THE RECENT report that coffee may cause pancreatic cancer1 was presented in a pattern that has become distressingly familiar. The alleged carcinogen is a commonly used product. The report was given widespread publicity before the supporting evidence was available for appraisal by the scientific community, and the public received renewed fear and uncertainty about the cancerous hazards lurking in everyday life.
The research on coffee and pancreatic cancer was done with the case-control technique that has regularly been used in epidemiologic circumstances where the more scientifically desirable forms2 of clinical investigation—a randomized controlled trial or a suitably performed observational cohort study—are either impossible or unfeasible. In case-control studies, the investigators begin at the end, rather than at the beginning, of the cause-effect pathway. The cases are selected from persons in whom the target disease has already developed. The controls are selected from persons in whom that disease has not