What is a Cohort Study?

1. Start with a group of healthy people
2. Divide them based on whether they are exposed or not exposed to a risk factor
3. Follow them forward in time
4. See who develops the disease

Think of it like watching two trains leave the same station - one carrying smokers, one carrying non-smokers - and seeing which train has more passengers fall sick by the end of the journey.

Simple Analogy

You take 1,000 factory workers who are all currently healthy.

• 500 work near radiation (Exposed)
• 500 do not (Not exposed) You follow them for 10 years and count how many develop cancer in each group.

Three Key Features

When is it Used?

• When there is existing evidence of an association from descriptive or case-control studies
• When the exposure is rare but incidence is high (e.g., radiation workers, doctors)

Elements of a Cohort Study

Who is studied?

• General population - from a defined geographical, political, or administrative area
• Special groups - doctors, nurses, factory workers (homogeneous and easy to follow up)
• Exposed (high-risk) group - e.g., radiologists, industrial workers

How is data collected?

• Interviews, record reviews, medical examinations, environmental surveys
• Exposure is classified as: Exposed vs. Unexposed, or by level/degree

Comparison Groups

• Internal comparison (within the same group) - stratified by level/duration of exposure
• External comparison - e.g., smokers vs. non-smokers outside the cohort
• General population rate - used when no comparison group is available

Follow Up

• Periodic medical examination
• Reviewing physician/hospital records
• Death record surveillance
• Questionnaires, telephone calls, home visits

2x2 Table for Analysis

Key Measures Calculated

• Among exposed = a / (a+b) × 1000
• Among non-exposed = c / (c+d) × 1000

RR = Incidence rate in exposed / Incidence rate in non-exposed = a/(a+b) ÷ c/(c+d)

• RR = 1 → No difference between groups
• RR > 1 → Exposed group has higher risk
• RR < 1 → Exposure is protective

AR = IR in exposed - IR in non-exposed Tells how much of the disease is due to the exposure

Advantages of Cohort Study

1. Incidence can be calculated
2. Several possible outcomes can be studied
3. Direct estimate of relative risk
4. Dose-response relationship can be calculated
5. Minimizes misclassification bias (exposure assessed before disease)

Disadvantages of Cohort Study

1. Large number of subjects needed
2. Not suitable for uncommon/rare diseases
3. Long time required to complete
4. Attrition (loss to follow-up) is high
5. Selection bias - volunteers may not be representative
6. Ethical problems (diagnostic criteria may change over time)
7. Expensive
8. Concentrates on a limited number of factors relating to disease outcome

Cohort vs. Case-Control (Quick Comparison)

Steps of a Cohort Study (Think of it as a Story)

Step 1 - Select Your Study Subjects (The Cohort)

• 🟥 Exposed group - people who have the risk factor (e.g., smokers)
• 🟩 Non-exposed group - people without the risk factor (e.g., non-smokers)

Example: You pick 1,000 factory workers. 500 work near asbestos (exposed), 500 work in the office (not exposed). All 1,000 are currently healthy.

Step 2 - Obtain Data on Exposure

• Interviews
• Medical records
• Special tests or examinations
• Environmental surveys

• Simply: Exposed vs. Not exposed
• Or in detail: Low exposure / Medium / High exposure (for dose-response analysis)

Example: You measure how many hours per day each worker is near asbestos dust.

Step 3 - Select a Comparison Group

Step 4 - Follow Up the People Over Time

• Who develops the disease?
• Who stays healthy?
• Who dropped out? (loss to follow-up)

• Regular medical check-ups
• Hospital/physician records
• Death certificates
• Phone calls, questionnaires, home visits

Example: You track all 1,000 workers every year for 10 years and record who develops lung cancer.

Step 5 - Analysis (The Maths Part)

A) Incidence Rate (IR)

How many people got the disease per 1000?

• IR in exposed = a ÷ (a+b) × 1000
• IR in non-exposed = c ÷ (c+d) × 1000

B) Relative Risk (RR)

How many times MORE likely is the exposed group to get the disease?

Example: RR = 8 means smokers are 8 times more likely to get lung cancer than non-smokers.

C) Attributable Risk (AR)

How much of the disease is directly CAUSED by the exposure?

Example: If 300 per lakh smokers get lung cancer and 10 per lakh non-smokers get it, AR = 290 per lakh. That 290 is directly attributable to smoking.

D) Population Attributable Risk (PAR)

What proportion of ALL disease cases in the population is due to this exposure?

Useful for public health decisions - tells you how much disease would disappear if you removed the exposure.

Summary - The 5 Steps at a Glance

Step 1 → SELECT healthy people, divide into Exposed vs. Not Exposed
Step 2 → COLLECT data on their exposure level
Step 3 → CHOOSE a comparison group
Step 4 → FOLLOW UP over time, track who gets sick
Step 5 → ANALYSE using 2×2 table → calculate IR, RR, AR, PAR

Memory Tip

"S-O-S-F-A" Select subjects → Obtain exposure data → Select comparison group → Follow up → Analyse