Design options and methodological fallacies in the studies of reproductive failures.

Reproductive failures are at first sight well suited for epidemiologic research. The time of pregnancy is closely monitored, and failures such as spontaneous abortions and subfecundity are rather frequent. Although epidemiologists' interest in the field has been growing, there is still disappointingly little new information of relevance for prevention. A number of methodologic shortcomings may explain this. A large part of disease classification is not well suited for etiologic research, reduced fertility has diminished the populations at risk, close medical monitoring tends to mask causal links, and many scientific problems related to this area bring limitations to the research field. Still, much more could be learned from a systematic use of epidemiologic knowledge, existing registers, and the joint effort between different research groups.