2018 John Charnley Award: Analysis of US Hip Replacement Bundled Payments: Physician-initiated Episodes Outperform Hospital-initiated Episodes.

BACKGROUND: The Centers for Medicare & Medicaid Services (CMS) launched the Bundled Payment for Care Improvement (BPCI) initiative in 2013 to create incentives to improve outcomes and reduce costs in various clinical settings, including total hip arthroplasty (THA). This study seeks to quantify BPCI initiative outcomes for THA and to determine the optimal party (for example, hospital versus physician group practice [PGP]) to manage the program. QUESTIONS/PURPOSES: (1) Is BPCI associated with lower 90-day payments, readmissions, or mortality for elective THA? (2) Is there a difference in 90-day payments, readmissions, or mortality between episodes initiated by PGPs and episodes initiated by hospitals for elective THA? (3) Is BPCI associated with reduced total Elixhauser comorbidity index or age for elective THA?
METHODS: We performed a retrospective analysis on the CMS Limited Data Set on all Medicare primary elective THAs without a major comorbidity performed in the United States (except Maryland) between January 2013 and March 2016, totaling more than USD 7.1 billion in expenditures. Episodes were grouped into hospital-run BPCI (n = 42,922), PGP-run BPCI (n = 44,662), and THA performed outside of BPCI (n = 284,002). All Medicare Part A payments were calculated over a 90-day period after surgery and adjusted for inflation and regional variation. For each episode, age, sex, race, geographic location, background trend, and Elixhauser comorbidities were determined to control for major confounding variables. Total payments, readmissions, and mortality were compared among the groups with logistic regression.
RESULTS: When controlling for demographics, background trend, geographic variation, and total Elixhauser comorbidities in elective Diagnosis-Related Group 470 THA episodes, BPCI was associated with a 4.44% (95% confidence interval [CI], -4.58% to -4.30%; p < 0.001) payment decrease for all participants (USD 1244 decrease from a baseline of USD 18,802); additionally, odds ratios (ORs) for 90-day mortality and readmissions were unchanged. PGP groups showed a 4.81% decrease in payments (95% CI, -5.01% to -4.61%; p < 0.001) after enrolling in BPCI (USD 1335 decrease from a baseline of USD 17,841). Hospital groups showed a 4.04% decrease in payments (95% CI, -4.24% to 3.84%; p < 0.01) after enrolling in BPCI (USD 1138 decrease from a baseline of USD 19,799). The decrease in payments of PGP-run episodes was greater compared with hospital-run episodes. ORs for 90-day mortality and readmission remained unchanged after BPCI for PGP- and hospital-run BPCI programs. Patient age and mean Elixhauser comorbidity index did not change after BPCI for PGP-run, hospital-run, or overall BPCI episodes.
CONCLUSIONS: Even when controlling for decreasing costs in traditional fee-for-service care, BPCI is associated with payment reduction with no change in adverse events, and this is not because of the selection of younger patients or those with fewer comorbidities. Furthermore, physician group practices were associated with greater payment reduction than hospital programs with no difference in readmission or mortality from baseline for either. Physicians may be a more logical group than hospitals to manage payment reduction in future healthcare reform. LEVEL OF EVIDENCE: Level II, economic and decision analysis.