Combining Effect Estimates Across Cohorts and Sufficient Adjustment Sets for Collaborative Research: A Simulation Study.

摘要: Background Collaborative research often combines findings across multiple, independent studies via meta-analysis. Ideally, all study estimates that contribute to the meta-analysis will be equally unbiased. Many meta-analyses require measure same covariates. We explored whether differing minimally sufficient sets of confounders identified by a directed acyclic graph (DAG) ensures comparability individual estimates. Our analysis applied four statistical estimators multiple adjustment in single DAG. Methods compared obtained linear, log-binomial, and logistic regression inverse probability weighting, data were simulated based on previously published Results results show weighting generally provide estimate effect for different estimands are adjust confounding bias, with modest differences random error. In contrast, performed poorly, notable from unique sets, larger standard errors than other estimators. Conclusions do not support reliance collaborative meta-analyses. Use DAGs identify potentially can allow without requiring exact