Value-based payment (VBP) is a powerful and widespread health care transformation movement that has potential to improve health equity. The care delivery and payment flexibility possible under VBP (for example, to address social needs or hire community health workers), and the care coordination emphasized, are difficult at best or impossible at worst under fee-for-service reimbursement. Pairing that flexibility with VBP models’ financial incentives to improve care quality and value can make advancing equity a reality—if there are specific equity-focused design elements and incentives.
Accordingly, the Center for Medicare and Medicaid Innovation (the Innovation Center) redesigned its Global and Professional Direct Contracting Model earlier this year to be an equity-focused VBP model, ACO (ACO) REACH (Realizing Equity, Access, and Community Health). With the creation of ACO REACH, the federal government joined a handful of states and commercial payers taking steps to directly include equity in VBP model design. Additionally, the Centers for Medicare and Medicaid Services (CMS) is seeking comments on potential proposals for rolling equity-focused VBP principles into a variety of Medicare programs, including the Medicare Shared Savings Program.
Without an explicit focus on equity in model design, however, VBP models could perpetuate health inequities by not including historically marginalized populations and their providers or by not encouraging providers to identify and address health disparities. In a recent two-part article in Forefront, we discussed and categorized equity-focused design elements for use in VBP, highlighting examples, evidence, and early lessons from how CMS, states, and commercial payers do this in practice. For CMS to achieve the five priorities from its new Framework for Health Equity, though, equity-focused design is just the first part of the puzzle—it must be paired with equity-focused implementation.
After speaking with VBP and equity experts to identify priority topics, we scanned policy documents and reviewed literature to accomplish three goals in this article. First, following up on our previous design-focused two-part article centered primarily on ACO REACH, we identify additional equity-focused VBP design components that other models may need to consider that were not employed in ACO REACH—specifically, equity-focused patient attribution methods. Second, we highlight key implementation considerations for how CMS, states, and commercial payers can maximize equity in VBP model rollout by better engaging patients, providers, and other implementation partners. To this end, similar to how we synthesized a framework for VBP-specific design considerations in the previous pieces, we present here a brief framework for categorizing the spectrum of approaches we identified for improving equity through VBP model implementation. And while critical to advancing health equity, these new equity-focused models are novel, and we need more evidence on what works, especially in early years of implementation. Accordingly, we end this article with a brief overview of where equity-focused evaluation is needed to assess the success of equity-focused design and implementation.
Identifying Patients To Attribute Into VBP Models
One area of VBP design that influences equity but that we did not explore in our previous work—due to finding no extant examples—is patient attribution. Patient attribution methods are critical for equitable inclusion of populations in VBP, yet to our knowledge, little to no equity-focused considerations exist for attribution. Therefore, below, we explore theoretical mechanisms for payers to explore; our recommendations are then summarized in exhibit 1.
For context, about 75 percent of Medicare beneficiaries self-identify as non-Hispanic White race, yet 87 percent of beneficiaries attributed to the most recent year of the Medicare Shared Savings Program (MSSP, the largest VBP program in the country) were non-Hispanic White race. This indicates lower representation of Hispanic ethnicity, and of non-Hispanic Black, Asian, Native America, and Other race populations (source: authors’ analysis of public data). As CMS notes in a recent analysis of VBP models for implicit bias, some VBP models are associated with treating fewer historically marginalized populations and fewer dually eligible people (a marker of lower socioeconomic status). Changes to attribution could help mitigate these discrepancies.
Most population-based VBP models, such as ACOs, attribute patients based on claims data analyses of where patients receive the plurality of primary care-type services. Episode or bundled payments models also rely on use of a specific intervention (such as a hip or knee replacement), a specific event (such as a hospitalization for AMI/heart attack), or use that demonstrates a specific diagnosis. Because these attribution approaches depend on people having medical encounters, there could be lower attribution of historically marginalized populations who have insufficient access (for example, cost of care, mobility issues, transportation), limited trust in the health care system, or overall limited engagement in health care. For context, non-Hispanic White populations are the least likely racial and ethnic group to say they have no usual source of care—about 24 percent, compared to 26–38 percent of respondents self-reporting their race and ethnicity as Hispanic; non-Hispanic Black; non-Hispanic Asian, Hawaiian, or Pacific Islander, or non-Hispanic American Indian, Alaska Native, or multiple race (source: authors’ analysis of the most recent year of Medical Expenditures Panel Survey public data).
