Effects of Comprehensive Care for Joint Replacement model on hospitals serving low socioeconomic status patients

Background: Medicare has placed great emphasis value-based payments to improve cost and quality of care. However, value-based payments may unfairly penalize hospitals serving patients with low socioeconomic status (SES). Caring for low SES patients is expensive and makes it challenging to perform well on quality metrics. In line with this concern, this paper assesses whether hospitals serving a high percentage of low SES patients performed worse under the Comprehensive Care for Joint Replacement (CJR) model. Implemented in April 2016, CJR holds hospitals accountable for the cost and quality of care for patients receiving hip or knee replacements (lower extremity joint replacement or LEJR) during care episodes that include hospitalization and 90-day post-discharge care. CJR provides bonus payments if hospitals meet cost and quality thresholds. These cost thresholds do not account for patients’ SES, but SES affects patients’ capacity to recover at home and rehabilitation costs.

Study Design: We used 2014-2016 Medicare claims and compared health service use and quality of care related to LEJR between hospitals located in 67 metropolitan statistical areas (MSAs) affected by CJR and those in 104 comparable MSAs elsewhere, across hospitals’ proportion of dual-eligible patients (as a proxy of low SES patients). CJR was implemented in 67 randomly selected MSAs, enabling us to estimate the causal effect of the CJR model. There were no significant differences in baseline outcomes when comparing CJR and non-CJR MSAs, suggesting that randomization was effective. Therefore, we used only post-CJR data for our main analyses. We conducted hospital-level analyses using multivariate regressions where we controlled for hospital and MSA-level characteristics unrelated to SES.

Results: Among top 10% hospitals in the proportion of dual-eligible patients, 22% met CJR’s cost and quality thresholds and received bonus payments, compared to 26% for bottom 90% hospitals in the proportion of dual-eligible patients. We also found that top 10% hospitals reduced patients’ 90-day emergency department visit rates by 5 percentage points (from 23 percentage points in non-CJR MSAs) under the CJR while the corresponding rates among the bottom 90% hospitals did not change. Patient discharge to home health care increased by 6 percentage points only among the bottom 90% hospitals under CJR. CJR effects did not vary for other outcomes between those top 10% and bottom 90% hospitals because CJR had no or relatively small effects on those outcomes for both top 10% and bottom 90% hospitals.

Conclusions: CJR had heterogeneous effects between top 10% and bottom 90% hospitals in the proportion of dual-eligible LEJR patients. Particularly, top 10% hospitals reduced emergency department visit rates. This finding contrasts with the concern that hospitals serving a high proportion of low SES patients may struggle more to improve cost and quality of care than other hospitals. In addition, similar percentages of top 10% and bottom 90% of hospitals received bonus payments. However, in 2016, CJR hospitals faced less restrictive cost thresholds without any down-side risk. Further evaluation on CJR’s heterogeneous effects is needed as the CJR model becomes fully implemented.