Baseline Racial and Ethnic Differences in Access to Transplantation in Medicare’s ESRD Treatment Choices Payment Model

Introduction

In July 2019, Medicare—the predominant payer for dialysis in the United States(US)¹—announced the End-Stage Renal Disease Treatment Choices (ETC) model to enhance the value of care for beneficiaries with kidney failure. The model mandated new financial incentives to promote kidney transplantation and home dialysis for dialysis facilities and managing nephrologists in a random sample of 30% of US hospital referral regions. Following a measurement period in 2021, the ETC model awarded payment bonuses or penalties beginning in July of 2022 (Supplementary Figure S1) to ETC-assigned providers based on their use of living donor transplant, transplant waitlisting, and home dialysis compared to control facilities, and ETC-assigned facilities improvements in these outcomes. The model also implemented health equity incentives to reduce socioeconomic disparities in access to home dialysis and transplantation.² Although an assessment of the effectiveness of this model on transplant outcomes is premature, it is critical for clinicians, researchers, and federal policymakers to understand how well randomization worked with respect to important outcomes.

In this analysis, we sought to identify preintervention differences in timely access to referral and evaluation for transplantation, placement on the deceased-donor waiting list, living-donor kidney transplant, as well as racial and ethnic disparities in these outcomes between ETC-assigned and control regions in Georgia, North Carolina, and South Carolina, the only states where referral and evaluation data are completely collected.^3,4 Although referral and evaluation start are not directly incentivized by the ETC model, they represent more proximal steps in the pathway leading to waitlisting and living-donor transplant and may reflect early practice changes implemented in response to ETC model’s transplant incentives. Characterizing these differences prior to an evaluation of the effectiveness of the policy may inform providers about their opportunities to improve, as well as provide important insight to both federal policymakers and to researchers on important methodological considerations that may be needed to evaluate the impact of ETC in later years.⁵ Study methods are described in the Supplementary Material.

Results

In our sample of 747 dialysis facilities in Georgia, North Carolina, and South Carolina treating 60,322 individuals initiating dialysis between 2014 and 2019, 293 facilities (39%) treating 24,313 patients (40%) were assigned to ETC model participation (Supplementary Figure S2 and Supplementary Table S1). Sample construction and patient characteristics are described in Supplementary Methods (Supplementary Figure S3 and Supplementary Table S1). Compared to patients in control facilities, pooled crude rates of 1-year referral, 3-month evaluation start, and 2-year waitlist or living-donor transplant among patients dialyzing in ETC-assigned facilities were lower by 1.8 percentage points (6.4% relative difference; P < 0.001), 5.6 percentage points (34.7% relative difference; P < 0.001), and 1.6 percentage points (16.4% relative difference, P < 0.001), respectively (Table 1 and Supplementary Table S2). The significant differences in these rates were larger in magnitude among non-Hispanic Black patients relative to non-Hispanic White patients (Table 1, Figure 1a).

Table 1.

Mean percentage point differences in crude rates of timely completion of transplant pathway steps among patients initiating dialysis in ETC-assigned versus control facilities 2014–2019

Patient race or ethnicity	Transplant pathway outcome
	1-yr referral			3-mo evaluation start			2-yr waitlist or living-donor transplant
	ETC-assigned	Control	Difference (ETC-control)	ETC-assigned	Control	Difference (ETC-control)	ETC-assigned	Control	Difference (ETC-control)
All	25.9%	27.7%	−1.7a	10.6%	16.2%	−5.6a	8.2%	9.8%	−1.6a
Non-Hispanic Black	30.3%	32.4%	−2.1a	11.9%	18.8%	−6.8a	8.3%	10.5%	−2.2a
Non-Hispanic White	20.7%	21.5%	−0.8	8.8%	12.3%	−3.4a	8.1%	8.9%	−0.8

ETC, end-stage renal disease treatment choices model.

Percentage point differences in average crude rates of transplant evaluation referral within 1 year of dialysis initiation, evaluation start within 3 months of referral, and waitlisting or living-donor transplant receipt within 2 years of dialysis initiation among patients with new-onset kidney failure initiating dialysis at ETC-assigned versus control facilities in Georgia, North Carolina, and South Carolina between 2014 and 2019. T-tests were used to assess statistically significant differences in the proportion of patients achieving an outcome between ETC-assigned and control facilities overall and by race or ethnicity. Total rates are calculated among non-Hispanic-Black and non-Hispanic White patients only (n = 57,109). The denominator for 2-year waitlisting or transplant included n = 45,694 eligible patients (dialysis initiation before November 13, 2018), because follow-up ended November 13, 2020 for these outcomes.

