Case-control studies using other diseases as controls: problems of excluding exposure-related diseases.

Exclusion criteria are examined in case-control studies which include persons with other diseases in the control group. Theoretically, a sample of the study base should not exclude persons who develop exposure-related diseases before, during, or after the time period in which the case occurred. This principle also generally applies when controls are sampled from the subset of persons with other diseases generated by the same study base. In particular, if the incidence rates (assuming completeness of ascertainment) for all other disease combined are similar in the exposed and nonexposed populations, then controls should be sampled from all other diseases, regardless of their relation to exposure. On the other hand, if the composite incidence rate for other diseases is greater (or less) in the exposed population, then it may be necessary to exclude controls with diseases positively (or negatively) associated with exposure. Additionally, considerable care should be taken when sampling controls from persons with specific diseases which are apparently "unrelated" to exposure because confounding may be introduced by exposure-related determinants of the control disease(s). To remove such confounding, it is necessary to control for risk factors for both the study disease and the control disease(s).