Dissecting heritability, environmental risk, and air pollution causal effects using > 50 million individuals in MarketScan

This study uses the MarketScan dataset, containing EHRs from over 50 million individuals, to investigate the interplay of genetics and environment in disease risk. A spatial mixed linear effect (SMILE) model, incorporating genetic relatedness and spatially correlated environmental factors (using geographical location as a proxy), was developed. The study analyzed 257,620 nuclear families and 1083 disease outcomes, augmented with environmental data (PM2.5, NO2, climate, sociodemographics). The SMILE model refined heritability estimates and quantified environmental contributions. Using wind speed and direction as instrumental variables, the causal effects of air pollution were assessed. PM2.5 and NO2 showed significant causal effects on 135 diseases, affecting distinct disease categories. The study demonstrates robust strategies for jointly modeling genetic and environmental effects using large EHR datasets.

Daniel McGuire, Havell Markus, Lina Yang, Jingyu Xu, Austin Montgomery, Arthur Berg, Qunhua Li, Laura Carrel, Dajiang J. Liu, Bibo Jiang