Following exposed and unexposed forward through time
Cohort studies identify groups differing in exposure status (smokers vs. non-smokers) at baseline and follow both forward in time, documenting outcome occurrence. This prospective direction mirrors causality (exposure precedes outcome), strengthening causal inference compared to retrospective designs. Participants start disease-free and are tracked to disease onset.
Cohort studies measure relative risk directly and suit common exposures and outcomes. They excel at identifying multiple outcomes from a single exposure and at detecting rare complications from common exposures. However, prospective follow-up is expensive, time-consuming, and subject to loss-to-follow-up bias when some participants drop out.
Confounding and adjustment challenges
People choosing different exposures often differ on many other factors (socioeconomic status, diet, exercise) that independently affect disease risk. These confounders create spurious associations if not adjusted. Cohort studies measure many variables at baseline, allowing statistical adjustment, but unmeasured confounding remains possible.
Cohort studies are stronger than case-control designs for causal inference because exposure precedes outcome and confounding can be quantified at baseline. However, cohort studies cannot distinguish correlation from causation without additional reasoning. Large sample sizes, long follow-up, and substantial costs limit cohort studies to high-priority research questions.