Improving Self-reported Prescription Medicine Data Quality with a Commercial Database Lookup Tool and Claims Matching

Abstract

Estimating prescription medicine use is challenging due to recall bias associated with surveys and coverage bias in administrative data. This study assesses how making operational improvements and combining both survey and administrative data sources can increase data quality on filled prescriptions. We use data from the Medicare Current Beneficiary Survey (MCBS) and administrative data from the Centers for Medicare and Medicaid Services (CMS). First, we investigate improvements from a prescription medicine lookup (PMLU) tool integrating a commercial medicine database into the MCBS. We then examine impacts of matching survey-reported medicines to Part D claims. We find that the PMLU improves accuracy and reduces measurement bias. Claims matching identifies additional medicines, especially for beneficiaries with more chronic conditions and medicines. This study shows that integrating a commercial database and supplementing with administrative data improves data quality and reduces sources of error.