Cross-cohort mixture analysis: a data integration approach with applications on gestational age and DNA-methylation-derived gestational age acceleration metrics

Abstract

ABSTRACTBackgroundData integration of multiple epidemiologic studies can provide enhanced exposure contrast and statistical power to examine associations between environmental exposure mixtures and health outcomes. Extant studies have combined population studies and identified an overall mixture-outcome association, without accounting for differences across studies.ObjectiveTo extend the novel Bayesian Weighted Quantile Sum (BWQS) regression to a hierarchical framework to analyze mixtures across multiple cohorts of different sample sizes.MethodsWe implemented a hierarchical BWQS (HBWQS) approach that (i) aggregates sample size of multiple cohorts to calculate an overall mixture index, thereby identifying the most harmful exposure(s) across cohorts; and (ii) provides cohort-specific associations between the overall mixture index and the outcome. We showed results from six simulated scenarios including four mixture components in five and ten populations, and two real case-examples on the association between prenatal metal mixture exposure—comprising arsenic, cadmium and lead—and both gestational age and gestational age acceleration metrics.ResultsResults from simulated scenarios showed good empirical coverage and little bias for all parameters estimated with HBWQS. The Watanabe-Akaike information criterion (WAIC) for the HBWQS regression showed a better average performance across scenarios than the BWQS regression. HBWQS results incorporating cohorts within the national Environmental influences on Child Health Outcomes (ECHO) program from three different sites (Boston, New York City (NYC), and Virginia) showed that the environmental mixture—composed of low levels of arsenic, cadmium, and lead—was negatively associated with gestational age in NYC..ConclusionsThis novel statistical approach facilitates the combination of multiple cohorts and accounts for individual cohort differences in mixture analyses. Findings from this approach can be used to develop regulations, policies, and interventions regarding multiple co-occurring environmental exposures and it will maximize use of extant publicly available data.