De la validité des études épidémiologiques

Abstract

The physician must act as primary medical advisor on behalf of the patient. Because the basis of prevention is epidemiology, problems of validity in the epidemiologic literature ought to be recognized by the primary medical advisor, so that he may optimize prevention. This article addresses the question of validity from the standpoint of the material, analysis, and interpretation. Validity of the material is discussed from four main points of reference: the diagnostic criteria of the disease, selection biases, misclassification of exposure, and control group relevance. Validity of the analysis is considered from the standpoint of the etiologic hypothesis; the notion of the “expected minimum analysis” is discussed. The three main aspects of this are the control of confounders, the introduction of test-variables, and the concept of refinement of the independent and dependent variables. Validity of interpretation is presented as a function of the introduction of test-variables. A relevant analysis may help discriminate between alternate interpretations, defined as the inverse path relationship, the confounded relationship, and the common cause pathway.