Association of Medicare Mandatory Bundled Payment Reform With Joint Replacement Surgery Use for Beneficiaries With Alzheimer Disease and Related Dementias

Key Points

Question  Is the Comprehensive Care for Joint Replacement (CJR) model associated with changes in hip and knee replacement use for Medicare beneficiaries with Alzheimer disease and related dementias (ADRD)?

Findings  In this cohort study of 24 598 729 beneficiary-year observations among 9 624 461 unique beneficiaries, CJR was statistically significantly associated with a decrease in hip replacement use for beneficiaries with and without ADRD; however, the gap in use between these groups did not change with CJR implementation. The CJR model was not associated with changes in knee replacement use.

Meaning  This study found that the CJR model was not associated with a disproportionate reduction in joint replacement use for Medicare beneficiaries with ADRD.

Importance  Medicare beneficiaries with Alzheimer disease and related dementias (ADRD) are a particularly vulnerable group in whom arthritis is a frequently occurring comorbidity. Medicare’s mandatory bundled payment reform—the Comprehensive Care for Joint Replacement (CJR) model—was intended to improve quality and reduce spending in beneficiaries undergoing joint replacement surgical procedures for arthritis. In the absence of adjustment for clinical risk, hospitals may avoid performing elective joint replacements for beneficiaries with ADRD.

Objective  To evaluate the association of the CJR model with utilization of joint replacements for Medicare beneficiaries with ADRD.

Design, Setting, and Participants  This cohort study used national Medicare data from 2013 to 2017 and multivariable linear probability models and a triple differences estimation approach. Medicare beneficiaries with a diagnosis of arthritis were identified from 67 metropolitan statistical areas (MSAs) mandated to participate in CJR and 104 control MSAs. Data were analyzed from July 2020 to July 2021.

Exposures  Implementation of the CJR model in 2016.

Main Outcomes and Measures  Outcomes were separate binary indicators for whether or not a beneficiary underwent hip or knee replacement. Key independent variables were the MSA group, before-CJR and after-CJR phase, ADRD diagnosis, and their interactions. The linear probability models controlled for beneficiary characteristics, MSA fixed effects, and time trends.

Results  The study included 24 598 729 beneficiary-year observations for 9 624 461 unique beneficiaries, of which 250 168 beneficiaries underwent hip and 474 751 underwent knee replacements. The mean (SD) age of the 2013 cohort was 77.1 (7.9) years, 3 110 922 (66.4%) were women, 3 928 432 (83.8%) were non-Hispanic White, 792 707 (16.9%) were dually eligible for Medicaid, and 885 432 (18.9%) had an ADRD diagnosis. Before CJR implementation, joint replacement rates were lower among beneficiaries with ADRD (hip replacements: 0.38% vs 1.17% for beneficiaries with and without ADRD, respectively; P < .001; knee replacements: 0.70% vs 2.25%; P < .001). After controlling for relevant covariates, CJR was associated with a 0.07-percentage-point decline in hip replacements for beneficiaries with ADRD (95% CI, −0.13 to −0.001; P = .046) and a 0.07-percentage-point decline for beneficiaries without ADRD (95% CI, −0.12 to −0.02; P = .01) residing in CJR MSAs compared with beneficiaries in control MSAs. However, this change in hip replacement rates for beneficiaries with ADRD was not statistically significantly different from the change for beneficiaries without ADRD (percentage point difference: 0.01; 95% CI, −0.08 to 0.09; P = .88). No statistically significant changes in knee replacement rates were noted for beneficiaries with ADRD compared with those without ADRD with CJR implementation (percentage point difference: −0.03, 95% CI, −0.09 to 0.02; P = .27).

Conclusions and Relevance  In this cohort study of Medicare beneficiaries with arthritis, the CJR model was not associated with a decline in joint replacement utilization among beneficiaries with ADRD compared with beneficiaries without ADRD in the first 2 years of the program, thereby alleviating patient selection concerns.