[Problems of the number of cases in drug surveillance].

The purpose of drug surveillance studies is to ascribe the adverse event observed to a pharmacological cause and to try to evaluate its frequency in order to draw reliable conclusions. The present study shows that epidemiological investigations of the cohort or case-control type not only are of poor warning value but also require so many patients that the information required is obtained very slowly and at a very high cost. Despite its imperfections, spontaneous reporting seems to be the most effective method of post-marketing drug surveillance.