Value-Based Kidney Care

Value-based kidney care continues to accelerate. With over a decade behind us since the passage of the Affordable Care Act, the Center for Medicare & Medicaid Innovation (CMMI) has implemented 54 value-based payment and delivery system models that incentivize high-quality, cost-efficient care (1). The Comprehensive End-Stage Renal Disease Care (CEC) model was CMMI’s sole alternative payment model focused on kidney disease, including only patients with kidney failure on dialysis. In the CEC model, participants form End-Stage Renal Disease Seamless Care Organizations (ESCOs) consisting of dialysis facilities, nephrologists, and other providers, and they assume one-sided or two-sided financial risk for a population of patients on dialysis.

In year 4 of the CEC model, there were 37 ESCOs encompassing 1172 dialysis facilities and 63,640 patients (2). ESCOs were preferentially located in more affluent markets with a greater percentage of Black patients compared with non-ESCO markets (3). The Lewin Group estimated the effect of the CEC model by comparing ESCO facilities with propensity-matched non-ESCO facilities, finding that ESCOs were associated with a 3% decrease in the number of hospitalizations and a 5% decrease in catheter use (2). ESCOs were associated with a $960 yearly decrease per beneficiary in Medicare Part A and B spending, which was predominantly driven by Wave 1 ESCOs and reductions in inpatient hospitalizations and readmissions (2). However, there were no net savings to Medicare after accounting for shared savings payments back to ESCOs.

Although the CEC model did not fully succeed in CMMI’s overall goal of reducing net health care costs, the effect of ESCOs on quality of care and hospitalizations provides important insights—especially as we prepare for Kidney Care Choices, CMMI’s new value-based kidney care model starting in January 2022, which includes patients with CKD stages 4 and 5. Studying the heterogeneity of performance under the CEC model could help us understand which ESCOs, facilities, nephrologists, and patients are most likely to succeed and, perhaps more importantly, why.

There are several hypotheses why certain ESCOs may have better quality and cost performance than others. Vascular surgeons participating in ESCOs would be financially incentivized to place and intervene upon vascular accesses earlier, with the goal of avoiding costly hospitalizations for catheter-associated bloodstream infections and arteriovenous fistula/graft malfunction. Primary care providers participating in ESCOs may have contributed to better coordination of care, resulting in fewer hospitalizations, and thus, greater cost savings. Health system affiliation of nephrologists may be associated with cost savings by facilitating care pathways and a broader alignment of systems and resources to avoid emergency room visits and short-stay hospitalizations. Socially disadvantaged patients or those living in neighborhoods lacking health care access may disproportionately benefit from more intensive care management available in ESCOs.

In this issue of CJASN, Drewry et al. (4) provide a unique and insightful analysis that examines ESCO characteristics and their association with financial performance and mortality. The authors constructed a novel dataset by linking three data sources: (1) publicly available ESCO-level performance data on the CMMI website; (2) ESCO-participating physicians listed on ESCO websites linked to physician specialty information in the National Plan and Provider Enumeration System; and (3) the racial/ethnic composition, percentage Medicaid eligible, and median income of counties where ESCO-participating dialysis facilities were located. Outcome measures examined included financial performance as measured by the gross savings or losses percentage, which compares actual Medicare expenditures with expected (“benchmark”) expenditures. For example, if actual health care costs were $95,000/yr per patients in ESCO and the benchmark was $100,000/yr per patient, the gross savings/losses percentage would be +5%. Second, the standardized mortality ratio was examined, which captures observed to expected mortality, risk adjusted for age, race, ethnicity, sex, comorbidities, and other clinical factors (5).

In multivariable adjusted analyses, the authors find that early ESCO participants were 48% more likely to achieve higher than median financial performance than later entrants, confirming prior analyses (2,6). ESCOs with fewer dialysis facilities were associated with better financial performance. County-level characteristics of ESCO-participating dialysis facilities were also associated with outcomes; counties with fewer Black or Hispanic residents were associated with better financial performance, and counties with higher-income residents were more likely to have below-median mortality.

What can be learned from the results of Drewry et al. (4) that can inform our approach to Kidney Care Choices and future models? First, we may expect a similar pattern of early participants in voluntary payment models generating greater cost savings. Importantly, cost savings were already seen among first-wave ESCO participants in the first year of the program (7), which likely reflects highly motivated participants selecting into the program rather than greater years of experience within the program resulting in improved financial performance. This finding may suggest that early participants recognized low-hanging opportunities for cost savings but were perhaps unable to make further headway as the program progressed. As CMMI expands participation in voluntary models and introduces more mandatory models (8), greater support through effective learning collaboratives must be provided, perhaps modeled after the Fistula First Breakthrough Initiative or the recently proposed Kidney Transplant Learning Collaborative. Reducing complexity of payment model design and ensuring the relevance of all quality measures to nephrology care could also help the performance of subsequent participants. Providing greater support for model success or introducing outlier exceptions may dissuade behaviors such as “cherry-picking” or reluctance on the part of participants to accept certain patients.

