Time for Value-Based Payment Models to Adopt a Disparities-Sensitive Frame Shift

More than 1500 hospitals recently signed the American Hospital Association's 123forEquity pledge and have begun implementing strategies to address economic and racial disparities in health care provision and outcomes. Despite these institutional commitments to equity, evidence suggests that the current growth in value-based payment models may counter these efforts, unintentionally worsening health care disparities by disadvantaging hospitals caring for socially at-risk populations (1). Because alternative payment models account for a mounting proportion of payments, equity will remain elusive unless explicitly incorporated into what we measure and pay for in health care. How can we move toward universal high performance standards while narrowing disparities? Much of the onus has fallen on how to actuarially adjust payments and penalties using socioeconomic risk factors (24). This method is insufficient for a few reasons. First, current socioeconomic risk measures have considerable limitations: We can adjust only what we can measure, and we may not be measuring the right exposures. Take Medicare data, a key driver of payment reform, in which sociodemographics are limited to race/ethnicity, dual eligibility for Medicaid and Medicare, and residency in rural or low-income areas. Although these measures are often good proxies for socioeconomic risk, others may better explain between-population variations in health outcomes, cost, and quality. Second, some experts are justifiably concerned that adjusting for socioeconomic risk, regardless of the measures used, may lower quality standards altogether, disincentivizing hospitals to improve performance (2). Therein lies the rub: The health care industry cannot ignore true instances of poor quality, but it also should not worsen health care for at-risk populations. To address this tension, value-based payment models should adopt a disparities-sensitive frame shift to integrate measures of equity into hospitals' financial calculus, incentivizing hospitals to tackle the disparities challenge without losing sight of quality. Achieving this frame shift requires us to continue determining which metrics matter most when addressing disparities in health care delivery and outcomes, both to improve risk adjustment and to establish which measures are actionable. To avoid underpaying hospitals that disproportionately serve socially at-risk patients, we should assess how risk adjustments would perform if they accounted for factors like socioeconomic position, social relationships, and community context, which the National Academy of Medicine has identified as key domains affecting health care outcomes (1, 5). The Medicaid payment formula used in Massachusetts is an example of how incorporating medical and social risk factors (such as mental health and housing instability) improved payments to hospitals by matching predicted to actual costs for high risk patients better (6). Practice-level coding of existing International Classification of Diseases, 10th Revision, codes for conditions like homelessness (Z59.0) and extreme poverty (Z59.5) may improve documentation and shift us toward adopting these billable diagnoses as a substantive component of hospital payment. These examples also highlight real-world laboratories in which to continue refining our approach to health equity: Which social risk factors most affect health outcomes? Which (alone or combined) provide the highest value, when modified, to produce cost savings or better outcomes? Concurrently, we should start paying hospitals to reduce disparities directly. Evidence indicates that hospitals with previously poor performance made the most gains in quality within alternative payment models (7), but we do not universally reward them for these gains. We certainly do not reward them for gains in equity. Furthermore, hospitals with poor patients are at a disadvantage. For example, under Medicare's Hospital Readmission Reduction Program, 86% of hospitals with poorer patients have been penalized in 2017, compared with 66% of those serving wealthier patients, a consistent pattern across the program's first 5 years (8). We believe that improving the financial outlook for hospitals that disproportionately care for high-risk populations (particularly rewarding hospitals that do this well) can reduce such stark disparities between hospitals. One approach is to pay for improvements in health care outcomes for populations we know have high social risksminority, low-income, and dual-eligible patientsa strategy already embedded in high-risk management programs used by accountable care organizations (9). Another approach ensures that hospitals address disparities by paying directly for services that disproportionately affect high-risk communities. For example, mental health is underpaid and likely a driver of disparities (10). Yet rather than incentivizing mental health treatment, value-based payment models have promoted spending on higher-cost services, such as oncology and cardiovascular care, to shift outcomes. We would rather see hospitals paid directly both for taking on high-risk populations and for making exceptional reductions in disparities. If we used financial incentives to drive hospitals toward reducing disparities, how they chose to intervene could rely on locally informed initiatives with potential for health returns. Which interventions have high value and how hospitals can better align with community-based initiatives is a growing field of study. More data will soon be available: The Centers for Medicare & Medicaid Services recently invested $157 million in testing the Accountable Health Communities Model to screen Medicaid and Medicare beneficiaries for health-related social needs and connect them with local services. Meanwhile, the Camden Coalition of Healthcare Providers and the Abdul Latif Jameel Poverty Action Lab at the Massachusetts Institute of Technology are conducting a randomized trial on the effect of care management initiatives with community linkages on readmissions for high-risk patients. Such rigorous approaches are needed to inform next steps, but in the meantime, to further drive the adoption of innovative health delivery solutions that address social risks, a payment frame shift is needed to promote equity as a health care goal. Payment reform is not a panacea to our health care system's challenges, but it is a critical vehicle for aligning our goals with incentives to do better. In this context, our suggestions are not without shortcomings, nor are they comprehensive. We need to continue monitoring for unintended effects, such as patient selection, and for the possibility that payment incentives improve process measures but not health outcomes. Inevitable tradeoffs between equity and efficiency also warrant further discussion. But as a starting point, we can shift our conversation away from health care disparities as an unintended consequence of payment models and toward incentivizing hospitals to tackle this problem directly. If we want a rising tide to lift all boats, a disparities-sensitive frame shift in payment reform may help us get there sooner.