To help mitigate these challenges, policy makers can expand the data acceptable for attribution to more completely identify marginalized populations for equitable inclusion in VBP models. One approach is expanding the settings where health care encounters can turn into claims beyond in-person visits. The MSSP now includes telehealth visits (online digital as well as telephonic evaluation and management services) in the definition of claims used for attribution, which is helpful in fostering attribution equity by including marginalized patients who have trouble accessing traditional physical locations of care. Given the focus on primary care, most ACO programs limit their attribution algorithms to exclude inpatient settings—for example, the MSSP counts primary care services for attribution that are provided in outpatient settings, the patient’s home, federally qualified health centers, rural health clinics, or nursing homes. And CMS is currently proposing to include certain chronic pain management service codes and prolonged nursing facility or home services as counting toward the definition of primary care. While excluding inpatient care from attribution makes sense (that is, to not incentivize costly, potentially unnecessary care in a value-focused accountable model), this policy has one blind spot:
The hospital comprises 22–24 percent of usual source-of-care settings for respondents self-reporting their race and ethnicity as Hispanic; non-Hispanic Black; non-Hispanic Asian, Hawaiian, or Pacific Islander, or non-Hispanic American Indian, Alaska Native, or multiple race. This range is higher than non-Hispanic White populations (source: authors’ analysis of the most recent year of Medical Expenditures Panel Survey public data). While it makes sense to focus on primary care in these settings, it may leave out chronic care management or other types of primary care services provided by hospitalists or internal medicine clinicians in inpatient settings.
Another approach to expanding attribution beyond historical utilization patterns is to leverage patient voluntary alignment, where patients have the choice to align themselves with an ACO. Through this approach, ACOs notify patients of the opportunity to choose to be in an ACO, and the patient can select their health care provider, who is the primary person responsible for coordinating their care; this can override claims-based attribution and result in alignment with an ACO. For people with currently limited health care use, voluntary alignment into an ACO is an opportunity to improve patient engagement, and could help reward early prevention as well as overcoming existing barriers to access. The Innovation Center studied voluntary attribution, finding that it was increased by sending a letter to Medicare beneficiaries, although there are concerns that few people have used voluntary alignment.
To mitigate concerns that people with lower health literacy or lack of trust in the health care system would be less able or likely to take advantage of formal voluntary alignment outreach methods, payers should experiment in two ways. First, we recommend that payers use additional outreach methods beyond letters that are accessible to wider populations, such as email, telephone, and text message methods. Second, we recommend that researchers and payers work together to test better plain language explanations of what VBP is in ways that patients understand and value. These outreach approaches will need to be structured to ensure that recipients, especially people with substantial medical and social needs, feel they have freedom to choose whether or not to participate.
There is ongoing research by CMS and the National Institutes of Health (NIH) to examine implicit bias in model rules, processes, or algorithms—including attribution—that can negatively affect marginalized populations. The Innovation Center recently released its findings on implicit bias within three of its specialty care VBP models. In one of the examples, participants in the Comprehensive Care for Joint Replacement (CJR) model were less medically complex and less likely to be duals and/or Black race compared to joint replacement patients in the same hospitals before the model was in effect. One hypothesis was that providers may be influencing the patient mix of those attributed by selecting patients who are not as likely to need institutional postacute care. Accordingly, the Innovation Center revised the CJR benchmark beginning in 2022 to include risk adjustment for dual-eligibility status. Similar concerns were raised recently that CMS will need to actively monitor the types of investments providers make in response to ACO REACH’s health equity benchmark to see if they invest in improving care delivery or, counterproductively, in aggressively marketing to maximize profit.
