Association Between Hospital Voluntary Participation, Mandatory Participation, or Nonparticipation in Bundled Payments and Medicare Episodic Spending for Hip and Knee Replacements

Medicare has used both voluntary and mandatory participation to engage health care organizations in value-based payment models. Compared with mandatory participants, voluntary participants are assumed to achieve greater savings because they self-select into programs due to greater opportunity to reduce spending. However, no empirical data exist comparing savings under mandatory vs voluntary programs.

In 2013, Medicare started the voluntary Bundled Payments for Care Improvement (BPCI) program, which included hip and knee joint replacements. In 2016, under the Comprehensive Care for Joint Replacement (CJR) program, Medicare randomized hospitals in 75 metropolitan statistical areas to receive mandatory hip and knee joint replacement bundled payments, whereas hospitals in 121 metropolitan statistical areas continued receiving fee-for-service payments. This created an opportunity to examine the association between voluntary vs mandatory hospital participation in a bundled payment program for joint replacements and episodic spending changes.^1-3

The University of Pennsylvania institutional review board approved the study with a waiver of consent. Following prior work,^3,4 we used 2011-2017 Medicare claims to evaluate spending among hospitals in metropolitan statistical areas randomized via the CJR program and weighted by strata used in the Medicare randomization protocol. The hospitals randomized to mandatory bundled payments in the 75 metropolitan statistical areas in the CJR program were divided into hospitals (1) that self-selected into bundled payments for joint replacement via the BPCI program before being subject to the CJR program (voluntary hospitals) or (2) that were not participating in the BPCI program for joint replacement prior to CJR program participation (mandatory hospitals). Hospitals were assigned to these groups regardless of subsequent dropout.

Hospitals in the 121 metropolitan statistical areas randomized to continue receiving fee-for-service payments were defined as the comparison group if they did not participate in the BPCI program for joint replacements prior to CJR program participation (hospitals not participating in the bundled payment program) or were excluded if they did. Consistent with prior evidence that joint replacement bundle programs target institutional postacute care,^2,3 the outcome measure selected was episodic spending on institutional care (the sum of index hospitalizations and 90-day postdischarge readmissions, skilled nursing, inpatient rehabilitation, and long-term acute care facility spending). All spending estimates were adjusted for inflation into 2017 dollars.

To account for time-varying entry into the BPCI program and the influence of BPCI program participation on a hospital’s episodic spending during the period preceding CJR program participation, we used event study methods to examine changes in spending before and after starting the bundled payment program (ie, the exposures were time-varying indicators of BPCI or CJR program participation). Risk-adjusted values were estimated for each hospital group during their common periods before and after starting the bundled payment program using generalized linear difference-in-differences models that included quarter-year fixed effects, hospital fixed effects that adjusted for time-invariant hospital characteristics, patient and time-varying market characteristics, and standard errors clustered at the metropolitan statistical area level. There were nondivergent trends in episodic spending across hospital groups during the period before starting the bundled payment program.

Statistical tests were 2-sided and considered significant at α = .05. Analyses were conducted using Stata version 16.0 (StataCorp).

The sample consisted of 1 346 756 Medicare fee-for-service beneficiaries undergoing joint replacement surgery at 92 voluntary hospitals, 752 mandatory hospitals, and 894 hospitals not participating in the bundled payment program (Table 1). Voluntary hospitals were larger than mandatory and nonparticipant hospitals and were more likely to be nonprofit, teaching institutions. Compared with patients at other hospitals, patients at voluntary hospitals were more likely to be non-White individuals in markets with a higher penetration of accountable care organizations.

Risk-adjusted episodic spending decreased among voluntary hospitals from $21 182 before bundled payments to $18 452 after bundled payments; among mandatory hospitals, spending decreased from $18 390 to $15 652; and among hospitals not participating, spending decreased from $17 132 to $14 871 (Table 2). Compared with hospitals not participating, both voluntary hospitals (difference-in-differences estimate, −$469 [95% CI, −$795 to −$142]; P = .005) and mandatory hospitals (difference-in-differences estimate, −$477 [95% CI, −$771 to −$183]; P = .002) exhibited differentially lower risk-adjusted episodic spending. Voluntary and mandatory hospitals did not exhibit differential changes in risk-adjusted episodic spending (difference-in-differences estimate, −$8 [95% CI, −$337 to $322]; P = .96).

Hospitals in bundled payment programs achieved lower episodic spending for hip and knee replacements than hospitals not participating in the programs, but spending changes did not differ between the voluntary and mandatory hospitals. This result does not support the concept that organizations perform better when self-selecting into programs.^5,6

Study limitations include residual confounding and limited generalizability to other payment models or non–joint replacement bundled payment programs. Nonetheless, these findings may inform policy debates about the benefits of mandatory vs voluntary payment models.

Corresponding Author: Amol S. Navathe, MD, PhD, Department of Medical Ethics and Health Policy, University of Pennsylvania, 423 Guardian Dr, 1108 Blockley Hall, Philadelphia, PA 19104 (amol@wharton.upenn.edu).

Accepted for Publication: June 2, 2021.

Published Online: June 16, 2021. doi:10.1001/jama.2021.10046

Author Contributions: Dr Navathe had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Liao and Gupta contributed equally to this work.

Concept and design: Liao, Gupta, Martinez, Navathe.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Liao, Gupta, Navathe.

Critical revision of the manuscript for important intellectual content: Liao, Zhu, Zhao, Martinez, Cousins, Navathe.

Statistical analysis: Gupta, Zhu, Zhao, Martinez, Navathe.

Obtained funding: Liao, Cousins, Navathe.

Administrative, technical, or material support: Martinez, Cousins, Navathe.

Supervision: Navathe.

Conflict of Interest Disclosures: Dr Liao reported receiving personal fees from the Kaiser Permanente Washington Health Research Institute; receiving textbook royalties from Wolters Kluwer; and receiving honoraria from the Journal of Clinical Pathways and the American College of Physicians. Dr Navathe reported receiving grants from the Hawaii Medical Association, the Anthem Public Policy Institute, the Commonwealth Fund, Oscar Health, Cigna Corporation, the Robert Wood Johnson Foundation, the Donaghue Foundation, Pennsylvania Department of Health, Ochsner Health System, United Healthcare, Blue Cross Blue Shield of North Carolina, Blue Shield of California, and Humana; receiving personal fees from Navvis Healthcare, Agathos Inc, Yale New Haven Hospital Center for Outcomes Research and Evaluation, Maine Health Accountable Care Organization, Maine Department of Health and Human Services, Singapore National University Health System, Singapore Ministry of Health, Elsevier Press, Medicare Payment Advisory Commission, Cleveland Clinic, Analysis Group, VBID Health, Embedded Healthcare Equity, and Integrated Services Inc; and receiving personal fees and equity from Nava Health. No other disclosures were reported.

Funding/Support: This work was funded by grant R01MD013859 from the National Institutes of Health and grant R01HS027595-01A1 from the Agency for Healthcare Research and Quality.

Role of the Funder/Sponsor: The funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: Presented virtually at the annual research meeting of AcademyHealth, June 16, 2021.