Perspective: cohort studies.

This issue of Epidemiologic Reviews on cohort studies is published five decades after the initiation of the Framingham study, one of the landmark investigations of epidemiologic research and, more generally, contemporary biomedical research. Within a few years other cohort studies—for example, of British physicians and of the atomic bomb survivors—were initiated. As follow-up of the participants in some of these studies has continued, the studies provide information on changing risks over time and the modification of risks by increasing age. The findings from many subsequent cohort studies have been widely applied as bases for developing public health policy. Evidence from cohort studies has tended to have greater credibility as a basis for decision-making than that from other epidemiologic designs, in part because cohort studies can directly describe the sequence from exposure to disease. Thus, the place of cohort studies in epidemiologic research is assured, and cohort studies are certain to figure prominently in future epidemiologic and clinical research. In spite of its widespread application, there have been remarkably few texts and collections of papers on this central study design. Also lacking are contributions to the literature that link new developments in statistical methods for longitudinal data analysis with epidemiologic applications. The contributions to this issue of Epidemiologic Reviews provide a broad review of epidemiologic aspects of the cohort study that should complement the literature on statistical techniques for longitudinal data. The papers address both the conduct of cohort studies and conceptual issues in the design of cohort studies and in the interpretation of their findings. Two contributions, those authored by Thomas and by Munoz and Gange, focus on analytical issues. Considered together, the papers in this volume provide a starting point for learning about the use of the cohort