Abstract
This study compared monetary penalties and rewards among safety net vs non–safety net hospitals under Medicare’s Comprehensive Care for Joint Replacement (CJR) model, a bundled payment plan for hip and knee replacements intended to incentivize health care quality and savings, in 2016 and 2017.
Medicare’s Comprehensive Care for Joint Replacement (CJR) model is a 5-year bundled payment reform that was mandated for about 700 hospitals in 67 Metropolitan Statistical Areas to improve quality and lower costs of care for fee-for-service beneficiaries undergoing hip and knee replacements. In year 1 (2016), hospitals spending below Medicare’s quality-adjusted target price earned back the difference between the target price and their actual spending (reward).1 Beginning with year 2 (2017), hospitals exceeding the target price repaid Medicare for excess spending (penalty).
The CJR model does not account for worse outcomes and higher costs of socially vulnerable patients.2 We assessed whether safety net hospitals, which serve disproportionately large numbers of these vulnerable patients, were more likely to be penalized and less likely to consistently receive rewards compared with hospitals that serve few vulnerable patients (non–safety net hospitals).
Methods
We linked Medicare’s CJR payment data3 for 2016 and 2017 with the 2016 Medicare Impact File. The first outcome was receipt of penalties in year 2 (hospitals could not be penalized in year 1). The second outcome was receipt of rewards in both years. We categorized hospitals into quartiles based on their disproportionate patient percentage (DPP), which determines a hospital’s eligibility to earn Disproportionate Share Hospital payments.4 Hospitals in the first and fourth DPP quartiles were classified as non–safety net and safety net hospitals, respectively. Hospitals in the middle 2 quartiles served as comparators. Different specifications of the key predictor (excluding the middle 2 quartiles, grouping quartiles 1-3 as non–safety net hospitals, and classifying hospitals as above or below the DPP median) to test the influence of the middle 2 quartiles provided consistent findings.
We used χ2 and Kruskal-Wallis tests to compare hospital characteristics and outcomes across the DPP quartiles. We estimated separate multivariable hierarchical logistic regression models to determine whether safety net status was independently associated with CJR penalties and rewards. These models controlled for hospital-level characteristics and for Metropolitan Statistical Area–level random effects.
University of Rochester’s Research Subject Review Board approved this study and waived the consent requirement. Data were analyzed using Stata version 14.2 (StataCorp). A 2-sided P < .05 was considered statistically significant.
Results
Of 721 hospitals in the data set, 684 were included in the main analysis (see Table 1 footnote); 171 were classified as safety net and 171 as non–safety net (Table 1). Safety net hospitals were more likely to be located in the West, medium and large sized, medical school affiliated, and government owned, and had lower median annual joint replacement volume (24 vs 92 episodes per year, respectively) compared with non–safety net hospitals.
Table 1. Characteristics and Key Outcomes for Hospitals Mandated to Participate in the CJR Model in 2016 and 2017a.
| Hospital Characteristics | DPP Quartileb | Total (n=684) | P Valuec | |||
|---|---|---|---|---|---|---|
| 1 (Non–Safety Net) (n=171) | 2 (n=171) | 3 (n=171) | 4 (Safety Net) (n=171) | |||
| DPP, mean (SD) | 9.28 (4.51) | 21.20 (2.37) | 32.01 (3.69) | 62.26 (18.89) | 31.19 (22.04) | <.001 |
| Geographic region, No. (%) | ||||||
| Northeast | 47 (27.49) | 27 (15.79) | 29 (16.96) | 42 (24.56) | 145 (21.20) | <.001 |
| Midwest | 50 (29.24) | 49 (28.65) | 29 (16.96) | 17 (9.94) | 145 (21.20) | |
| South | 37 (21.64) | 60 (35.09) | 68 (39.77) | 44 (25.73) | 209 (30.56) | |
| West | 37 (21.64) | 35 (20.47) | 45 (26.32) | 68 (39.77) | 185 (27.05) | |
| Size (No. of beds), No. (%) | ||||||
| Small (<200) | 114 (66.67) | 87 (50.88) | 66 (38.60) | 53 (30.99) | 320 (46.78) | <.001 |
