[Oxygen therapy and prematurity: lessons from the past].

The history of oxygen therapy in the newborn clearly demonstrates the importance of the methodology used in clinical research. It has revealed the tragic mistakes which haunt every clinician confronted with pathology of unknown etiology. It has allowed us to determine the relative strength of the methods used in the search for various etiologies: passive or active observation, retrospective or prospective study and prospective randomized controlled trial. In addition, history has illustrated the danger of utilization of statistical tests in the absence of controlled interventions. The aim of this paper is to familiarize the practising physician with the basic concepts of epidemiology. This should allow him to be a better judge of the quality of recommendations made by "authorities". In the future, clinical epidemiology should be an essential part of medical training.