This important work by the Innovation Center shows that even medically based risk-adjustment methods for the benchmark can generate implicit biases in patient attribution and patient selection. Current risk-adjustment models are limited in their ability to account for patients’ functional status or socioeconomic characteristics, which are important predictors of health need and spending, so improved risk-adjustment methods can improve equitable attribution through reduced adverse selection. The National Institute on Minority Health and Health Disparities is currently sponsoring new research into how to mitigate biases associated with utilization data-based algorithms to prevent systemic exclusion of marginalized populations. These efforts by the Innovation Center and the NIH should continue. Payers and researchers should generate practical evidence on new data sources and methodological attribution approaches; this should include testing the viability of using new risk adjusters that can better capture socioeconomic factors, the use of artificial intelligence-enabled tools to identify patients at high risk of social marginalization who would benefit from care coordination, and new design approaches in VBP models to improve equitable attribution.
Beyond changing the technical attribution specifications, payers can use incentives and outreach tools to help ensure that VBP models reach people of diverse race, ethnicity, and socioeconomic status, as well as people with greater medical and social needs. CMS has developed additional incentives for VBP participants to reach out to and care for broader populations, including the health equity benchmark in the ACO REACH model, which adds additional monthly payments for caring for more dually eligible beneficiaries and people in lower-resourced geographic areas. CMS is currently considering including a similar method in the MSSP, along with several other health equity incentives, such as bonus points in quality scores to support ACOs serving a high proportion of underserved patients.
Incentives will not be effective to achieve equity unless health care organizations have strategies they can use to reach out to people commonly excluded from VBP or health systems in general. These strategies may include working with providers to identify patients they see who may benefit from inclusion in the VBP program, a tactic that has been used in combination with algorithmic identification in serious illness programs within ACOs. Once identified, health care organizations could reach out to patients for preventive care, which would increase claims data-based attribution methods for population-based models, or encourage voluntary alignment. Another strategy would be for health care organizations to examine where they can form partnerships with safety-net providers, such as community health centers or safety-net health systems, that care for traditionally medically underserved communities. The best approach is to use multiple approaches at once.
Exhibit 1: Summary of VBP design recommendations for equity-focused attribution
Source: Authors’ analysis (from literature and policy document scans, and interviews with equity-focused value-based payment model implementors and experts).
A Framework Of Equity-Focused VBP Implementation Elements
A range of frameworks exist to help assess how health payment or delivery reforms can improve equity, and we reviewed three that specifically incorporate concepts of VBP. We leverage these existing frameworks to outline specific considerations for VBP implementation, as shown in exhibit 2:
Exhibit 2: Synthesized framework of equity-focused VBP implementation elements
Source: Authors’ analysis (from literature and policy document scans, interviews with equity-focused value-based payment model implementors and experts, and concise synthesis of concepts from the Institute for Healthcare Improvement’s Framework for Achieving Health Equity, the Health Care Payment Learning and Action Network’s guidance for advancing health equity through VBP, and the Robert Wood Johnson Foundation’s and National Alliance to impact the Social Determinants of Health’s Raising the Bar framework for promoting equity and excellence in health care).
Engaging Patients And People In VBP Design And Implementation
Two related critical equity-focused VBP implementation considerations are better provider communication to maximize patient engagement and financial incentives for patients that are equitably accessible. Our recommendations in this area are discussed below and summarized in exhibit 3 at the end of this section.
Patient And Person Engagement In VBP Model Design
One important way to make sure that VBP meets the needs of different communities is to engage people from those communities in the design and implementation of VBP models. This helps to ensure that the model is focused on improving areas of health care that are meaningful to patients and people broadly. Key considerations for policy makers weighing community input on outcomes with tradeoffs among what is easy to measure (such as measures easily calculated from health care claims data), those factors that are directly under a clinician’s control (such as adherence to specific guideline-recommended care), and broader aspects of a person’s experience of care and health outcomes.
