Issues in reporting epidemiological studies: Authors' reply.

EDITOR—We agree with Gillman that estimates and their confidence intervals contain the key information when reporting results. However, our observation that many epidemiological studies lack statistical power is still highly pertinent. The relevance of P values in epidemiologic publications will continue to be debated. Our view is that they are of value if used sparingly and interpreted wisely. We are surprised that Blackburn questions our premise that studies in clinical epidemiology in people …