Physician and Hospital Performance in Medicare's Updated Bundled-Payment Model for Joint Replacement.

Importance: Independent evaluations of Bundled Payments for Care Improvement Advanced (BPCI-A) have focused on hospitals and have not assessed the performance of physicians in participating physician group practices (PGPs). However, PGPs are accountable for a larger proportion of surgical procedures, including for lower-extremity joint replacement, in the BPCI-A model than are hospitals.

Objective: To evaluate the association of treatment by BPCI-A-participating physicians and hospitals with health care spending, quality, and utilization for joint replacement procedures compared to nonparticipants.

Design, setting, and participants: This cohort study used Medicare claims of beneficiaries receiving lower-extremity joint replacement between April 2016 and September 2019 and data on BPCI-A-participating PGPs and hospitals to assess spending, quality, and utilization. Differences-in-differences methods adjusting for patient and market characteristics (aDID) were used with matched comparison groups of nonparticipating physicians and hospitals. Data analysis was performed from January 2023 to January 2025.

Exposures: Lower-extremity joint replacement by a physician in a PGP or hospital that began BPCI-A participation in October 2018.

Main outcomes and measures: Ninety-day total episode spending for joint replacement. Secondary outcomes were postacute care utilization, mortality, hospital readmissions, and joint replacement complications.

Results: The matched cohort included 846 529 Medicare beneficiaries (mean [SD] age, 73.7 [8.3] years; 63.8% female) who obtained a joint replacement in April 2016 to September 2019, of whom 281 189 were treated by 2820 physicians in BPCI-A-participating PGPs, and 69 107 by 174 BPCI-A-participating hospitals. An additional 28 309 beneficiaries were treated by physicians and hospitals both participating in BPCI-A. The remaining 467 924 were treated by 4671 nonparticipating physicians and 432 nonparticipating hospitals. Before BPCI-A participation, total unadjusted baseline episode spending was $26 483 for participating physicians and $29 854 for participating hospitals. Treatments by BPCI-A participating physicians and hospitals were each associated with differentially lower total spending (physician aDID, -$855; 95% CI, -$1074 to -$636; hospital aDID, -$613; 95% CI, -$1039 to -$187). Treatment by a BPCI-A-participating physician or hospital was associated with differentially lower institutional postacute care utilization. Physician participation was associated with a differential increase in outpatient visits 7 days postdischarge (aDID, 2.9 percentage points; 95% CI, 2.0 to 3.8), while hospital participation was not associated with a change in outpatient visits. Differential changes in mortality, readmissions, and complications were not observed for either participant type.

Conclusions and relevance: This cohort study found that participation in BPCI-A for joint replacement was associated with differentially lower total spending for both physicians and hospitals. Given that physicians in PGPs accounted for 73% of all the joint replacement episodes, these findings highlight the importance of facilitating alignment between hospitals and physicians in future bundled-payment models, including those that allow only hospitals.