Cohort study
- Study design
Exposed and unexposed people are followed over time and investigators count who falls ill. It measures incidence and the relative risk.
What it is
In a cohort study people free of the outcome are enrolled, classified by exposure and followed to see who develops it. Because exposure is known before the outcome, it establishes the time sequence and measures incidence and the relative risk (RR). It can be prospective (followed into the future) or retrospective (historical, using past records). It is ideal for rare exposures but slow and costly for rare diseases.
How it is built
- People free of the outcome are recruited.
- Exposure is measured at baseline (and sometimes during follow-up).
- Everyone is followed for the same time, or each person's person-time is recorded.
- Outcomes are counted and incidences compared with the RR or the rate ratio.
Worked example
- A cohort of 2 000 workers without tuberculosis: 500 exposed to silica dust and 1 500 unexposed, followed for 10 years.
- Tuberculosis develops in 20 exposed and 15 unexposed workers.
- Cumulative incidence: 20/500 = 4% in the exposed; 15/1 500 = 1% in the unexposed.
- RR = 4% / 1% = 4.0.
- Risk difference = 4% − 1% = 3 percentage points over 10 years.
- Reading: silica exposure was associated with 4 times the 10-year risk of tuberculosis in this cohort.
Strengths and weaknesses
| Strengths | Weaknesses |
|---|---|
| Establishes the time sequence | Slow and costly |
| Measures incidence, RR and risk difference | Inefficient for rare diseases |
| Ideal for rare exposures | Loss to follow-up |
| Studies several outcomes of one exposure | Still observational: confounding is possible |
Markers
- RR
- Incidence
- Person-time
When to use it
- Much of the major evidence on tobacco, blood pressure and cholesterol comes from cohorts followed for decades.
- A retrospective cohort uses records (occupational, clinical) to rebuild exposures and outcomes that have already happened.
Common pitfalls
- Including people who already had the outcome at baseline: they inflate incidence and break temporality.
- Ignoring that those lost to follow-up may differ from those who stay.
- Calling a retrospective cohort a case-control study: the design is defined by selection on exposure.
Test yourself
Think of the answer, then tap the question to see it.
What is the typical measure of association in a cohort study?
The relative risk (or the rate ratio when person-time is used).
What separates a retrospective cohort from a case-control study?
A cohort selects by exposure and looks at the outcome; a case-control study selects by outcome and looks at exposure.
For what kind of exposure is a cohort especially useful?
For rare exposures, such as a specific occupational exposure, because the exposed can be recruited on purpose.
What is the most characteristic bias of cohort studies?
Bias from loss to follow-up, above all when losses differ between exposed and unexposed.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Principles of Epidemiology | Lesson 1 - Section 7 (Centers for Disease Control and Prevention, 2012); Principles of Epidemiology | Lesson 3 - Section 5 (Centers for Disease Control and Prevention, 2012); Attrition bias | Catalog of Bias (Catalogue of Bias Collaboration, 2017).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.