The sources and effects of bias and confounding in epidemiologic studies, and strategies to minimize their impact.