[Pharmacoepidemiology: definitions, problems, methodology].

Pharmacoepidemiology aims to complete the evaluation of drugs made before approval, by providing reliable information concerning effectiveness, safety and utilization of medicines in realistic conditions. The goal may be only descriptive or aetiological. In the latter, the conclusions from observational studies can be jeopardized by systematic errors and cannot achieve the robustness of experimental designs. According to directionality, three main types of studies can be identified: cross-sectional, prospective and retrospective. Prospective and retrospective studies can be based on a single group (descriptive studies) or include a reference group (comparative or aetiologic studies). The interest of prospective studies is reduced when (1) the incidence of the considered event becomes low or (2) one intends to assess the effects of various causal factors. Retrospective studies are approaching their limits when (1) the prevalence of the exposure is low in the source-population or (2) several events or outcomes are concerned.