Validity of reported Medicare Part D enrollment in the Medical Expenditure Panel Survey

Abstract

The authors validate reported Part D coverage in the Medical Expenditure Panel Survey (MEPS) and assess the impact of misreporting on descriptive and behavioral analyses. MEPS participants with Medicare coverage during 2006 to 2007 were matched to Medicare administrative data. A summary measure of Part D coverage based on several questions has substantial validity (κ = .70) and an agreement rate of 85.1%. Some beneficiaries confused Part D and private drug coverage, leading to both under- and overreported Part D coverage. Accuracy varies little by sociodemographic group. Standard regression models of the determinants of Part D enrollment were estimated with both MEPS-based and administrative data-based measures of Part D enrollment. In this analysis, the signs of the marginal effects were the same, the magnitudes were similar, and mostly the same variables had statistically significant effects in both regressions. Thus, behavioral analyses are largely unaffected by misreporting.