| Medium (200-399) | 52 (30.41) | 56 (32.75) | 67 (39.18) | 72 (42.11) | 247 (36.11) | |
| Large (≥400) | 5 (2.92) | 28 (16.37) | 38 (22.22) | 46 (26.90) | 117 (17.11) | |
| Medical school affiliation, No. (%) | 55 (32.16) | 70 (40.94) | 89 (52.05) | 123 (71.93) | 337 (49.27) | <.001 |
| Hospital ownership, No. (%) | ||||||
| Private for-profit | 53 (30.99) | 36 (21.05) | 30 (17.54) | 42 (24.56) | 161 (23.54) | <.001 |
| Private not-for-profit | 109 (63.74) | 114 (66.67) | 114 (66.67) | 89 (52.05) | 426 (62.28) | |
| Government | 9 (5.26) | 21 (12.28) | 27 (15.79) | 40 (23.39) | 97 (14.18) | |
| Transfer-adjusted case-mix index, mean (SD)d | 1.74 (0.40) | 1.64 (0.26) | 1.63 (0.24) | 1.64 (0.28) | 1.66 (0.30) | .15 |
| Medicare days as percentage of total inpatient days, mean (SD) | 37.71 (15.07) | 36.76 (10.35) | 34.06 (9.24) | 25.90 (11.69) | 33.59 (12.64) | <.001 |
| Joint replacement episodes per year, median (IQR) | 92 (30-171) | 81 (33-159) | 57 (25-169) | 24 (9-67) | 57 (20-143) | <.001 |
| Outcomese | ||||||
| Penalties in year 2, No. (%) | 24 (14.04) | 26 (15.20) | 45 (26.32) | 64 (37.43) | 159 (23.25) | <.001 |
| Total penalties in year 2, median (IQR), $ | 21 963 (4649-54 190) | 38 693 (9094-65 820) | 19 913 (14 403-33 674) | 22 254 (8466-39 847) | 22 144 (8480-49 069) | .57 |
| Penalty per episode in year 2, median (IQR), $ | 911 (347-1364) | 1255 (397-1706) | 1122 (458-1658) | 1590 (1133-2035) | 1264 (592-1777) | .002 |
| Rewards in both years, No. (%) | 101 (59.41) | 75 (43.86) | 72 (42.60) | 52 (30.95) | 300 (44.25) | <.001 |
| Total rewards over 2 years, median (IQR), $ | 298 115 (145 748-525 856) | 248 822 (115 771-586 330) | 280 968 (96 417-532 097) | 141 202 (46 238-285 639) | 250 661 (94 779-484 333) | .001 |
| Reward per episode over 2 years, median (IQR), $ | 1167 (1054-1369) | 1180 (1003-1361) | 1221 (996-1448) | 1202 (993-1701) | 1175 (1018-1469) | .52 |
Abbreviations: CJR, Comprehensive Care for Joint Replacement; DPP, disproportionate patient percentage; IQR, interquartile range.
Based on Medicare’s 2016 and 2017 payment data for the CJR model and the 2016 Impact File. The original data set downloaded from the CJR website in January 2019 included 721 hospitals.3 The main analytic cohort was limited to hospitals with financial (reward/penalty) data in both years (n = 678) or hospitals that were missing financial data in year 1 but were penalized in year 2 (n = 6). One hospital that was penalized in year 2 but for which DPP data were missing was excluded. Also excluded from the main analytic cohort were nonpenalized hospitals that were missing financial data in either year 1 (n = 25) or year 2 (n = 11).
DPPs are used to represent hospital safety net status.
P values for χ2 (for categorical variables) and Kruskal-Wallis (for continuous variables) tests comparing the distribution of characteristics and outcomes across DPP quartiles.
The transfer-adjusted case-mix index is the average Diagnosis-Related Group weight for all Medicare discharges in a hospital and is adjusted for discharges to post–acute care facilities. This index is used to represent the clinical complexity of patients and the resources used in care.
Penalty and reward amounts were computed only for hospitals that received penalties or rewards, respectively.
Thirty-seven percent of safety net hospitals were penalized in year 2 compared with 14% of non–safety net hospitals (difference, 23%; 95% CI, 14%-32%; P < .001). In contrast, 31% of safety net hospitals received rewards in both years vs 59% of non–safety net hospitals (difference, −28%; 95% CI, −39% to −18%; P < .001). The median penalty and reward per episode among safety net hospitals were $1590 and $1202, respectively, vs $911 and $1167, respectively, among non–safety net hospitals. Safety net hospitals had higher odds of being penalized in year 2 (adjusted odds ratio, 2.50; 95% CI, 1.25-5.01; P = .009) and lower odds of receiving rewards in both CJR years (adjusted odds ratio, 0.44; 95% CI, 0.24-0.81; P = .008) compared with non–safety net hospitals (Table 2).
