Impact of Part D on previously uninsured Medicare beneficiaries with hypertension

Objectives US Medicare Part D began in January 2006, but beneficiaries could wait until May to sign up without being assessed a premium penalty. The availability of this penalty-free period offers a unique opportunity to study selection patterns and demand response to Part D coverage. The objective is to examine factors associated with early and late Part D enrolment and assess impact of enrolment timing on changes in drug use. Methods Nationally representative sample of community-dwelling Medicare beneficiaries with hypertension and no drug coverage in 2005 (n = 641).We used 2005 and 2006 Medicare Current Beneficiary Survey data with a multinomial logistic model to predict Part D enrolment and a difference-in-difference (DID) model to compare changes in monthly antihypertensive drug fills pre- and post-enrolment. Key findings Of the sample enrolled in Part D in 2006, 63.3% (23.5% by January and 39.8% late). Early enrolees had higher antihypertensive use prior to Part D (1.6 monthly fills) compared to late enrolees (1.2) and non-enrolees (0.9). Post-Part D monthly antihypertensive use averaged 2.2 fills for early enrolees, 1.4 for late enrolees and 1.0 for non-enrolees. The estimated DID programme effect was 0.54 additional monthly antihypertensive prescription fills for early enrolees (P < 0.01) and 0.26 fills for late enrolees (P < 0.01). Conclusions The estimated Part D demand response from early enrolees was unrealistically high suggesting selection bias. The price elasticity for late enrolees (−0.29 to −0.43) was consistent with best evidence in the literature. Failure to consider selection can lead to overestimating the impact of Part D on drug use.