La Revue de médecine interne
-
This paper describes the hierarchy of analytical studies, from the observational studies to the experimental ones. ⋯ Case-control studies, the most frequently performed among analytical studies, may generate or confirm hypotheses. They are of particular use for rare diseases. Cohort studies aim mainly at confirming hypotheses already tested, and at precisely quantifying the magnitude of effect. Randomisation aims at the elimination of confounding factors, and meta-analysis, at decreasing potential selection biases and increasing power. Those goals may not always be achieved. PERSPECTIVES AND PROJETS: The study power, or the beta error, are major factors to be determined when designing a study, before any attempt of realization.