (Validity low)
 Evaluation of Diagnostic/Screening test
 Investigation of disease outbreak
 Case Control Studies
Limitations -
 Inefficient for evaluation of rare exposures
 Temporal relationship between exposure

and disease may be difficult to establish

 Particularly prone for selection bias and recall bias
 No measurement of incidence

Thank you