Abstract
This study uses Medicare data to compare the characteristics of hospitals that did vs did not realize episodes savings under the Centers for Medicare & Medicaid Services’ Comprehensive Care for Joint Replacement program, which bundled payments for hip and knee surgery and paid bonuses to hospitals that exceeded quality and cost benchmarks.
Since April 2016, Medicare has bundled payments for hip and knee surgery at 799 hospitals through the Comprehensive Care for Joint Replacement (CJR) program, which combined payments for hospitalization and postdischarge care in the following 90 days into a single benchmark. CJR incentivizes quality and cost containment by providing retrospective bonus payments that increase as hospitals exceed their quality and cost benchmarks. Initially, Medicare required participation from hospitals in 67 urban markets defined by metropolitan statistical areas (MSAs), favoring MSAs with above-average episode spending and adequate procedural volume. However, Medicare recently indicated that most hospitals in 34 of 67 MSAs are required to continue in the CJR program (mandatory markets) while low-volume hospitals and others in the remaining 33 MSAs could elect to continue or withdraw (voluntary markets). Given this policy shift and its implications for the national debate about mandatory bundled payment, understanding how different hospitals fared in the first year of CJR is salient for policy makers, clinicians, and health care organizations.
Methods
The University of Pennsylvania institutional review board approved the study.
We used Medicare data to identify and describe the proportion, distribution, and quality performance of CJR hospitals that achieved episode savings (savings hospitals) in performance year 1 (April 2016-March 2017).
We compared organizational characteristics and safety-net status between savings and other nonsavings CJR hospitals using data from the American Hospital Association. To compare baseline episode volume, quality (risk-standardized readmission and mortality), and spending, we defined the pre-CJR baseline period from April 2015 to March 2016. Volume was assessed using 100% Medicare claims for 2015-2016. For other measures, we constructed joint replacement episodes using a 20% random sample of Medicare claims and CJR’s methodology, combining all health care services paid by Medicare for joint replacement hospitalization and 90 days of postdischarge care. We described the proportion of savings hospitals in mandatory and voluntary markets.
We compared categorical variables using χ2 tests and continuous variables using t tests, Kruskal-Wallis, and Wilcoxon rank sum tests. We compared baseline performance using hierarchical, multivariable, generalized linear models adjusted for patient characteristics (Table) with hospital savings group (savings and nonsavings hospitals) fixed effects and standard errors clustered at the hospital level. Analyses were performed using SAS (SAS Institute), version 9.4. Statistical tests were 2-tailed and significant at an α level of .05.
Table. Characteristics of Savings Hospitals vs Nonsavings Hospitals in the First Year of Comprehensive Care for Joint Replacement (CJR).
| Characteristics | CJR Savings Hospitals, No. (%) (n = 382) |
CJR Nonsavings Hospitals, No. (%) (n = 417) |
P Value |
|---|---|---|---|
| No. of beds, mean (SD) | 301 (285) | 230 (201) | <.001 |
| Hospital sizea | |||
| Small (<100 beds) | 64 (16.9) | 109 (26.6) | <.001 |
| Medium (100-399 beds) | 223 (59.0) | 240 (58.5) | |
| Large (≥400 beds) | 91 (24.0) | 61 (14.9) | |
| Ownership statusa | |||
| For-profit | 71 (18.8) | 126 (30.7) | <.001 |
| Nonprofit | 263 (69.6) | 219 (53.4) | |
| Government | 44 (11.6) | 65 (15.9) | |
| Teaching statusa | |||
| Major teaching | 49 (13.0) | 30 (7.3) | .004 |
| Minor teaching | 185 (48.9) | 185 (45.1) | |
| Nonteaching | 144 (38.1) | 195 (47.6) | |
| Geographic regiona | |||
| South | 115 (30.4) | 143 (34.9) | .03 |
| Midwest | 83 (22.0) | 79 (19.3) | |
| West | 90 (23.8) | 120 (29.3) | |
| Northeast | 90 (23.8) | 68 (16.6) | |
| Annual Medicare volume, mean (SD) | 6242.4 (5112.9) | 4361.7 (3746.7) | <.001 |
| Annual baseline joint replacement volume, mean (SD) | 216.9 (290.1) | 133.3 (182.2) | <.001 |
| Low-volume hospitalsb | 8 (2.1) | 94 (23.2) | <.001 |
| Safety-net hospitalsc | 90 (21.9) | 153 (37.3) | <.001 |
| Hospitals in mandatory marketd | 233 (61.0) | 257 (61.6) | .80 |
| Hospitals in voluntary marketd | 149 (39.0) | 160 (38.4) | .80 |
| Patient mix | |||
| % Medicare by admissions, mean (SD) | 46.4 (11.9) | 45.8 (12.7) | .46 |
| Elixhauser comorbidity score, mean (SD)e | 4.1 (9.9) | 4.4 (9.9) | .06 |
| Post–acute care integration | 211 (55.8) | 162 (40.0) | <.001 |
| Quality performance categoryf | |||
| Acceptable | 31 (8.1) | ||
| Good | 201 (52.6) | ||
| Excellent | 148 (39.3) | ||
| Savings per episode, mean (SD), $ | 1174 (550) | ||
| Hospital-Specific Baseline Performance Measuresg | |||
| Unadjusted episode cost, mean (95% CI), $ | 22 145 (21 894-22 395) |
23 560 (23 182-23 939) |
<.001 |
| Risk-adjusted episode cost, mean (95% CI), $ | 22 974 (20 922-23 142) |
23 977 (21 836-24 153) |
.001 |
| Risk-standardized mortality rate, % (95% CI) | 1.5 (0.1-2.7) | 1.4 (0.0-2.6) | .63 |
| Risk-standardized readmission rate, % (95% CI) | 8.5 (4.8-10.5) | 9.4 (5.7-11.4) | .10 |
Of the 799 hospitals in CJR, 378 savings hospitals and 410 nonsavings hospitals had complete data in the Annual Hospital Association annual survey.