Some public and private payers have started to include patients on VBP advisory boards (implementation focused), such as the Community Advisory Councils for Oregon’s Coordinated Care Organizations. CMS has highlighted its intention to increase outreach to beneficiaries, patient advocacy groups, and groups with expertise in caring for traditionally underserved populations. We recommend CMS continue to include patients and communities on implementation-focused VBP advisory boards and use focus groups to better understand the beneficiary perspective on VBP model design, outreach, implementation, and evaluation at a program level (outside of the context of a specific ACO’s advisory board).
Communications And Engagement Of People In VBP Implementation
A long-standing challenge is explaining to people what VBP means to them and how it will affect the care they receive. VBP program implementers have historically experienced challenges explaining value-based care even to patients already involved in VBP programs. In terms of equity, communications about VBP can be important for building trust (especially for communities that may historically mistrust the health care system).
Clinicians serve as the most trusted messengers with patients on matters of health care, and VBP implementation should not be any different. Payers can provide clinicians with technical support on outreach and communication strategies specifically for VBP programs: clear, customized messages (for example, tailoring messages to different linguistic groups or explaining how VBP can improve equity); feedback channels (for example, hotline, suggestion box) to receive patient feedback during the VBP implementation process; and user-friendly platforms to communicate with patients (for example, patient portals, interactive texting platforms, or follow-up information using photos with bite-size information) that simultaneously serve as a resource portal on the nuances of VBP model benefits for patients. To help explain the benefits of VBP beyond clinician voices, policy makers should also consider as outreach and education partners the community voices that have communications skills, experience working with the health system, and who people trust the most, such as faith-based organizations and local advocacy organizations with experience canvassing or phone-banking.
Cultural competency and cultural humility are essential in engaging people in VBP design and implementation. CMS has highlighted several ways it currently promotes culturally competent care, which includes providing materials in multiple languages and identifying communities’ cultural and social needs. There is still limited evidence on how best to train providers in cultural competency. By comparison, the practice of cultural humility involves continuously seeking to understand patients’ complex identities and their experiences. Payers can encourage cultural competency and cultural humility through quality measures, such as Consumer Assessment of Healthcare Providers and Systems measures related to patient-provider communication and health literacy.
Exhibit 3: Summary of implementation recommendations for equity-focused engagement of patients and people in VBP
Source: Authors’ analysis (from literature and policy document scans, and interviews with equity-focused value-based payment model implementors and experts).
Implementation Considerations For Cross-Sectoral Partnerships And Addressing Social Drivers Of Health
One critical way to improve health equity is by addressing social drivers of health, which requires collaborations across the health care and social and human services sectors as well as social needs screening, referral, and resolution. Our recommendations in this area are discussed below and summarized in exhibit 4.
Provide Incentives To Screen For And Address Social Drivers Of Health
Some VBP models have provided incentives for health care providers who engage patients and community-based organizations (CBOs) on health-related social needs. The Innovation Center previously tested the Accountable Health Communities (AHC) Model to encourage providers to conduct universal screening for social needs, although financial incentives were limited to infrastructure and staffing payments to conduct screenings. The model successfully facilitated screening patients for social needs, noting the importance of community health workers to this success, and this highlights that quality measures around social needs can be a starting point for VBP models to incentivize progress in social needs screening. And to this end, CMS is currently asking for more information about using two new health equity-focused quality measures—around social needs screening and patient survey questions around discrimination. Moreover, the National Quality Forum is developing measures for food insecurity that go beyond screening to assess referrals and improvements in food insecurity outcomes, but further research is necessary on developing new measures to encourage providers to “close the loop” on a wider array of social needs.
Connecting Providers To Social And Human Service Sector
Screening for social needs is only a first step. The AHC Model evaluation revealed that many organizations were limited in their ability to resolve social needs for those screened, largely due to limited referral and follow-up ability along with a general lack of community partnership resources. This implies that VBP models will need to improve the broader infrastructure and connections between the health care sector and the social and human services sector to facilitate referrals and care coordination.
