As payers are increasingly holding providers financially accountable for outcomes under value-based payment arrangements, there is growing concern that organizations caring for populations with greater social risk factors are unfairly penalized. Those voicing this concern often suggest that it is necessary to adjust performance scores for social risk factors to fairly assess and reward providers caring for populations with differing levels of socioeconomic vulnerability. The underlying premise is that achieving favorable outcomes for patients with greater social risk is more difficult or requires different resources than achieving the same level of outcomes in a more socially advantaged population. Other observers object to adjusting performance scores for social risk factors, seeing this as excusing or accepting of lower-quality care being delivered to socially at-risk populations and as masking lower performance with statistical adjustments.
We propose two policy approaches designed to satisfy these seemingly divergent perspectives. Our approaches acknowledge the differential resources that may be required to achieve a given level of outcomes in a socioeconomically vulnerable population and compensate providers differently for this care. That is, rather than adjusting the performance scores to account for a more at-risk population, we propose to adjust payment.
One payment adjustment would be applied prospectively. Much as the Centers for Medicare and Medicaid Services (CMS) adjusts payment to Medicare Advantage plans based on medical risk, this model would increase upfront payments to providers who care for populations with greater social risk in recognition of the extra resources required to care for this population. The second payment method would be applied as part of performance incentive payments—establishing a multiplier such that those serving a more socially at-risk population would be rewarded more for achieving the same level of performance as their peers serving a more advantaged population.
Below, we outline pros and cons for each approach, illustrate how adjusting budgets rather than performance standards for social risk is an optimal approach, and suggest options for policy makers’ consideration.
Value-Based Payment And Controversy Around Social Risk
Over the past two decades, public and private payers have introduced payment innovations that aim to motivate and reward provider performance. The initiatives were, in part, a response to widely publicized research showing significant failings of the US health care system on multiple dimensions of quality and more recently to the unrelenting rise in health care spending. As these programs moved from the earliest generation of metrics, which focused largely on a handful of process-based measures, to more robust models focused on accountability for outcomes and total costs of care, providers have grown increasingly vocal about concerns that the models do not fairly account for the difficulty of achieving good performance in populations that have high levels of social risk.
CMS has been an important leader in performance-based payment innovations. Since 2008, CMS has implemented a series of programs that reward and penalize providers based on a variety of process and outcomes measures. In 2010, the Affordable Care Act set the stage for an array of value-based payment (VBP) programs for hospital and physician payments, including the Hospital Readmissions Reduction Program. In 2014, the Protecting Access to Medicare Act established the Skilled Nursing Facility VBP, and a Home Health VBP began in 2016 in selected states. One persistent concern cutting across these programs has been the absence of any mechanism to account for differences in the social risk of providers’ populations when evaluating and rewarding performance.
Some observers argue that not adjusting for these factors unfairly penalizes safety-net providers, which may require more—or at least, different—resources to achieve a given level of performance than providers treating more advantaged populations. Others argue that adjusting for social risk would create a two-tiered system, excusing lower-quality care provided to those with higher social risk.
Prospective Social Risk Adjustment
One way to account for the different type or number of resources required to achieve good outcomes for disadvantaged populations is to provide upfront supplemental payments to providers who care for disproportionately higher-risk populations. This is analogous to how Medicare accounts for medical risk. For example, benchmarks used for rate-setting in Medicare Advantage are adjusted for medical risk. Similarly, CMS pays hospitals based on diagnosis-related groups that account for medical condition severity in the base rate. In these ways, the upfront payments to providers are greater when medical risk is deemed higher.
Implicit in these upfront payment adjustments is the notion that all providers can be held to the same performance standard once the additional resources have been provided. For example, the readmission performance targets under the Hospital Readmission Reduction Program for heart failure patients are the same for those with or without comorbidities. Rather than setting a different performance target based on medical complexity—for example, allowing for 25 percent readmission rates for heart failure patients with medical comorbidities but only 15 percent for those without comorbidities—a common performance standard is applied; but greater resources are afforded to providers caring for the medically complex population.
Social risk can be treated similarly. A health plan or provider group that cares for a population with greater average social risk would receive higher monthly payments than if they enrolled a population with equivalent medical risk but lower social risk. Quality performance targets would be identical. Although achieving the same outcomes may be a greater challenge in the population with greater social risk, the additional upfront payments would give plans and providers the resources to invest in activities that support their populations’ needs. This would have the additional benefit of creating opportunities for plans and providers to incorporate the social determinants of health into the care they deliver. For example, Geisinger Health System has found that patients with diabetes who receive healthy food directly from their providers have improved control of blood sugars, blood pressure, and cholesterol. This payment approach would create not only the opportunity but the obligation to seek out these kinds of solutions.
One advantage to this approach is that it is consistent with how patients’ medical risk is accounted for in payment adjustments and as such could be fully incorporated into existing risk-adjustment methodologies. Most importantly, it ensures that different standards for high-quality care are not applied to different populations. Disadvantages are that it may be difficult to implement, may require additional funding to initiate, and demands accountability from providers who would be responsible for making the needed investments to achieve the desired outcomes.
