Medication Utilization of Dual Eligibles Before and After Medicare Part D: Cases of Antidepressants and Antipsychotics

Abstract

Objectives: The study objective is to examine whether medication utilization among dual eligibles was different under Part D compared to Medicaid period by focusing on states which vary baseline state Medicaid policies and wrap-around programs for dual eligibles to access drugs under Part D. The transition of prescription drug benefits to federal Medicare Part D from state Medicaid has potential to affect medication utilization for dual eligibles, beneficiaries for both Medicaid and Medicare programs. Changes in prescription drug benefits under Part D will not equally affect the dual eligible and will differ by states with differences in baseline Medicaid policies and availability of wrap-around programs to access drugs under Part D. The study focuses on antidepressants and antipsychotics. Methods: This study is a pre-post study design with a longitudinal dataset by linking Medicaid data for 2004-2005 and 5% random sample of Medicare data for 2006-2007. The study population is dual eligibles, existing users of antidepressants and of antipsychotics in 2004 and with enrollment from 2004-2007 in eight states. I employ a state-fixed effect model to estimate medication utilization using proportion of days covered (PDC), adjusting for beneficiaries characteristics and health status. I adopt generalized estimation equation (GEE) model for estimating PDC and spline regression for investigating whether changes in PDCs were related to Part D. A stratified analysis is conducted for community based dual eligibles (n=4,703 for antidepressants, n=2,301 for antipsychotics) and nursing home dual eligibles (n=1,504 for antidepressants, n=1,011 for antipsychotics) separately. Results: For antidepressants, adjusted changes in PDC were not significant for most states, except Arkansas community based dual eligibles (p=0.00), Florida nursing home dual eligibles had a significant increase (p=0.03), New Mexico nursing home dual eligibles(p=0.01) under Part D. For antipsychotics, adjusted changes in PDC were not significant in all states except Arkansas community based dual eligibles (p=0.03) and Florida for both groups (p=0.00 for both community based dual eligibles and nursing home dual eligibles). A separate spline regression for states that had significant changes in PDC from the state-fixed effect model indicated that trends in PDC before and after Part D were not significantly different, suggesting that changes were not related to transition to Part D. Conclusions: I did not find empirical support for concerns regarding disruption of medication utilization of the dual eligible under Part D. Although states had different baseline Medicaid polices and wrap-around programs under Part D, lack of significant changes in utilization suggest that minor changes in copayments and refill/prescription limits etc. do not have large effect on medication utilization to antidepressants and antipsychotics.