^aP ≤ 0.001.

Figure 1.

Crude annual rates and pooled odds of timely access to steps in the kidney pathway among patients with new-onset kidney failure in Georgia, North Carolina, and South Carolina by patient race or ethnicity, 2014–2019. (a) present the crude annual rates of referral for transplant evaluation within 1 year of dialysis initiation (a.i), evaluation start within 3 months of referral (a.ii), and waitlisting or living-donor transplant receipt within 2 years of dialysis initiation (a.iii) among patients initiating treatment in Georgia, North Carolina, and South Carolina between 2014 and 2019. Crude annual rates were calculated as the proportion of patients with new-onset end-stage kidney disease during that year who had a given outcome within the specified time of incidence. Pooled odds were estimated among all patients with incident end-stage kidney disease from 2014 to 2019 initiating dialysis at a facility in Georgia, North Carolina, or South Carolina achieving outcome events within specified time periods since incidence. The denominator for 2-year waitlisting or transplant included n = 48,218 eligible patients initiating dialysis on or before November 13, 2018, because follow-up ended November 13, 2020 for these outcomes. (b) present the pooled (2014–2019) crude and adjusted odds of referral for transplant evaluation within 1 year of dialysis initiation (b.iv), evaluation start within 3 months of referral (b.v), and waitlisting or living-donor transplant receipt within 2 years of dialysis initiation (b.vi) among patients initiating treatment in Georgia, North Carolina, and South Carolina. In each figure, the reference group is non-Hispanic Black patients in control regions. Adjusted odds calculated as the odds of each outcome among all patients controlling for patients’ sex, age at kidney failure onset, and the median household income of their county of residence. Standard errors were clustered at the dialysis facility level.

Pooled differences in the adjusted odds of 1-year referral by patient race or ethnicity were similar in ETC-assigned facilities (0.65 for non-Hispanic White compared to non-Hispanic Black patients, 95% confidence interval [CI]: 0.59–0.71) and control facilities (0.68; 95% CI: 0.63–0.72; p_ETC=Control = 0.41), as were differences in the odds of 3-month evaluation start (0.85 [95% CI: 0.78–0.94] and 0.71 [95% CI 0.66–0.77] for non-Hispanic White compared to non-Hispanic Black patients in ETC-assigned and control facilities, respectively (p_ETC=Control = 0.09). The adjusted odds for 2-year waitlisting or living-donor transplant were 1.29 (95% CI: 1.15–1.45) and 1.14 (95% CI: 1.04–1.26) for non-Hispanic White compared to non-Hispanic Black patients in ETC-assigned and control facilities, respectively (p_ETC=Control = 0.23) (Figure 1b). Sensitivity analyses using Cox proportional hazard regression models yielded similar results (Supplementary Figure S4). Racial and ethnic differences in these outcomes did not change substantially over time and were consistent between ETC-assigned and control facilities (Supplementary Table S4).

In supplemental analyses, differences in outcomes by ETC assignment were similar among Hispanic patients (Supplementary Figure S5, Supplementary Tables S3 and S4): rates of 1-year referral, 3-month evaluation start, and 2-year waitlisting or living donor transplant were lower in ETC-assigned (vs. control) regions. Adjusted odds of 1-year referral were significantly lower among Hispanic (vs. non-Hispanic Black) patients, regardless of ETC exposure (0.43 [95% CI: 0.34–0.54] and 0.68 [95% CI: 0.56–0.81] for Hispanic patients in ETC-assigned and control facilities compared to non-Hispanic Black patients in control facilities, respectively); however, adjusted odds of 3-month evaluation start and 2-year waitlisting or transplant were not significantly different from those of non-Hispanic Black patients.

Discussion

As more data become available on transplant-related outcomes, policymakers, clinicians, and researchers will evaluate the impact of the ETC payment model on access to kidney transplantation. To identify baseline differences in early steps in the transplant pathway among ETC-assigned and control dialysis facilities, we examined the preintervention rates of 1-year referral, 3-month evaluation start, and 2-year waitlist or living-donor transplant in the southeastern US, and found that ETC-assigned facilities had lower rates of all outcomes during the preintervention period. Moreover, differences in these rates by racial and ethnic group were consistent between ETC-assigned and control regions for all outcomes.

