A structured approach to evaluating life course hypotheses: Moving beyond analyses of exposed versus unexposed in the omics context

BackgroundLife course epidemiology provides a framework for studying the effects of time-varying exposures on health outcomes. The structured life course modeling approach (SLCMA) is a theory-driven analytic method that empirically compares multiple prespecified life course hypotheses characterizing time-dependent exposure-outcome relationships to determine which theory best fits the observed data. However, the statistical properties of inference methods used with the SLCMA have not been investigated with high-dimensional omics outcomes. MethodsWe performed simulations and empirical analyses to evaluate the performance of the SLCMA when applied to genome-wide DNA methylation (DNAm). In the simulations, we compared five statistical inference tests used by SLCMA (n=700). For each, we assessed the familywise error rate (FWER), statistical power, and confidence interval coverage to determine whether inference based on these tests was valid in the presence of substantial multiple testing and small effect sizes, two hallmark challenges of inference from omics data. In the empirical analyses, we applied the SLCMA to evaluate the time-dependent relationship of childhood abuse with genome-wide DNAm (n=703). ResultsIn the simulations, selective inference and max-|t|-test performed best: both controlled FWER and yielded moderate statistical power. Empirical analyses using SLCMA revealed time-dependent effects of childhood abuse on DNA methylation. ConclusionsOur findings show that SLCMA, applied and interpreted appropriately, can be used in the omics setting to examine time-dependent effects underlying exposure-outcome elationships over the life course. We provide recommendations for applying the SLCMA in high-throughput settings, which we hope will encourage researchers to move beyond analyses of exposed versus unexposed. Key messagesO_LIThe structured life course modeling approach (SLCMA) is an effective approach to directly compare life course theories and can be scaled-up in the omics context to examine nuanced relationships between environmental exposures over the life course and biological processes. C_LIO_LIOf the five statistical inference tests assessed in simulations, we recommend the selective inference method and max-|t|-test for post-selection inference in omics applications of the SLCMA. C_LIO_LIIn an empirical example, we revealed time-dependent effects of childhood abuse on DNA methylation using the SLCMA, with improvement in statistical power when accounting for covariates by applying the Frisch-Waugh-Lovell (FWL) theorem. C_LIO_LIResearchers should assess p-values in parallel with effects sizes and confidence intervals, as triangulating multiple forms of statistical evidence can strengthen inferences and point to new directions for replication. C_LI