The committee’s first four reports detail what the Assistant Secretary for Planning and Evaluation (ASPE) and the Centers for Medicare & Medicaid Services (CMS) could do if they choose to account for social risk factors in Medicare payment. Recommending what ASPE and CMS should do is beyond the scope of the committee’s task. Thus, the committee’s reports aim to inform analyses the ASPE is conducting to identify appropriate strategies to account for social risk factors in Medicare payment should CMS choose to do so. To that end, throughout the report series, the committee presented an array of optional methods to account for social risk factors as well as a variety of data sources for social risk factor indicators ASPE and CMS could use. At the same time, although Chapter 3 provides some additional guidance on how CMS could use the methods identified individually or in combination to achieve policy goals, the committee recognizes that implementing any approach to accounting for social risk factors in Medicare quality measurement and payment can be complex and will require substantial analyses to identify the best approaches to do so for different Medicare incentive programs. Implementing any approach to accounting for social risk factors in any of Medicare’s many value-based payment (VBP) programs will also require considerable resources—including costs. This final chapter provides some general conclusions and other considerations to guide ASPE and CMS if they choose to begin accounting for social risk factors in Medicare payment programs, and to help them to identify priorities and preferences from among the options presented.
As the committee concluded, it is possible to improve on the status quo with regard to the effect of VBP on patients with social risk factors (NASEM, 2016a).1 Furthermore, the committee found that some social risk factors are already collected by CMS and could be used in the near term to account for social risk factors in Medicare payment, should CMS choose to do so. Three indicators of social risk are readily available in existing CMS data resources for use in the near term: dual eligibility, nativity, and urbanicity/rurality. Other indicators of social risk are available for use in the near term either for some outcomes or with some limitations. These include race and ethnicity, income, education, neighborhood deprivation, and housing. (The committee’s specific recommendations are described in detail in Appendix D.) Other social risk factors might require new data collection efforts, some of which CMS could start collecting in a standardized way in the near term. For example, although area-level education data that could be used as a proxy for individual education are available from census data in the near term, CMS could begin collecting individual education at the time of enrollment for use in the longer term. Similarly, CMS has some existing data on preferred language. However, to improve accuracy, CMS could begin collecting preferred language information in a standardized, consistent way at the time of enrollment.
Existing analyses do not address which of the social risk factors that may influence performance indicators used in VBP must be individually accounted for to ensure adjustments to performance measures and payment are accurate. It may be that some are not adequately measured using current data or data collection techniques. It may also be that a smaller set of indicators is sufficient. If the latter, the literature does not currently indicate which factors should be included. Thus, in order to determine which social risk factors should be incorporated in any given VBP system, their usefulness in explaining variation in outcomes should be investigated. ASPE has already conducted some analyses of the associations of different types of social risk factors with certain outcomes (Filice and Joynt, 2016; Samson et al., 2016; Snyder et al., 2016).
A related consideration pertains to one of the committee’s conclusions about new data collection. As the committee concluded, if there are substantial barriers to collecting social risk factor data (such as high cost) and/or if early pilot testing or modeling in a multivariable model suggests only marginal gains from including any given indicator in any method of accounting for social risk factors in Medicare performance measurement and payment, inclusion of that indicator may not be warranted (NASEM,
2016b).2 Applying this in practice can be challenging, because determining what constitutes a “marginal gain” can be difficult. Although it can be informed by pilot testing and modeling, a marginal gain does not reflect a strictly quantitative value—indeed, there is unlikely to be a threshold under which the effects of any social risk factor are considered to be “marginal.” Rather, a marginal gain reflects a balance of several competing considerations. These considerations include not only tradeoffs between collection burden and accuracy, but also the extent to which an indicator meets other selection criteria—for example, the extent to which there is a strong conceptual relationship and evidence of an empirical association between the indicator and the outcome(s) of interest. However, certain empirical analyses can help inform decision making about which social risk factors must be included in Medicare VBP and which may not warrant inclusion. Such analyses would be similar to those used to identify the set of clinical adjusters included in the CMS-Hierarchical Condition Category (HCC) clinical risk adjustment model (Pope et al., 2004). For the CMS-HCC model, CMS considered analyses on the predictive power of diagnostic categories individually and together, consulted with clinicians, and weighed the results against other selection criteria for inclusion of diagnostic categories (Pope et al., 2004). Similarly, for social risk factors, ASPE/CMS could conduct analyses on the predictive power of social risk factor indicators individually and together, and then they could weigh these results with advice from technical experts (including clinicians but also those with expertise in the social determinants of health), as well as the selection criteria the committee identified for inclusion of social risk factors in Medicare quality measurement and payment.
Further research could inform ASPE or CMS as it determines the optimal way in which to adjust indicators used in VBP for social risk factors. Specifically, further work would address the following questions:
- How can ASPE/CMS implement the use of an initial set of social risk factors on a rapid timeline?
- Which social risk factors present in claims or enrollment data, and already examined by ASPE/CMS, explain variation in outcomes when added to the clinical variables already included in risk adjustment models?
- Are some of these social risk factors so correlated with each other that a more parsimonious set would explain a substantively equivalent amount of variation in outcomes?
