Association of the Comprehensive Care for Joint Replacement Model With Disparities in the Use of Total Hip and Total Knee Replacement

This cohort study evaluates whether the implementation of a large-scale Medicare policy is associated with unequal use of and access to joint replacement care among older adults from various racial/ethnic groups and socioeconomic backgrounds.

Beneficiaries who met dual-eligibility criteria for twelve months in the year were classified as dual-eligible beneficiaries.
Test for parallel trends assumption for triple differences models To assess whether the trends in the use of joint replacements for Medicare beneficiaries from the various race-dual-eligibility groups were parallel in the period before the CJR was implemented (parallel trends assumption for the triple differences models), we estimated the following models (separate for hip and knee replacements). The data for these models was limited to 2013-2015 (pre-CJR period). Results of the parallel trends tests for triple differences models The parallel trends tests were significant for both hip and knee replacements.
Model estimation for triple differences models We estimated the following triple differences models (separate for hip and knee replacements) to assess the differential effect of the CJR. The triple differences approach includes the estimation of three differences: First, the difference between the pre-and post-CJR rates for each race-dual-eligibility group in CJR and non-CJR MSAs (difference 1). Second, the difference in difference 1 between CJR and non-CJR MSAs for each race-dual-eligibility group (difference 2). Third, the difference in difference 2 between each race-dual-eligibility group (except for White non-dual-eligible beneficiaries) and White non-dual-eligible beneficiaries Because of the violation of the parallel trends assumption, we included interactions of with _ and to account for the differential trends in the pre-CJR period. 6

Sensitivity analysis
We checked for the robustness of the main findings using the following strategies. First, we re-estimated the main models using the cohort of 75 MSAs (intention-to-treat analysis) that were originally mandated to participate in the CJR. 2 We used 121 control MSAs that originally met CJR inclusion criteria but were not selected for participation through randomization.
Second, because the Research Triangle Institute (RTI) indicator for race is more sensitive for the identification of Hispanic beneficiaries, we re-estimated the main models using the RTI indicator. This indicator uses a combination of names and geographic locations to improve the identification of Hispanic and Asians/Pacific Islander beneficiaries in comparison to their identification from the Social Security Administration records. 9, 10 Third, we redefined dualeligibility using the Medicare entitlement/buy-in code, which identifies dual-eligible beneficiaries as those with state buy-in for Parts A or B or both of Medicare. 5 Fourth, the intent to implement the CJR was published in July 2015, and the rule was finalized in November 2015. Because changes in joint replacement use patterns may have set-in in advance of the April 2016 start date, we re-estimated the main models by including data from 2016 in the post-CJR period.
Fifth, to refine the identification of elective surgeries, we used Medicare's definition for elective joint replacements which was designed for identifying the cohort for risk-standardized readmission and complication rates. 11,12 This approach uses ICD-9-PCS and ICD-10-PCS codes instead of MS-DRGs (used by the CJR) for identification of joint replacements. Sixth, to examine the effects of race and dual-eligibility separately, we estimated the differential effect models by interacting race (instead of race-dual-eligibility combination) with the CJR/non-CJR MSA and CJR phase indicators. We conducted similar analysis by interacting dual-eligibility with CJR/non-CJR MSA and CJR phase indicators. Finally, to examine whether racial minorities may have been directed to undergo joint replacements in MSAs not participating in the CJR, we estimated multivariable logistic regression models that modeled a binary indicator of whether the patient underwent surgery in his/her residence MSA. This analysis was limited to beneficiaries residing in CJR MSAs. The key independent variables were the CJR phase, the race-dual-eligibility indicator, and the interactions between these variables.  We used the method by the Lewin Group to account for an MSA's probability of selection into the treatment (CJR MSAs) or control (non-CJR MSAs) group. 13 In this approach, the CJR MSAs were assigned a weight of 1 and the non-CJR MSAs were assigned weights to represent the CJR MSAs. These weights for the non-CJR MSAs were obtained by dividing the number of CJR MSAs in each of the 8 strata (constructed by the CMS using quartiles of pre-period episode spending and whether the MSA had above or below median population) by the number of non-CJR MSAs in that stratum. These weights were then used in our regression models using