An observational study that follows a defined group of people over time to track health outcomes.A cohort study is a form of observational research in which a defined group of people — the cohort — is identified and then followed forward in time while researchers record exposures and health outcomes. The defining feature is direction: participants are grouped before anyone knows who will develop the outcome of interest, and the study watches what happens. This contrasts with case-control studies, which start from people who already have a disease and look backwards, and with randomised trials, which assign the exposure rather than merely observing it.
Cohort studies matter because they can measure risk directly. If a cohort of non-smokers and a cohort of smokers are followed for twenty years, the proportion developing lung cancer in each group is an actual incidence, and their ratio is a genuine relative risk. They can also track many outcomes at once from a single exposure, and they can study exposures that would be unethical to assign, such as radiation or pollution. Their main weaknesses are cost and time — following thousands of people for decades is expensive — and the fact that unmeasured differences between exposed and unexposed people can masquerade as an effect.
Cohort studies explain much of what is known about the long-term health effects of smoking, diet, occupation and environmental exposure. They are also the backbone of longitudinal research on ageing and child development, where the same individuals are measured repeatedly across a lifetime.
The design is usually traced to the British Doctors Study begun by Richard Doll and Austin Bradford Hill in 1951, which followed tens of thousands of physicians and provided decisive evidence linking smoking to lung cancer. The Framingham Heart Study, launched in 1948 in a Massachusetts town, established the cohort approach as central to cardiovascular epidemiology.
Researchers first define the population and the exposure or exposures to be compared, then recruit participants and measure their baseline characteristics. Participants are followed at intervals, with outcomes recorded from medical records, examinations, questionnaires or death registries. Because exposure is measured before the outcome occurs, recall bias is largely avoided. Analysis compares outcome rates between exposure groups, adjusting statistically for confounding variables such as age, sex and socioeconomic status. A variant, the prospective cohort, follows people forward from the present; a retrospective cohort uses existing records to reconstruct a group's past and follows it to the present, which is faster but limited to data that were already collected.
Cohort studies underpin public-health guidelines on smoking, diet, air pollution and occupational hazards, and they supply the long-term safety data that regulators use for drugs and vaccines. Large biobanks such as UK Biobank are essentially very large cohorts with stored samples for genetic analysis.
The British Doctors Study ran for fifty years and was still publishing findings in the 2000s, long after most of its original participants had died.