Design and conduct of epidemiologic studies.

This article reviews the advantages and disadvantages of the experimental vis-a-vis the nonexperimental (observational) epidemiologic approach in the assessment of causal relationships. Nonexperimental epidemiologic strategies, including the study of ecologic correlations, concurrent and nonconcurrent prospective designs, as well as case-control and cross-sectional designs, are briefly discussed. Among the case-control designs, reference is made to matched and to "nested" case-control studies. These different design approaches are discussed in relation to issues of external validity, bias, confounding, interaction, and precision.