Under bundled payment programs, such as Medicare’s Comprehensive Care for Joint Replacement (CJR) model, hospitals bear financial risk for posthospitalization care for beneficiaries in traditional fee-for-service Medicare. It is unknown whether participating hospitals change care patterns only for patients subject to the payment bundle or if changes spillover onto care for other patients undergoing joint replacement. Spillovers to Medicare Advantage (MA) patients would indicate that clinicians have a consistent approach to discharge planning regardless of payor1,2 and would suggest Medicare’s payment reforms have had a broader societal effect.3
The CJR model was implemented in a random sample of communities and only includes patients with traditional Medicare who received a lower extremity joint replacement (LEJR). In prior evaluations of CJR and other LEJR bundled payment programs, spending reductions have primarily been driven by fewer discharges to institutional post–acute care settings.4,5 We evaluated whether the reductions in institutional posthospitalization care observed in the first year of the CJR program among traditional Medicare patients were also observed among MA patients who underwent LEJR.
Methods
Using 2013 to 2016 Medicare MedPAR data, which include hospitalizations for traditional Medicare and MA-insured patients at hospitals that receive disproportionate share subsidies,6 we identified all LEJRs (diagnosis related groups 469 and 470) in acute care hospitals located in the 75 metropolitan statistical areas (MSAs) initially randomized to CJR (“treatment” hospitals) and the 121 MSAs randomized to usual payment (“control” hospitals). The study was approved by the Harvard Medical School Committee on Human Studies, which granted a waiver of informed consent because the study analyzed deidentified secondary data. We excluded patients not enrolled in Medicare parts A and B during the month of their admission, patients with end-stage renal disease, and patients who died before discharge. We limited hospitals to those that received disproportionate share subsidies in 2013 and had at least 10 MA and 10 traditional Medicare LEJR discharges during the study period. These criteria eliminated 360 of 1674 hospitals (22%) that provided 96 109 of 1 556 823 LEJR discharges (6%) in these MSAs. We identified comorbidities defined by Medicare’s readmission program for LEJR using diagnoses on each hospitalization record. Our primary outcome was discharge to institutional post–acute care (including skilled nursing facilities, inpatient rehabilitation facilities, long-term care hospitals, or other institutional settings, such as hospice) vs home.
The CJR program began in the second quarter of 2016. We defined the preintervention period as 2013 to 2015 and the post-intervention period as the last 3 quarters of 2016. We conducted a difference-in-differences analysis using a linear regression with hospital and quarter fixed effects to measure the differential change in institutional post–acute care use in the traditional Medicare and MA samples. We did not observe substantive differences in the preintervention trends between intervention and control hospitals in either population. All models were adjusted for patient characteristics and included sample weights.4 Standard errors were clustered within MSAs and P values less than .05 were considered significant. All analyses were conducted using Stata, version 15 (StataCorp).
Results
Before CJR program implementation, MA patients received less institutional post–acute care, including inpatient rehabilitation, compared with patients in traditional Medicare (Table 1).6 In hospitals affected by the CJR program, the use of institutional post–acute care differentially decreased by 2.1% (95% CI, −3.9% to −0.2%; P = .03) among MA patients and 2.3% (95% CI, −4.0% to −0.7%; P < .01) among traditional Medicare patients (Table 2).
Table 1.
