A study that makes use of historical information. For example, if we are interested in the life expectancy of patients then it is convenient if all the patients have died, since only then will the lengths of their lives be known.

In a case-controlled study it is the outcomes (for example, death from a particular disease) that determine the sampling procedure. In effect, time is reversed and the early background of those with the outcome (the cases) is compared with those without the outcome (the controls). The approach enables oversampling of the cases of interest.