CMS Releases Bundled Payment Evaluation

The Centers for Medicare and Medicaid Services (CMS) released last week an evaluation of the Bundled Payments for Care Improvement (BPCI) program. The study, conducted by the Lewin Group, evaluated the performance of BPCI Models 2, 3, and 4, in the first year in which participants assumed financial risk. Overall, it found modest reductions in Medicare spending, some instances of reduced quality, and fewer instances of improved quality. Performance varied across the four models, with Model 2 showing the best results.

Under Model 2, all expenditures arising from an episode, including the inpatient stay in an acute care hospital plus the post-acute care and all related services, up to 90 days after hospital discharge are reconciled against a target price. Model 2 represented three-quarters of the episodes and half of the provider participants in BPCI. The study found that for Model 2 orthopedic surgery episodes, Medicare payments for the hospitalization and 90-days post-discharge were estimated to have declined more for episodes initiated at BPCI-participating hospitals than for episodes initiated at comparison hospitals. Most of the savings was attributed to reduced use of institutional post-acute care following the hospitalization. In addition, beneficiaries who received care at participating hospitals reported greater improvement in two mobility measures than beneficiaries from comparison hospitals. Among spinal surgery episodes, however, average Medicare payments increased more for the hospitalization and the 90-day post-discharge period for the BPCI beneficiaries than for a comparison population.

Model 3 similarly involves retrospectively reconciling expenditures associated with an episode against a target price, but the episode begins after hospital discharge with the initiation of post-acute care services. Model 3 bundles resulted in reductions in standardized skilled nursing facility (SNF) payments and SNF days for SNF-initiated BPCI episodes across almost all clinical episode groups. However, unlike Model 2, it did not drive statistically significant declines in total episode payments. Model 3 bundles generally maintained or improved quality, except in three instances where quality declined relative to the comparison group.

In Model 4, payments are made prospectively rather than reconciled retrospectively. The episode encompasses all services furnished by the hospital, physicians, and other practitioners during the inpatient stay and any readmissions. This model experienced the lowest participation of the four BPCI models. Orthopedic surgery and cardiovascular surgery accounted for 81 percent of all Model 4 episodes. For the orthopedic surgery episodes, there were no statistically significant relative changes in Medicare payments, quality, or utilization. In the cardiovascular surgery episodes, payments and utilization increased, and certain functional outcomes declined, relative to the comparison.

The authors of the study cautioned that the short timespan covered by the analysis and the small sample sizes limit the ability to generalize from these results. With growing participation over time, the ability to discern and validate changes in payment and/or quality will grow.

Contact: Karen Lipson, klipson@leadingageny.org, 518-867-8383 ext. 124