Reducing population stratification bias: stratum matching is better than exposure

ObjectiveGenetic studies of complex human diseases rely heavily on the epidemiologic association paradigm, particularly the population-based case–control designs. This study aims to compare the matching effectiveness in terms of bias reduction between exposure matching and stratum matching.Study Design and SettingFormulas for population stratification bias were derived. An index of matching effectiveness was constructed to compare the two types of matching.ResultsIt was found that exposure matching can paradoxically increase the magnitude of population stratification bias sometimes, whereas stratum matching can guarantee to reduce it.ConclusionThe authors propose two simple rules for genetic association studies: (a) to match on anything that helps to delineate population strata such as race, ethnicity, nationality, ancestry, and birthplace and (b) to match on an exposure only when it is a strong predictor of the disease and is expected to have great variation in prevalence across population strata.