cohort study

The Centre for Evidence-Based Medicine at the University of Oxford describes a cohort study as one where data are obtained from groups who have been exposed, or not exposed, to the factor of interest, often from existing databases. The defining feature is what the researchers do not do: they observe an exposure that people already have, rather than allocating it. That makes the design available where a trial would be unethical or impossible, which covers most questions about how people live.

Timing separates the two kinds. In a prospective cohort the outcomes are determined after the exposure is recorded, so the sequence is fixed by the calendar rather than by anyone's memory. A retrospective cohort reaches back, establishing exposure from records or recall after the outcomes are already known. The centre places cohort studies below randomised controlled trials in the hierarchy of designs, because there is no random allocation, and above case-control studies, which start from the outcome and work backwards and so find the temporal order harder to establish.

Their strengths are the clear direction of time, the ability to study natural progression, standardised assessment and a cost below that of a trial. Their weakness is the one the absence of randomisation guarantees: hidden confounders linked to the exposure, which no amount of adjustment can be shown to have removed. Finding a suitable comparison group is difficult, blinding is usually impossible, and a rare outcome demands very large numbers.