Adjustments To Performance Incentive Payments
Under the second approach, performance standards would be the same for all providers, irrespective of their population mix, but a multiplier would be applied such that organizations serving a population with greater social risk would receive larger rewards for an equivalent level of outcome, compared to providers caring for lower-risk populations.
The Blue Cross Blue Shield of Massachusetts Alternative Quality Contract (AQC) is an example of a program that used the same performance standards for all providers, irrespective of social risk. For each performance measure, the AQC established a range of performance targets (called gates), and higher payment was provided for each increment of improvement across the continuum from the lowest to highest “gate.” The model drove significant performance improvements among providers serving populations with greater social risk and, as a result, closed long-standing disparities in quality and outcomes. Similarly, under the Medicare program that penalized providers with the highest rates of unplanned 30-day readmissions, hospitals serving populations with higher social risk showed greater gains in performance than other hospitals. In both cases, prior to these performance incentive programs, those serving a more socioeconomically disadvantaged population had long-standing and seemingly intractable deficits in quality and outcomes relative to those serving more advantaged populations. Had the programs set a different, lower bar for performance standards for these providers, it seems unlikely that the resulting improvements would have occurred.
While results of both programs demonstrate the feasibility of achieving performance as high or higher in populations with greater social risk, even in the absence of added financial reward, widespread concerns about fair financial treatment of providers caring for socially disadvantaged populations compel a policy solution. By adjusting payment rather than performance scores, the proposed policy solutions address this concern without lowering performance expectations and implicitly accepting lower-quality care for patients with higher social risk.
Continuing the illustration of hospital readmissions for a population of patients with heart failure, a single target could still be the goal for all patients. Under our proposal, however, if the top performance was a readmission rate less than 10 percent, a hospital with that level of performance and a population with social risk factors in the highest quartile might receive twice the reward of that given to a hospital with similar performance but a socioeconomically advantaged population.
Advantages to this approach are that it would help guard against providers avoiding socially disadvantaged populations, support more person-centered care by incentivizing providers to address patients’ social needs, and may be relatively straightforward to implement. Since, by design, providers caring for populations with higher social risk will have higher rewards at any given level of performance, one disadvantage is this could dampen motivation for continued improvement by some providers.
Social Risk Metrics
Either of these policy approaches—adjusting budgets upfront or having a multiplier on performance payments—will require a robust indicator of social risk. At present, owing to sparse and inconsistent data on key markers of social risk in providers’ populations—such as race, ethnicity, income, and educational attainment—a proxy will be required. One such proxy recommended by the Medicare Payment Advisory Commission is the percentage of a provider’s patient population that is dually enrolled in Medicare and Medicaid. While this may be a good first step, there are limitations to this approach, including the variability among states’ Medicaid program eligibility and the fact that dual-eligible status does not fully correlate with the presence or absence of social risk factors.
Another approach worth exploring is use of sociodemographic data drawn from the US census. In particular, data at the Census Block Group (CBG) level is highly proximate to the individual patient/person and corresponds closely with self-reported data for these same variables. The census provides hundreds of variables measured at the CBG level, many of which could offer a rich profile of population social risk (for example, language spoken at home, percent of households below 200 percent of the federal poverty level, race, ethnicity). Address information for patients treated by a given provider, linked to CBG data, could be used to construct a provider’s social risk index.
Over the long term, the ideal approach is systematic collection and use of data on social risk factors from patients as a routine part of care delivery—although using such data for payment purposes will require validation that would not be necessitated with use of exogenous indicators such as dual Medicare/Medicaid status or CBG-derived data. Nonetheless, given increasing awareness of the importance of addressing social risk factors as part of good health care, the routine collection and use of these data for purposes of the clinical interaction seems necessary.
Suggestions For Policy Makers
Given the increased role of value-based payment in health care and the importance of assuring that such programs do not exacerbate health disparities, policy makers should take concrete steps to implement models that adjust payment to account for differential social risk in provider populations. By adjusting payment rather than performance scores, such programs can appropriately reward and resource those caring for socioeconomically vulnerable populations. In so doing, these approaches would enable providers to be held to a common performance standard, regardless of the population served, and avoid perpetuating health disparities and accepting lower standards of care for those with greater social risk. Accordingly, we recommend that policy makers:
- Act now. Both approaches discussed in this post could be implemented immediately within many existing value-based payment programs. Using the proportion of dually eligible individuals as an interim indicator of social risk would enable payment adjustments to begin right away, while a more robust indicator of social risk is developed.
- Begin work on a more robust indicator of social risk. This should include enhanced integration of data on social determinants of health, potentially leveraging CBG-level data. Patients and families with high social risk should be part of the development process.
- Evaluate program results, assure these new approaches achieve the desired impact, and be prepared to adjust course as needed.