Baseline Characteristics of Medicare Advantage and Traditional Medicare LEJR Discharges Before the Implementation of Mandatory Bundled Payment Program, 2013 to 2015a
| Characteristic | No. (%) | |||
|---|---|---|---|---|
| Medicare Advantage | Traditional Medicare | |||
| Bundled Payment Hospitals | Control Hospitals | Bundled Payment Hospitals | Control Hospitals | |
| MSAs, No. | 74 | 119 | 74 | 119 |
| Hospitals, No. | 597 | 717 | 597 | 717 |
| Discharges, No. | 151 592 | 179 490 | 306 754 | 423 616 |
| Patient characteristics | ||||
| Age, mean, y | 73.52 | 73.33 | 73.82 | 73.63 |
| Male | 52 602 (34.7) | 63 414 (35.3) | 109 266 (35.6) | 152 078 (35.9) |
| White | 130 687 (86.2) | 156 228 (87.0) | 273 747 (89.2) | 382 991 (90.4) |
| Metro residence | 140 783 (92.9) | 157 897 (88.0) | 262 336 (85.5) | 350 500 (82.7) |
| Entitlement disability | 28 575 (18.9) | 34 677 (19.3) | 51 228 (16.7) | 71 125 (16.8) |
| Dual enrollment in Medicaid | 21 268 (14.0) | 23 082 (12.9) | 39 909 (13.0) | 48 758 (11.5) |
| Total condition categories | 1.89 | 1.88 | 1.86 | 1.86 |
| Procedure characteristics | ||||
| LEJR with complications | 6928 (4.6) | 7718 (4.3) | 15 982 (5.2) | 19 995 (4.7) |
| Fracture | 20 814 (13.7) | 23 495 (13.1) | 47 087 (15.4) | 60 026 (14.2) |
| Total knee | 87 726 (57.9) | 104 732 (58.4) | 169 727 (55.3) | 242 308 (57.2) |
| Total hip | 45 341 (29.9) | 54 403 (30.3) | 94 818 (30.9) | 128 906 (30.4) |
| Partial hip/knee | 17 842 (11.8) | 19 726 (11.0) | 40 338 (13.2) | 50 495 (11.9) |
| Discharge status | ||||
| Any institutional post–acute care | 66 458 (43.8) | 70 396 (39.2) | 141 199 (46.0) | 189 017 (44.6) |
| Skilled nursing facility | 59 045 (39.0) | 62 714 (35.0) | 111 168 (36.2) | 149 070 (35.2) |
| Inpatient rehabilitation | 6033 (4.0) | 5833 (3.3) | 25 767 (8.4) | 34 821 (8.2) |
| Long-term care hospital | 106 (0.1) | 215 (0.1) | 368 (0.1) | 890 (0.2) |
| Other | 1289 (0.9) | 1615 (0.9) | 3865 (1.3) | 4236 (1.0) |
Abbreviations: LEJR, lower extremity joint replacement; MSA, metropolitan statistical area.
Counts of metropolitan statistical areas, hospitals, and discharges are unweighted; characteristic averages used MSA sample weights described in Barnett et al.4
Table 2.
Changes in Institutional Post–Acute Care Use Before and After Implementation of Bundled Payment Program, 2013to 2015 vs the Last 3 Quarters in 2016a
| Characteristic | Medicare Advantage | Traditional Medicare | ||||
|---|---|---|---|---|---|---|
| % | Adjusted Differential Change, % (95% CI) | % | Adjusted Differential Change, % (95% CI) | |||
| Change Among Bundled Payment Hospitals | Change in Control Hospitals | Change Among Bundled Payment Hospitals | Change in Control Hospitals | |||
| Any institutional post–acute care | −9.6 | −7.4 | −2.1 (−4.0 to −0.2) | −10.9 | −8.9 | −2.3 (−3.9 to −0.7) |
| Skilled nursing facility | −7.9 | −6.3 | −1.4 (−3.3 to 0.5) | −7.2 | −6.8 | −0.7 (−2.3 to 0.8) |
| Inpatient rehabilitation | −1.7 | −0.8 | −1.0 (−2.1 to 0.2) | −3.2 | −1.8 | −1.4 (−2.3 to −0.5) |
| Long-term care hospital | −0.1 | −0.1 | 0 (−0.1 to 0.2) | 0 | −0.1 | 0.1 (0 to 0.1) |
| Other | 0 | −0.2 | 0.2 (0 to 0.4) | −0.5 | −0.3 | −0.3 (−0.9 to 0.4) |
The change among bundled payment hospitals and change in control hospitals columns show the changes in unadjusted average institutional post–acute care use (weighted) between the preperiod (2013–2015) and the postperiod (April 2016-December 2017). Differential changes were adjusted by hospital and quarter fixed effects, procedure characteristics (diagnosis-associated group, fracture, and procedure type), patient characteristics, including indicators for age group (<65, 65–69 y, 70–74, 75–79, 80–84, and $85 years), male, race/ethnicity (white, black, Asian, other, and Hispanic), metro residence, entitlement reason (age, disability, and end-stage renal disease), Medicaid enrollment in prior 12 months, and 19 comorbidity indicators (condition categories) used by Medicare’s readmission program for risk adjustment. Metropolitan statistical area sample weights were used and are described in Barnett et al4 and errors were clustered at the metropolitan statistical area level.
