Voluntary Bundled Payment for Coronary Artery Bypass Surgery: Do Effects for Joint Replacement Extend to Other Surgical Conditions

Tuesday, June 25, 2019
Exhibit Hall C (Marriott Wardman Park Hotel)

Presenter: Joshua Rolnick

Co-Authors: Joshua Liao; Xinshuo Ma; Eric Shan; Jingsan Zhu; Erkuan Wang; Qian Huang; Amol Navathe

Background: With the launch of its new voluntary bundled payment model, Medicare has signaled that episode-based payment is a key strategy in its quest for value. The Bundled Payments for Care Improvement (BPCI) program tested the impact of voluntary bundled Medicare payments for medical and surgical conditions. However, most prior work on BPCI has focused on knee and hip replacement surgery. CABG was the first condition for which Medicare tried bundles (starting in 1992) and is the highest-volume surgical BPCI condition after LEJR. CMS has proposed, canceled, and subsequently reconsidered a mandatory bundled payment program for CABG. Yet little is known about whether the BPCI program was associated with changes in spending or utilization for CABG.

Methods: We used Medicare claims data from 2011–2016 to identify index hospitalizations for CABG participants and non-participants. Episodes attributed to BPCI physician groups and non-BPCI hospitals within BPCI markets were excluded. The primary outcome was standardized total episode spending. Second outcomes included spending by provider type, readmission rates, and mortality. Models were adjusted for patient characteristics, time-varying market characteristics (hospital referral region), and hospital and quarterly time fixed effects. Ordinary least squares models were used for utilization and generalized linear models with a log-link function and gamma distribution for spending, with robust standard error.

Results: 49 hospitals enrolled in CABG BPCI between October 2013 and October 2015 (11,108 episodes in baseline period, 5,305 treatment episodes). The comparison group included 844 hospitals (42,595 episodes baseline, 41,002 treatment). Mean follow up for BPCI participants was 23.9 months. There were no baseline differences in patient age, race, or 90-day readmission rate. Elixhauser index at baseline was 11.5 in BPCI vs. 10.9 (p < 0.001). Discharge to post-acute care from index hospitalization at baseline was 33.3% in BPCI group vs. 26.3% control (p < 0.001). Total episode spending at baseline was $48,326 in the BPCI group vs. $47,907 in controls (p < 0.001) and spending for post-acute institutional care was $5,318 BPCI vs. $4,256 controls (p < 0.001). The largest category of baseline post-index spending was physician fees (15%, $6,932 BPCI vs. $6,718 controls, p < 0.001), followed by skilled nursing facility (SNF) (6%, $3,144 BPCI vs. $2,257 controls, p < 0.001).

Difference-in-difference analysis showed BPCI participation was associated with non-significant decreases of 0.9% (95% CI -7.5%-6.0, p-value=0.79) in total spending, 68.5% (95% CI -39.0%-365.1%, p-value=0.31) in spending for post-acute institutional care, and 56.3% (95% CI -85.1%-27.8%, p-value=0.13) in spending for SNF. However, compared to non-BPCI hospitals, BPCI hospitals experienced a significant 26% (95% CI 12.3%-35.2%, p<0.001) decrease in physician fees. For utilization, there was a significant 19.5% (95% CI 8.4%-36.0%, p=0.020) decrease in the likelihood of at least one ER visit. Changes in readmission rate and mortality were not significantly associated with participation.

Conclusion: Although BPCI participation was not associated with overall savings, there was a significant decrease in ER utilization. This study provides the first look at the impact of the BPCI program on surgical conditions outside joint replacement.