Participation in a Voluntary Bundled Payment Program by Organizations Providing Care After an Acute Hospitalization

Spending on post–acute care (PAC), or care provided after a stay in an acute care hospital, is the largest driver of variation in total per capita Medicare spending.¹ To address this, Medicare has targeted PAC spending in payment reforms such as Model 3 of Medicare’s Bundled Payments for Care Improvement (BPCI) initiative, a voluntary bundled payment program. Model 3 participants receive a target price for 30-, 60-, or 90-day PAC episodes initiated in the 30 days after hospital discharge. Model 3 is risk-bearing: participating organizations spending less than the target price receive a portion of the savings; those spending more must pay Medicare some of the difference.² Model 3 is unique because it targets PAC organizations as the lead organizations that bear the risk, rather than acute care hospitals. Eligible PAC organizations include skilled nursing facilities (SNFs), inpatient rehabilitation facilities, long-term care hospitals, physician group practices, and home health agencies. We describe risk-bearing participation in Medicare’s BPCI Model 3 and factors associated with persistent risk-bearing participation by the largest group of organizations, SNFs.

We used Medicare quarterly BPCI participation reports from the first quarter of 2014 through the second quarter of 2017 to identify changes in BPCI participation over time. We restricted our sample to risk-bearing participants. Participants had to advance to the risk-bearing phase by the third quarter of 2015 or exit the program.

For SNFs, which make up 89% of organizations that ever participated in Model 3, we used data from Medicare Nursing Home Compare (2017) and from Online Survey, Certification, and Reporting/Certification and Survey Provider Enhanced Reporting (OSCAR/CASPER) surveys³ (2015) to compare the characteristics of SNFs stratified by their BPCI participation status (risk-bearing participation category: never, ever, and persistent). Persistent risk-bearing participation was defined as remaining in the program through the second quarter of 2017. A SNF-level multivariable logistic regression model was created to measure variables associated with persistent participation: quality (stars: 1 [lowest]-5 [highest]), patient age, facility type and size (number of beds), acuity index, and other SNF demographics. This analysis was limited to SNFs because quality and demographic data were unavailable for other PAC organizations.

Analyses were conducted using Stata (StataCorp), version 14.2. A 2-tailed P value less than .05 was considered statistically significant.

BPCI Model 3 risk-bearing participation peaked at 1177 PAC organizations (3.7% of the 31 600 eligible US PAC organizations)⁴ in the third quarter of 2015 (Figure), including 1059 SNFs (6.8% of the 15 513 eligible SNFs) and 118 non-SNFs. Of these PAC organizations, 669 (581 SNFs and 88 non-SNFs) were persistent participants (2.1% of eligible PAC organizations and 56.8% of PAC organizations that ever participated in Model 3).⁵

Compared with SNFs with a quality rating of 1 star, SNFs with 2 to 3 stars (odds ratio [OR], 1.58 [95% CI, 1.13-2.20]) or 4 to 5 stars (OR, 2.38 [95% CI, 1.72-3.31]) were more likely to be persistent participants (Table). For-profit SNFs (OR, 1.38 [95% CI, 1.08-1.77]) and SNFs that were part of a multifacility organization (OR, 5.11 [95% CI, 3.88-6.72]) were more likely to be persistent participants. Hospital-based SNFs (OR, 0.26 [95% CI, 0.08-0.86]) were less likely to be persistent participants. Organizations located in the south were less likely to be persistent participants (OR, 0.42 [95% CI, 0.31-0.57]; P < .001).

Only 3.7% of PAC organizations ever participated in the risk-bearing phase of BPCI Model 3, and 43.2% of those participants discontinued participation by the second quarter of 2017. Higher-quality, for-profit, and multifacility organizations were more likely to be persistent BPCI participants. Whether such organizations have an advantage in controlling costs should be assessed.

This study has limitations. The analyses cannot assess causality. The data did not include information on which BPCI participants met payment targets, the size of incentives, or baseline costs. Lack of quality and demographic data on non-SNF organizations led to their exclusion from the multivariable model.

Although policy makers are actively considering new PAC payment reform models,⁶ results suggest that it may be difficult to achieve wide participation with voluntary PAC-initiated bundles. To have a broader effect on PAC, policy makers may wish to explore programs that are either mandatory for PAC organizations or hold acute care hospitals responsible for PAC spending.

Accepted for Publication: June 4, 2018.

Corresponding Author: A. Jay Holmgren, MHI, Harvard Business School, Wyss House, Soldiers Field Road, Boston, MA 02163 (aholmgren@hbs.edu).

Published Online: June 25, 2018. doi:10.1001/jama.2018.8666

Author Contributions: Mr Holmgren 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.

Concept and design: All authors.

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

Drafting of the manuscript: Holmgren, Adler-Milstein.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Holmgren.

Obtained funding: Chen.

Supervision: Adler-Milstein, Chen.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Chen reported grant funding from National Institute on Aging, American Heart Association, Blue Cross Blue Shield of Michigan Foundation, and University of Michigan’s Institute for Healthcare Policy and Innovation (IHPI); personal fees from Robert Wood Johnson Foundation, National Institutes of Health, the Commonwealth Fund, and University of Michigan’s IHPI; and serving as a senior advisor to the deputy assistant secretary for health policy of the US Department of Health and Human Services (HHS). No other disclosures were reported.

Funding/Support: This work was supported by R01HS024698 from the Agency for Healthcare Research and Quality (AHRQ).

Role of the Funder/Sponsor: AHRQ and HHS 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: This article was presented at the AcademyHealth annual research meeting; June 25, 2018; Seattle, Washington.

Additional Contributions: We thank Bailey Green, MPH (University of Michigan), for her assistance on the manuscript. She was compensated for her effort.