Low-volume hospitals were defined based on CJR rules: fewer than 20 episodes for Medicare Severity Diagnosis Related Group (MS-DRG) 469/470 in 3 prior years immediately before CJR started.
Safety-net hospitals were defined as being in the top quartile of disproportionate share hospital payment percentage, using the fiscal year 2017 Centers for Medicare & Medicaid Services (CMS) Improving Medicare Post–Acute Care Transformation of 2014 (IMPACT) file.
Markets were defined per CJR program rules using metropolitan statistical areas.
The Elixhauser comorbidity score reflects mortality risk based on the presence of 29 conditions in claims data over the prior 12 months.
Data were obtained from CMS’ published list of hospitals and their results for CJR performance in year 1. Quality data were not available for hospitals that did not receive reconciliation payments under the CJR program.
There were 366 savings hospitals and 292 nonsavings hospitals that contributed claims to the analysis. All models have hospital group fixed effects, adjusting for patient sex, age, race, Medicare-Medicaid dual-eligibility status, Elixhauser comorbidities, MS-DRG of index admission, and time secular trend. Wilcoxon rank sum tests were used to test the difference in unadjusted cost. For episode cost, a multilevel generalized linear model assuming γ distribution was performed. For readmission and mortality rates, multilevel ordinary least square regressions were performed.
Results
Of 799 CJR hospitals, 382 (48%) were savings hospitals. Mean episode savings varied from $13.83 to $3590.97 (Figure), and 351 hospitals (92% of savings hospitals) achieved good or exceptional quality, as defined by Medicare. The proportion of savings hospitals in any given market ranged from 0% to 100%.
Figure. Distribution of Savings per Episode Among Hospitals Earning Savings in the First Year of Medicare’s Comprehensive Care for Joint Replacement.
Compared with nonsavings hospitals, savings hospitals were larger (mean No. of hospital beds, 301 for savings hospitals vs 230 for nonsavings hospitals; P < .001) (Table) and had higher volume (mean annual Medicare joint replacement volume, 216.9 for savings hospitals vs 133.3 for nonsavings hospitals; P < .001). Savings hospitals were more likely to be nonprofit (70% for savings hospitals vs 53% for nonsavings hospitals; P < .001), teaching (62% for savings hospitals vs 52% for nonsavings hospitals; P = .004), and integrated with a post–acute care facility (55.8% for savings hospitals vs 40% for nonsavings hospitals; P < .001) than nonsavings hospitals. A greater proportion of nonsavings hospitals were low volume (2% for savings hospitals vs 23% for nonsavings hospitals; P < .001) and safety-net hospitals (22% for savings hospitals vs 37% for nonsavings hospitals; P < .001). Case-mix severity did not differ by savings groups.
Savings hospitals exhibited similar baseline quality to nonsavings hospitals, but savings hospitals had lower baseline spending ($22 974 in savings hospitals vs $23 977 in nonsavings hospitals; P = .001) (Table). The proportion of savings hospitals did not differ between mandatory and voluntary markets (48% in mandatory markets vs 49% in voluntary markets).
Discussion
In the first year of the CJR program, savings and nonsavings hospitals differed with respect to several characteristics and baseline spending. Limitations include a lack of performance data for nonsavings hospitals and possible bias from a 20% claims sample. Nevertheless, these findings suggest that certain hospital types may be better positioned to succeed in mandatory bundled payment compared with others, particularly as policy changes affect the number and composition of hospitals and markets required to participate.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
References
- 1.Centers for Medicare & Medicaid Services Comprehensive care for joint replacement model. https://innovation.cms.gov/initiatives/cjr. Accessed December 17, 2017. [DOI] [PubMed]
- 2.Centers for Medicare & Medicaid Services Medicare program; cancellation of advancing care coordination through episode payment and cardiac rehabilitation incentive payment models; changes to comprehensive care for joint replacement payment model (CMS-5524-P). Fed Regist. 2017;82(230):57066-57104. [PubMed] [Google Scholar]
- 3.Centers for Medicare & Medicaid Services Overview of CJR quality measures, composite quality score, and pay-for-performance methodology. https://innovation.cms.gov/Files/x/cjr-qualsup.pdf. Accessed January 5, 2018.
- 4.Werner RM, Goldman LE, Dudley RA. Comparison of change in quality of care between safety-net and non-safety-net hospitals. JAMA. 2008;299(18):2180-2187. [DOI] [PubMed] [Google Scholar]
- 5.Annual Hospital Association AHA annual survey database. https://www.ahadataviewer.com/additional-data-products/aha-survey/. Accessed January 5, 2018.
- 6.Navathe AS, Troxel AB, Liao JM, et al. Cost of joint replacement using bundled payment models. JAMA Intern Med. 2017;177(2):214-222. [DOI] [PubMed] [Google Scholar]
