What makes a good case-control study? Design issues for complex traits such as endometriosis.

The combined investigation of environmental and genetic risk-factors in complex traits will refocus attention on the case-control study. Endometriosis is an example of a complex trait for which most case-control studies have not followed the basic criteria of epidemiological study design. Appropriate control selection has been a particular problem. This article reviews the principles underlying the design of case-control studies, and their application to the study of endometriosis. Only if it is designed well is the case-control study a suitable alternative to the prospective cohort study. Use of newly diagnosed over prevalent cases is preferable, as the latter may alter risk estimates and complicate the interpretation of findings. Controls should be selected from the source population from which cases arose. Potential confounding should be addressed both in studies of environmental and genetic factors. For endometriosis, a possible design would be to: (i) use newly diagnosed cases with 'endometriotic' disease; (ii) collect information predating symptom onset; and (iii) use at least one population-based female control group matched on unadjustable confounders and screened for pelvic symptoms. In conclusion, future studies of complex traits such as endometriosis will have to incorporate both environmental and genetic factors. Only adequately designed studies will allow reliable results to be obtained and any true aetiologic heterogeneity expected to underlie a complex trait to be detected.