Notably, the rates of 1-year referral and 3-month evaluation start we report are consistently higher among non-Hispanic Black than non-Hispanic White patients on dialysis in all facilities. Although these differences may be striking in light of well-established racial disparities in access to kidney transplant, they are consistent with findings from previous studies in southeastern US.³ The higher referral rate among non-Hispanic Black versus non-Hispanic White dialysis patients may result from the lower preemptive referral rates among non-Hispanic Black patients prior to dialysis initiation⁶ and regional efforts to improve racial or ethnic disparities in kidney transplantation in this region over more than 15 years.^4,7,8

We further show that the rates of timely completion of all steps, including the upstream steps of referral and evaluation start, in addition to waitlisting and transplantation were lower among patients treated at ETC-assigned dialysis facilities (relative to control facilities), regardless of patient race or ethnicity; and this is consistent with previously reported aggregate differences in preintervention prevalent rates of waitlisting and living donor transplantation in ETC-assigned and control facilities, nationally.⁵ Lower rates of referral and evaluation among ETC-assigned and control facilities may contribute to these national differences in waitlisting and living donor transplant, and merit further research. Evaluators should consider that these baseline differences may suggest that there is more opportunity for improvement among ETC-assigned facilities relative to control facilities when evaluating the ETC’s impact.

Our study has several limitations. Our data include incident patients in Georgia, North Carolina, and South Carolina; thus, our findings may not generalize to the prevalent patient population or patients in other regions. Hispanic, non-Hispanic Asian, and Native race or ethnicity were not included in our main analyses, because the risk of drawing spurious conclusions is greater among groups with small sample sizes. Broader collection of referral and evaluation data (e.g., at the national level) is necessary to enable research on these important outcomes among other minoritized populations and other US regions. Finally, our analyses were limited to evaluations of disparities at the dialysis facility level; similar analyses at the managing clinician level are merited, because the ETC model also randomized non-dialysis nephrology providers to similar financial incentives.

Our findings suggest that policy-makers and evaluators can attribute any changes in racial or ethnic disparities in timely access to referral, evaluation start, and waitlisting or living-donor transplant rates observed after implementing the ETC program to the program's incentives, though it will be important to address the systematic differences between ETC-assigned and control facilities we found. Although referral and evaluation start are not directly incentivized by the ETC model, it is likely that dialysis facilities will substantially increase referral rates and promote evaluation start in hopes of improving their waitlisting and living-donor transplant rates to garner bonus payments. Considering that our findings show consistency over time in these outcomes, evaluators should consider including referral and evaluation as key proximal outcomes in assessing dialysis facilities’ responses to the ETC’s financial incentives.

Disclosure

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Supplementary Material

Supplemental References.

Supplementary Methods.

Figure S1. End-Stage Renal Disease Treatment Choices model benchmark, performance, and payment period timeline.

Figure S2. ZIP codes randomly assigned to End-Stage Renal Disease Treatment Choices model participation in ESRD Network 6 (Georgia, North Carolina, and South Carolina).

Figure S3. Inclusion/exclusion criteria of the incident study population of patients with kidney failure, 2014–2019, with follow-up through 2020.

Figure S4. Crude and adjusted hazard ratios for timely access to steps in the kidney pathway among patients with new-onset kidney failure in Georgia, North Carolina, and South Carolina by patient race or ethnicity, 2014–2019.

Figure S5. Crude annual rates and pooled odds of timely access to steps in the kidney pathway among patients with new-onset kidney failure in Georgia, North Carolina, and South Carolina by patient race or ethnicity, 2014–2019.

Table S1. Characteristics of patients with new-onset end-stage kidney disease in Georgia, North Carolina, and South Carolina overall and by End-Stage Renal Disease Treatment Choices assignment, 2014–2019.

Table S2. Timely completion of steps in the transplant pathway among patients with new-onset ESKD in Georgia, North Carolina, and South Carolina overall and by End-Stage Renal Disease Treatment Choices assignment, 2014–2019.

Table S3. Estimates of time trends in timely access to early steps in the kidney transplant pathway by patient race or ethnicity and End-Stage Renal Disease Treatment Choices assignment, 2014–2019.

Table S4. Mean percentage point differences in crude rates of timely completion of transplant pathway steps among patients initiating dialysis in End-Stage Renal Disease Treatment Choices model-assigned versus control facilities, 2014–2019.