Background/Purpose:  Biologic drugs can be obtained both through Medicare’s Part B medical benefit (physician administered drugs) or Part D pharmacy benefit (self-administered drugs).  Coverage for biologics is additionally complicated by the low-income subsidy (LIS), whereby Medicare beneficiaries with low socioeconomic status face minimal costs for Part D biologics (while non-LIS beneficiaries pay less out of pocket for Part B drugs). Despite the complexity of biologic benefit design, variation in biologic use and coverage in the Medicare program has not been adequately studied.  We performed a nationwide study of Medicare beneficiaries with rheumatoid arthritis (RA) to investigate geographic variations in payment source for biologic drugs. We hypothesized that there would be consistent use of Part D biologics among LIS patients.

Methods: We used data from nationwide Medicare fee-for-service claims files for 2009 for a 5% random sample of beneficiaries, including all medical and pharmacy claims.  Included beneficiaries had RA (2 face-to-face visits coded as 714.xx within the calendar year), were continuously enrolled in a pharmacy plan, and had at least one prescription for a disease-modifying drug.  Source of biologic coverage (Part B or Part D) was tabulated for all biologic users according to LIS status and stratified by state.  We calculated the proportion of LIS vs. non-LIS beneficiaries receiving Part B biologics in each state and examined the correlation between these variables within states.  Individuals residing in states with fewer than 50 eligible beneficiaries were censored from this analysis to increase the precision of our estimates.

Results: 7190 beneficiaries received a DMARD for RA; 1901 (26%) received biologic drugs. 25% of biologic users received a low-income subsidy (LIS).  Among LIS recipients, 8% received their biologic through Part B and 20% through Part D.  Conversely, among non-LIS patients, 18% received their biologic through Part B and 8% through Part D.  Across states, Part B biologic use ranged from 5-26% for non-LIS patients (median 9, IQR 14-22) and 0-5% (median 2, IQR 1-3) for LIS patients (Figure). Proportions of non-LIS and LIS patients using Part B biologics within a state were modestly correlated (r²=0.15).

Conclusion: We found substantial variations between states in the proportion of patients with RA receiving biologics, and whether patients received biologics through Medicare Part B or Part D.  Variations were even observed for those receiving LIS, suggesting that factors other than cost-sharing influence biologic drug selection.   Further studies should examine how both insurance coverage policies and physician practice variation impact the choices available to patients.