Cohort Study

An observational study that follows an exposed and an unexposed group forward, measures outcome incidence in each, and compares them — yielding a relative risk. It can be prospective (assemble now, follow forward) or retrospective (use existing records to look back at exposure, then forward to outcome).

Two groups separated only by exposure status at the start, tracked over time, with the rate of new outcomes in each compared at the end. Because exposure is recorded before the outcome appears, a cohort establishes temporality — the cause precedes the effect.

Start at the left with people who do not yet have the outcome, split by whether they are exposed, then read rightward in time. Divide the outcome incidence in the exposed by that in the unexposed to get the relative risk: RR > 1 = exposure raises risk, RR < 1 = protective, RR = 1 = no association.

Cohorts are the natural design for prognosis and harm — where randomising the exposure would be unethical or impossible — and handle common and multiple outcomes well. But because allocation is not random, any difference may be driven by confounders rather than the exposure itself.

• Direction: exposure → outcome (forward in time)
• Effect measure: relative risk (or rate ratio)
• Best for: prognosis, harm, common & multiple outcomes

Framingham Heart Study (from 1948) — a prospective cohort that enrolled 5,209 outwardly healthy residents of Framingham, Massachusetts and followed them for decades. By comparing outcome incidence across exposure groups it established that smoking, hypertension and high cholesterol predict cardiovascular disease — it gave medicine the very phrase “risk factor”. No one could ethically randomise people to smoke for 30 years — this is exactly the kind of harm/prognosis question only a cohort can answer.

• Confounding — assuming an association is causal when an unmeasured factor differs between the groups.
• Attrition / loss to follow-up — if dropouts are related to exposure or outcome it biases the result.
• Healthy-worker effect — an “exposed” occupational group looks healthier simply because the sick never started or left the job.