Sample selection and validity of exposure–disease association estimates in cohort studies

Abstract

Using a restricted source population for a cohort study will, under a range of sensible scenarios, produce only relatively weak bias in estimates of the exposure-disease associations.