One approach to begin this process is to use cross-sector data exchanges to connect health and social needs resources. Examples from states include New York’s eHealth Collaborative, a nonprofit that received Department of Health and Human Services funding to support the exchange of community resources across existing statewide health information exchange networks, and Colorado’s Office of eHealth Innovation, which is working with the Metro Denver Partnership for Health to establish an interoperable Social-Health Information Exchange. Both programs use data exchanges to help clinicians make referrals to social programs in a community when a patient shares their social needs.
At the federal level, the Office of the National Coordinator for Health Information Technology (ONC) has also launched the Social Determinants of Health Information Exchange Learning Forum, a valuable first step toward developing policies to exchange health and social determinants of health data. We recommend that policy makers at the state and federal level encourage and materially support the creation of further social-health data exchanges. Furthermore, we recommend ONC develop regulations around cross-sector data interoperability to facilitate standardization of social-health data exchanges.
A second approach is to go beyond data exchanges and use technology platforms to facilitate cross-sector engagement across clinicians, social services, and CBOs. Through a state-led public-private technology partnership, North Carolina’s NCCARE360 was the first statewide network to unite health and local social service organizations on a shared technology platform to “close the loop.” The state embedded NCCARE360 within North Carolina’s current 1115 waiver program to address social needs through Medicaid managed care (the Healthy Opportunities Pilots program), including an invoicing function for CBOs to bill for health-related social needs services (although payment is handled through Medicaid managed care organizations). Lessons from early roll-out of NCCARE360 for policy makers to consider include using the technology to automate any program-required reporting, allowing for referrals to services instead of just to CBOs, and supporting languages commonly used among the patient population. We recommend that payers, especially states, work with technology partners to develop a closed-loop referral system and standardize its components and functionality.
Cross-Sectoral Technical Assistance
Many health equity-focused VBP design elements are new to the field, and so providers could be supported through equity-focused technical assistance programs. CMS’s Office of Minority Health developed the Health Equity Technical Assistance Program to provide coaching and resources for embedding health equity into provider strategic plans. This program includes resources to improve language access, care for historically marginalized populations and people with disabilities, and data collection and analysis. We recommend that CMS build off this program to provide a more prescriptive approach to creating technical guidance for the health equity plans that they require in the ACO REACH model, and in other VBP models where equity-focused care will soon be prioritized.
One helpful example to note is the Health Resources and Services Administration has funded National Training and Technical Assistance Partners (NTTAP) for health centers to receive training and technical assistance from external entities with specific subject matter expertise. NTTAP partners have helped increase access to high-quality primary care for special and vulnerable populations. This includes engaging specific communities (for example, working with the Association of Asian Pacific Community Health Organizations to engage Asian American Pacific Islander and Native Hawaiian communities) or addressing specific health or health-related disparities (for example, working with the National Health Care for the Homeless Council to provide technical assistance on addressing the needs of those with unstable housing).
Another example is the Partnership to Align Social Care, a National Learning and Action Network, funded by foundations, health payers, and health systems. This organization brings together partnership members and subject matter experts to share learnings and align priorities to further the Innovation Center’s strategy refresh goals pertaining to health transformation to address social needs. We recommend that CMS could build off of these types of infrastructure specific to VBP model equity in two ways: by linking equity-focused technical assistance learning collaboratives with networks of external partners such as the NTTAP and the Partnership, and by expanding the partnership linkage to not just key external stakeholders but key governmental entities as discussed two sections above.
Importantly, these technical assistance plans must not be “one-size-fits-all” and must be tailored to the needs of individual communities and involve regional technical assistance networks that best understand the priorities of the community. We recommend creating equity-focused learning collaboratives to tailor technical assistance plans to have the most impact on providers within specific regions, including sharing best practices for equity-focused communication, culturally competent care, and effective ways to collect social determinants of health data.
