A hierarchical meta‐analysis for settings involving multiple outcomes across multiple cohorts

Evidence from animal models and epidemiological studies has linked prenatal alcohol exposure \PAE) to a broad range of long-term cognitive behavioral deficits. However, there is paucity evidence regarding the nature levels PAE associated with increased risk clinically significant To derive robust efficient estimates effects on function, we have developed hierarchical meta-analysis approach synthesize information cognition, integrating data multiple outcomes six U.S. longitudinal cohort studies. A key assumption standard methods meta-analysis, effect sizes are independent, violated when intercorrelated synthesized across Our involves estimating dose-response coefficients for each outcome then pooling these correlated obtain an estimated \global" cognition. In first stage, use individual participant by fitting regression that adjust potential confounding variables using propensity scores. The correlation matrix characterizing dependence between outcome-specific within run, while accommodating incomplete some outcome. We also compare inferences based proposed full multivariate analysis.