Among traditional Medicare patients, there was a differential reduction in discharge to inpatient rehabilitation facilities of −1.4% (95% CI, -−2.3% to −0.5%; P < .01). The decline in discharge to inpatient rehabilitation facilities among MA patients was smaller and not significant.
Discussion
For patients who underwent LEJR, Medicare’s CJR program reduced use of institutional post–acute care among patients affected by the program (traditional Medicare) and those not affected by the program (enrolled in MA plans). Our finding is consistent with prior research in which clinicians’ responses to payment reforms were not limited by payer1,2 and suggests that the societal effect of CJR is broader than the traditional Medicare population. This is also notable, given that in MA plans incentives to lower post–acute spending were already present and the use of institutional post–acute care was lower at baseline, suggesting that capitated health plans may be limited in their ability to curb spending without aligning clinician incentives.
Acknowledgments
Funding/Support: This work was supported by Commonwealth Fund and National Institute on Aging/National Institutes of Health grants K23 AG058806 and P01 AG032952.
Role of the Funder/Sponsor: The funding organizations had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.
Conflict of Interest Disclosures: Dr McWilliams reported grants from the National Institute on Aging and serving as a member of the board of directors for the Institute for Accountable Care. Dr Grabowski reported personal fees from NaviHealth, Precision Health Economics, CareLinx, and Vivacitas. Dr Mehrotra reported leading a contract on physician payment not related to bundled payment for Medicare outside the submitted work. No other disclosures were reported.
Contributor Information
Andrew D. Wilcock, Harvard Medical School, Boston, Massachusetts.
Michael L. Barnett, Harvard T. H. Chan School of Public Health, Boston, Massachusetts.
J. Michael McWilliams, Harvard Medical School, Boston, Massachusetts; Brigham and Women’s Hospital, Boston, Massachusetts.
David C. Grabowski, Harvard Medical School, Boston, Massachusetts.
Ateev Mehrotra, Harvard Medical School, Boston, Massachusetts; Beth Israel Deaconess Medical Center, Boston, Massachusetts.
References
- 1.Newhouse JP, Marquis MS. The norms hypothesis and the demand for medical care. J Hum Resour. 1978;13(suppl):159–182. 10.2307/145251 [DOI] [PubMed] [Google Scholar]
- 2.Landon BE. Tipping the scale—the norms hypothesis and primary care physician behavior. N Engl J Med. 2017;376(9):810–811. 10.1056/NEJMp1510923 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.McWilliams JM, Landon BE, Chernew ME. Changes in health care spending and quality for Medicare beneficiaries associated with a commercial ACO contract. JAMA. 2013;310(8):829–836. 10.1001/jama.2013.276302 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Barnett ML, Wilcock A, McWilliams JM, et al. Two-year evaluation of mandatory bundled payments for joint replacement. N Engl J Med. 2019;380 (3):252–262. 10.1056/NEJMsa1809010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Dummit LA, Kahvecioglu D, Marrufo G, et al. Association between hospital participation in a Medicare bundled payment initiative and payments and quality outcomes for lower extremity joint replacement episodes. JAMA. 2016; 316(12):1267–1278. 10.1001/jama.2016.12717 [DOI] [PubMed] [Google Scholar]
- 6.Huckfeldt PJ, Escarce JJ, Rabideau B, Karaca-Mandic P, Sood N. Less intense postacute care, better outcomes for enrollees in Medicare advantage than those in fee-for-service. Health Aff (Millwood). 2017;36(1):91–100. 10.1377/hlthaff.2016.1027 [DOI] [PMC free article] [PubMed] [Google Scholar]