Case-control research. Temporal precedence and other problems of the exposure-disease relationship.

We assessed the principle of temporal precedence in recent case-control studies demonstrating the alleged associations between tampon use and toxic shock syndrome and between aspirin use and Reye's syndrome. For both relationships, we considered four components of the exposure-disease association, including: (1) establishing that the agent preceded the disease, (2) selecting an index time, (3) defining criteria for classifying a patient as "exposed," and (4) avoiding the bias that occurs when use of the etiologic agent was influenced by an early manifestation of the disease. The problems can be minimized by interviewing patients early during the course of their illness and by improving strategies for data analysis.