Second, these results reveal further insights about which types of participants may benefit from greater resources under these models. Interestingly, the authors find that dialysis facilities in counties with a greater percentage of racial/ethnic minorities had worse financial performance; however, ESCO annual reports indicate that Black patients may have had higher cost savings than White patients (Exhibit E-31 on p. 212 in ref. 2). Thus, the effect of neighborhood-level factors versus individual patient social risk factors must be further disentangled to understand these seemingly discordant findings, which underscore the limitation of analyses that rely on neighborhood-level factors alone. The finding of higher mortality in poor counties may reflect poorer quality of care, but also may reflect limitations inherent to the modeling of the standardized mortality ratio resulting from insufficient clinical covariates and/or social determinants of health. Although the authors identify that ESCOs overall had lower than expected mortality (standardized mortality ratio of 0.93), whether this reflected a selection of healthier patients into the model or improvements in quality requires further study.

One approach to narrow disparities in alternative payment model performance could be to universally screen for social needs of individual patients using validated tools, such as the Accountable Health Communities Health-Related Social Needs Screening Tool, and provide increased payments to address social needs, such as housing or food insecurity. Identification and recognition of social needs may allow for the integration of these measures in subsequent models. This strategy and its effect on downstream quality and costs could be tested in Kidney Care Choices but may be politically challenging to implement if increased payments are cost effective but not cost saving.

Third, the analysis of Drewry et al. (4) offers important findings on the effect of provider team composition on quality and cost savings in the ESCO program. Although 28 of 37 ESCOs had affiliated primary care providers, their involvement was not associated with financial performance or mortality in unadjusted results. ESCO annual reports did find that the number of primary care visits increased in patients in ESCOs versus patients not in ESCOs, but whether this contributed to quality improvements, such as reductions in preventable hospitalizations, is unclear (Exhibit ES-3 on p. 18 in ref. 2). Although patients on dialysis are in regular contact with their nephrology care team and the effect of primary care providers in ESCOs has not yet been demonstrated, patients with kidney disease have a high prevalence of diabetes, chronic pain, and depression, which are predominantly treated in primary care settings for other patients (9). Integrating primary care providers within nephrology care teams may improve patient quality of life (10) and might be worth additional consideration in Kidney Care Choices, especially for patients on home dialysis who see nephrologists less frequently. In contrast to ESCOs, which were limited to patients on dialysis, care coordination across primary care providers, general nephrologists, and transplant nephrologists will prove more crucial in Kidney Care Choices as patients transition across different settings. Additionally, it should be noted that an unintended consequence of ESCOs was to disincentivize kidney transplantation because healthier patients would exit the model—this disincentive has not only been dismantled but instead incentivized in Kidney Care Choices with a transplant bonus.

ESCO surgeon participation, seen in seven of 37 ESCOs, was also not associated with financial performance or mortality. It should be noted that additional ESCOs may have had informal partnerships with vascular surgeons not captured in these data if surgeons were not listed on ESCO websites as affiliated physicians. Characterizing the extent of these partnerships using survey studies or qualitative research could capture if vascular access care pathways changed meaningfully. Future analyses could examine the relationship between ESCO provider team composition and other outcomes, such as hospitalizations, readmissions, outpatient dialysis sessions, and catheter use, which were improved in patients in ESCOs compared with patients not in ESCOs (2). Furthermore, examining these outcomes, as well as financial performance and mortality, as continuous variables could expose greater variation and further elucidate their relationship with ESCO characteristics. Lastly, analyses using the National Plan and Provider Enumeration System and other provider datasets (11) could examine additional nephrologist characteristics (e.g., years in practice) and their association with participation and subsequent performance in ESCOs, Kidney Care Choices, and future models.

In summary, the examination of Drewry et al. (4) of ESCO organizational characteristics provides additional insights into the characteristics of programs that succeed in creating shared savings and improved patient outcomes in value-based kidney care. Coupling their analysis with greater research on the dynamic care delivery interventions implemented in ESCOs can inform how we structure and implement Kidney Care Choices, particularly for socially vulnerable populations. Ultimately, these models provide an opportunity to develop and test which care delivery interventions are effective in improving quality and reducing preventable health care costs.

Disclosures

S. Mohan reports consultancy agreements with Angion Biomedica; receiving research funding from Angion Biomedica and the National Institutes of Health (the National Institute of Biomedical Imaging and Bioengineering, the National Institute of Diabetes and Digestive and Kidney Disease, and the National Institute on Minority Health and Health Disparities); and serving as Deputy Editor of Kidney International Reports, as Vice Chair of the United Network for Organ Sharing Data Advisory Committee, as a member of Scientific Registry of Transplant Recipients Visiting Committee, as a member of the American Society of Nephrology Quality Committee, and on the Angion Pharma Scientific Advisory Board. S.L. Tummalapalli received consulting fees from Bayer AG and a subaward from Scanwell Health, unrelated to the submitted work.

Funding

S.L. Tummalapalli is supported by National Institute of Diabetes and Digestive and Kidney Disease grant F32DK122627 and a National Kidney Foundation Young Investigator grant.