Evaluation of Socioeconomic Status Indicators for Confounding Adjustment in Observational Studies of Medication Use

This methodologic study evaluated how different socioeconomic status (SES) indicators affect confounding adjustment in observational medication studies, finding that claims-based variables alone substantially improved balance on SES measures. While adding SES data further improved balance, it did not materially change outcome estimates for cardiovascular events or emergency department visits in a cohort of 7,109 Medicare patients initiating atorvastatin. The results suggest that claims-based proxies may be sufficient for mitigating SES confounding when aggregate-level SES information is unavailable in observational studies.