Genetic confounding in health disparities research

Abstract

Abstract Background Health disparities are defined as health differences that adversely affect socially disadvantaged populations, and health disparities research is focused primarily on social and environmental determinants of health. We hypothesized that unmeasured genetic differences between population groups are likely to be a major source of hidden confounding for observational studies of health disparities. Results Our study cohort consisted of 26,912 UK Biobank participants from Asian, Black, and White ethnic groups. We analyzed outcomes for 1,536 diseases and discovered numerous health disparities that affect socially disadvantaged Asian and Black UK ethnic groups. We modeled outcomes for the top twenty Asian-White and Black-White health disparities using genetic and socioenvironmental risk factors to test for genetic confounding. We found that genetic diversity and differences in socioenvironmental risk factors are correlated between UK ethnic groups and demonstrate how genetic confounding can lead to spurious associations between social disadvantage and genetically influenced disease disparities. Specifically, inclusion of genetic data in disease risk models attenuates the effect of socioeconomic deprivation (SED) on most of the top health disparities, including spurious associations of SED with sickle cell disease and skin cancer. Moreover, comparisons of disease models with SED alone versus models with SED and genetic data together indicate that all the top health disparities analyzed here are more accurately modeled when genetic confounders are included. Conclusions Our results support an integrated approach to health disparities research that incorporates genetic, social, and environmental data whenever possible.