Exhibit 4: Summary of VBP implementation recommendations to support equity-focused cross-sectoral collaboration to address social needs
Source: Authors’ analysis (from literature and policy document scans, and interviews with equity-focused value-based payment model implementors and experts).
Considerations For Future Work On Equity-Focused VBP Evaluation
New equity-focused VBP models are underway, but evidence is needed to refine the models and understand what works. Below, we provide a brief overview of directions for future work to embed equity principles within evaluation methods to assess the degree to which equity-focused VBP design and implementation worked.
An initial way to assess impact on equity at a program level is by stratifying cost and quality performance metrics, such as by race and ethnicity or by other sociodemographic factors. Such stratification depends on better and standardized data collection on race and ethnicity, disability status, and other populations who may have disparities. Further health equity performance measures could include process and outcome measures targeting specific subpopulations with known disparities (for example, appropriate antibiotic usage for children with sickle cell anemia) and process and outcome measures assessing the impact of disparity-reducing interventions (such as access to translator services).
The Assistant Secretary for Planning and Evaluation’s Office of Health Policy convened a technical expert panel to assess 10 approaches to health equity measures, releasing their report last year. The CMS Office of Minority Health’s Health Equity Summary Score received the panel’s highest rating due to its use of data from multiple performance measures, social risk factors, and types of data comparisons. We recognize that more work needs to be done to develop equity-focused performance metrics that do not rely on stratification, and we recommend that CMS focus on including composite indices that include data from multiple performance measures.
In addition, standard modeling approaches commonly used in VBP model evaluation are prone to poorly assessing equity. Researchers have described examples of how to apply a broader suite of more sophisticated statistical methods that can be used to properly study equity. VBP model evaluators, such as CMS, must use an expanded set of methods, including various types of regression analyses, matching and propensity score methods, instrumental variable analysis, marginal structural models, fixed-effects analyses, and multilevel modeling. Complex systems and simulation modeling methods—such as agent-based modeling—are particularly useful for assessing health equity interventions in a way that more fully accounts for the intricate relationships between upstream health determinants and downstream health disparities.
To effectively evaluate VBP effectiveness at improving health equity, we recommend that CMS embed evaluation design in conceptual models and use an expanded set of statistical methods to reduce biases in model assessment, capture spillover effects, and more thoroughly understand how models are (or are not) improving health equity.
Building off of our previous work, this piece discusses major equity-focused design, implementation, and evaluations considerations for VBP models. We discuss ways to ensure VBP models serve people from diverse communities, provide new tools for patient communications and engagement in model design and implementation, and provide tools for cultural competency and humility in engaging people in VBP. We also discuss cross-sector data exchange and referral platforms, incentives for meeting social needs, and equity-focused technical assistance programs. Finally, we outline how future evaluations might capture evidence on what is improving equity. The approaches and future directions summarized in this article will assist CMS, states, and commercial payers to maximize the potential of VBP health care transformation to advance health equity.
Support for this brief was provided by the Robert Wood Johnson Foundation. The views expressed here do not necessarily reflect the views of the Foundation. We would like to acknowledge specific support and interest from Katherine Hempstead, PhD. Dr. Bleser has previously received consulting fees from Merck for research for vaccine litigation unrelated to this work, from BioMedicalInsights, Inc., for subject-matter expertise on value-based cardiovascular research unrelated to this work, from Gerson Lehrman Group, Inc., on health policy subject-matter expertise unrelated to this work, and from StollenWerks, LLC, on health policy delivery system change unrelated to this work. He also serves as board vice president (uncompensated) for Shepherd’s Clinic, a clinic providing free health care to the uninsured in Baltimore, Maryland. Dr. Saunders has a consulting agreement with Yale-New Haven Health System for development of measures and development of quality measurement strategies for the Center for Medicare and Medicaid Innovation alternative payment models under CMS contract number 75FCMC18D0042/task order number 75FCMC19F0003, “Quality Measure Development and Analytic Support,” Option Year 2. The Duke-Margolis Center for Health Policy values academic freedom and research independence, and its policies on research independence and conflict of interest are available here.