The evaluation of negative epidemiologic studies: the importance of all available evidence in risk characterization

The importance of publishing and utilizing all available epidemiologic evidence in risk assessment is recognized. This recognition includes the findings from negative epidemiologic studies, described as well designed and executed studies where the hypothesized association with an adverse health effect was not found or found to be very weak. In assessing negative evidence, the random (chance) and nonrandom (bias, confounding) sources of variation need to be considered as well as dilution effects, dose-response patterns, biologic plausibility, and methods for data pooling (meta-analysis). Meta-analysis may be a useful statistical approach for a systematic review of results from multiple studies, but cannot overcome the bias of missing evidence.