Table 2. Key Estimates From Multivariable Hierarchical Logistic Regression Models Examining the Association Between Safety Net Status and Receipt of Rewards and Penalties in the First 2 Years of the CJR Modela.
| Penalties in Year 2 of the CJR (n=678)b | Rewards in Both CJR Years (n=673) | |||||
|---|---|---|---|---|---|---|
| Odds Ratio (95% CI) | P Value | Marginal Estimate, % | Odds Ratio (95% CI) | P Value | Marginal Estimate, % | |
| DPP quartilec | ||||||
| 1 (Non–safety net) | 1 [Reference] | 16.32 | 1 [Reference] | 55.71 | ||
| 2 | 1.17 (0.59-2.32) | .66 | 18.12 | 0.53 (0.31-0.88) | .01 | 42.80 |
| 3 | 2.26 (1.16-4.40) | .02 | 27.22 | 0.51 (0.29-0.87) | .01 | 42.09 |
| 4 (Safety net) | 2.50 (1.25-5.01) | .009 | 28.81 | 0.44 (0.24-0.81) | .008 | 39.36 |
Abbreviations: CJR, Comprehensive Care for Joint Replacement; DPP, disproportionate patient percentage.
See footnote “a” to Table 1. The models also controlled for geographic region (Northeast, Midwest, South, or West), bed number (small [<200 beds], medium [200-399 beds], or large [≥400 beds]), medical school affiliation, ownership (private for-profit, private not-for-profit, or government), transfer-adjusted case-mix index, Medicare days as a percentage of total inpatient days, and average annual case volume (specified as quartiles).
In sensitivity analyses for the outcome of penalties in year 2 of the CJR, analyses controlled for whether a hospital received rewards in year 1. The adjusted odds ratios for this model were 1.04 (95% CI, 0.52-2.11; P=.91) for DPP quartile 2, 1.95 (95% CI, 0.97-3.89; P=.06) for DPP quartile 3, and 2.15 (95% CI, 1.05-4.39; P=.04) for DPP quartile 4.
DPPs are used to represent hospital safety net status.
Discussion
Compared with non–safety net hospitals, safety net hospitals were more likely to be penalized in year 2 and less likely to receive rewards in both CJR years. These inferences are consistent with the analysis of year 1 rewards.5 Possible explanations for the findings6 are that safety net hospitals may provide lower-quality and higher-cost care or that CJR does not account for the increased social vulnerability of patients in safety net hospitals. Although CJR penalties may motivate hospitals to improve, the CJR may also exacerbate disparities by taking away financial resources that safety net hospitals need for improving quality. Medicare should consider changes to the CJR that increase, not decrease, the resources available to safety net hospitals. Study limitations include use of only the DPP to measure hospital safety net status, as well as a lack of hospital spending data.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
References
- 1.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 April 1, 2019.
- 2.Ellimoottil C, Ryan AM, Hou H, Dupree J, Hallstrom B, Miller DC. Medicare’s new bundled payment for joint replacement may penalize hospitals that treat medically complex patients. Health Aff (Millwood). 2016;35(9):1651-1657. doi: 10.1377/hlthaff.2016.0263 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Centers for Medicare & Medicaid Services Comprehensive Care for Joint Replacement model. https://innovation.cms.gov/initiatives/CJR. Data downloaded January 2, 2019. Accessed April 1, 2019.
- 4.Gilman M, Hockenberry JM, Adams EK, Milstein AS, Wilson IB, Becker ER. The financial effect of value-based purchasing and the hospital readmissions reduction program on safety-net hospitals in 2014: a cohort study. Ann Intern Med. 2015;163(6):427-436. doi: 10.7326/M14-2813 [DOI] [PubMed] [Google Scholar]
- 5.Thirukumaran CP, Glance LG, Cai X, Balkissoon R, Mesfin A, Li Y. Performance of safety-net hospitals in year 1 of the Comprehensive Care for Joint Replacement model. Health Aff (Millwood). 2019;38(2):190-196. doi: 10.1377/hlthaff.2018.05264 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Department of Health and Human Services Report to Congress: Social Risk Factors and Performance Under Medicare’s Value-Based Purchasing Programs https://aspe.hhs.gov/pdf-report/report-congress-social-risk-factors-and-performance-under-medicares-value-based-purchasing-programs. Accessed April 1, 2019.