Problems and pitfalls in conducting epidemiological research in the area of reproductive toxicology.

Defective reproductive outcome continues to represent a considerable public health burden, as well as a source of intense personal tragedy and enormous expense in our society. Evidence suggests that the rate of reproductive failure may be modifiable, in part by limiting human exposure to environmental agents toxic to the reproductive process. Methodological problems posing difficulties to researchers investigating the association between exposure to such agents and adverse reproductive outcome are discussed, including inadequacies in existing data bases available for analysis; bias in reporting by or in the collection of information from affected families; and weaknesses in the selection of study populations, "control" groups, and outcomes chosen for observation. Suggestions for possible resolution of some of these difficulties are discussed.