Reproducibility Package for "Beyond Point-and-Click: Interface-Mediated Analytic Constraints in Federated Electronic Health Record Research—A Monte Carlo Simulation Study"

This record contains the R code, replication-level simulation results, scenario-level performance metrics, workflow-decomposition results, conceptual-domain mapping, figures, and software-environment information supporting the manuscript “Beyond Point-and-Click: Interface-Mediated Analytic Constraints in Federated Electronic Health Record Research—A Monte Carlo Simulation Study.”

The Monte Carlo study included six scenarios defined by three levels of additional hidden confounding and two propensity-score overlap conditions. Each scenario comprised 1,000 replications of 20,000 simulated patients under a homogeneous null treatment effect with a true hazard ratio of 1.00.

Four analytic workflows were evaluated: propensity-score matching using a limited covariate set (Method A), overlap weighting using the same limited covariates (Method A-OW), overlap weighting using expanded EHR-accessible covariates (Method B), and oracle overlap weighting that additionally included an otherwise unavailable socioeconomic variable (Method C).

No patient-level or third-party data were used. The archive includes replication-level results for all four analytic workflows, aggregate performance results, descriptive workflow contrasts, final figures, software-version information, SHA-256 checksums, and a verified post-processing script that reproduces the aggregate results from the replication-level output.