Bias in intervention studies that enroll patients from high-risk clinics.

Abstract

It is important to evaluate the effects of proposed interventions to reduce the risk of disease among carriers of a highly penetrant mutation, such as the mutations in BRCA1 and BRCA2 for breast and ovarian cancers or in APC and MLH1 or MSH2 for colon cancer. However, some studies that evaluate the effects of interventions designed to reduce risk in mutation carriers may be susceptible to a serious selection bias when they are based in clinics that care for persons at high risk for the disease. A study design in which a large fraction of the case patients were diagnosed before being seen at the clinic and all control subjects are persons previously seen at the clinic can create a false impression of intervention efficacy if, as is likely, mutation carriers seen at the clinic were more likely to receive the intervention than mutation carriers in the general population.