Objective:
Although Medicare Part D has improved medication adherence among the elderly, its effect on adherence disparities remains unknown. We sought to estimate the impact of Part D on the racial/ethnic disparities in adherence to cardiovascular (CV) medications among Medicare seniors.
Approach:
We analyzed annual data (2002-2010) from the Medical Expenditure Panel Survey (MEPS) on Medicare recipients (65+, “treated” group) and the near-elderly (60-64, control group), who were white, black, or Hispanic, and used ACE inhibitors/Angiotensin receptor blockers, statins, beta blockers, calcium channel blockers, or diuretics. Pooled 2002-2005 and 2007-2010 data covered the pre- and post-Part D periods, respectively. Drug class-specific and average overall adherence were measured over survey year as the proportion of days covered (PDC), then dichotomized at an 80% PDC threshold. Since MEPS had no days’ supply data before 2010, we derived refill days’ supply from dispensed quantities and validated it in 2010 sample. In survey-adjusted logistic regressions, we estimated Part D impact on disparities using a difference-in-difference-in-difference interaction term of race, post period, and treatment group. Following the Institute of Medicine, we differentiated between racial differences in adherence due to variations in demographics, health status, and beliefs across groups, and the inequitable disparities created by the differentials in socioeconomic position, experience with the healthcare system, and discrimination. Empirically, we used a rank-and-replace procedure to replace minority distributions of demographic and health characteristics with their white counterparts. Disparities were then computed as the adjusted adherence differences relative to whites.
Findings:
Our sample included 17,566 respondents, nationally representing 25 million. In the 2010 sample, continuous PDC and binary adherence distributions, based on actual and derived days’ supply, were very similar: Lin’s concordance coeff. 0.83 and C-statistic 0.93, respectively. Part D was associated with a reduction in Hispanic-white disparity in overall CV medication adherence by 15.38 percentage points (95%CI: 2.41, 28.36; P=0.02), and a non-significant increase in the black-white disparity by 5.32% points (95%CI: -17.19, 6.56; P=0.38). In sensitivity analyses, these effects were robust to various adjustments, and to including 2006 data in the post period. The largest reduction in Hispanic-white disparities was observed in adherence to beta blockers (28.9% points; 95% CI: 5.11, 52.69; P=0.02), whereas black-white disparities in statin adherence increased the most (14.7% points; 95%CI: -31.92, 2.52; P=0.09).
Conclusion:
With Part D, Hispanic-white adherence disparities appear to have been mitigated. Significant black-white disparities still persist post Part D, meriting further attention.
Decomposing racial/ethnic disparities in influenza vaccination among the elderly