- How can ASPE/CMS implement the use of an expanded set of social risk factors?
- Which of those social risk factors not included in claims or enrollment data, but included in nationally representative surveys that include outcomes such as total costs of care or readmissions (e.g., the Medicare Current Beneficiary Survey, Medical Expenditure Panel Survey), appear to explain the most variation in outcomes (again, above and beyond clinical risk factors and currently available social risk factors)?
- Which methods of incorporating the promising social risk factors identified through the analyses above are the most feasible, accurate, and valid?
- How can ASPE/CMS monitor and refine the use of social risk factors in VBP?
- Are the methods to account for social risk factors achieving policy goals? What unintended adverse consequences might arise?
- What are the distributional effects of adjustment for social risk factors for providers? Which types of providers (e.g., safety-net hospitals or physicians in underserved communities who disproportionately serve Medicare beneficiaries with social risk factors) are the most likely to benefit or lose?
- Are there patient subgroups who appear to have benefited or been harmed by adjustment (e.g., through better access to care or information)?
Through research answering these questions, ASPE/CMS can determine the best path for implementing adjustments for social risk factors, and can ensure that doing so furthers the policy goals of VBP.
Even if it is appropriate to account for social risk factors in Medicare quality measurement and payment, incentivizing providers to find strategies to improve access, quality, and outcomes for socially at-risk populations is critical to the goal of promoting health equity (NASEM, 2016a). The committee’s second report, Systems Practices for the Care of Socially At-Risk Populations, shows that there are strategies health care providers (i.e., hospitals, clinics, and physician groups) and payers (i.e., health plans) can undertake to improve care and health outcomes for socially at-risk populations (NASEM, 2016c). To identify innovations, interventions, and other strategies providers disproportionately serving socially at-risk populations are implementing to improve the quality of care for and health care outcomes of their patients, the committee reviewed case studies from the peer-
reviewed and grey literature. Based on this review, as well as consideration of the peer-reviewed literature on quality improvement and health disparities and, in some cases, committee members’ empirical research or professional experience delivering care to patients with social risk factors, the committee identified a set of commonalities across the strategies identified. These themes describe a set of practices delivered within a health care system, where the system encompasses a set of related actors who collaborate to improve health equity and outcomes for socially at-risk populations. In this approach, the health care system includes not only medical providers, but also partnering public health and social service agencies, community-based organizations, and the community in which those medical providers are embedded. The medical providers may be formally connected to its partners through legal arrangements or informally connected, but all serve the same community or geographic region.
As the committee concluded, six community-informed and patient-centered practices show promise for improving care for socially at-risk populations (NASEM, 2016c).3Figure 4-1 illustrates these practices, which are grounded in community-informed and patient-centered care. These practices start with a commitment to health equity and make up an approach by which health care systems can promote equitable health outcomes by using data to identify unmet clinical and social needs and by addressing those needs through collaborative partnerships that coordinate care across time, sites of care, and intensity of needed services. In so doing, health care systems can support patients living in the community to engage in their health care. This systems approach provides a continuous process for improvement.
Complete descriptions of the six systems practices and highlighted case studies can be found in Appendix B. Based on its review of case studies, the committee found that, with adequate resources, providers can feasibly respond to incentives to deliver high-quality care. Some providers disproportionately serving socially at-risk populations achieved performance on par with the highest performers among all providers (NASEM, 2016c). Thus, it is possible to deliver high-quality care to socially at-risk populations, and patients with social risk factors need not experience low-quality care and poor health outcomes (NASEM, 2016c). If accounting for social risk factors in Medicare payment improves fairness in compensating providers disproportionately serving socially at-risk populations, this would increase the resources available to support delivery of high-quality care and quality improvement efforts. Thus, accounting for social risk factors in Medicare payment can be an important and necessary step toward improving health equity. At the same time, adjusting payment mechanisms will not
reduce health disparities unless providers implement systems that deliver high-quality care to all patients while responding to the particular needs of patients with social risk factors. Accounting for social risk factors in payment and designing and delivering care responsive to social risk factors are therefore complementary approaches to promoting health equity.
Although accounting for social risk factors in Medicare payment is critical for improving equity among Medicare beneficiaries, the committee
notes the limitations of the scope of its task—and, indeed, any endeavor to account for social risk factors in financial incentive design. In particular, to the extent that accounting for social risk factors in Medicare payment improves fairness in compensating providers disproportionately serving socially at-risk populations, doing so may increase the resources available to these providers. Although the committee’s charge, and therefore its findings, conclusions, and recommendations, are focused on Medicare payment, the approaches the committee identified to account for social risk factors in quality measurement and payment could be applied to other payers. Similar improvements to VBP by other payers could increase resources available to providers disproportionately serving socially at-risk populations via the same mechanisms as under Medicare. However, any policy that modifies incentive payments (as accounting for social risk factors in these payment does) does not (and cannot) fix or, for that matter, substantively address the payment system at large. Thus, accounting for social risk factors does not solve the problem of safety-net financing. Indeed, other payment reforms (for example, direct payments for quality improvement among safety-net providers and direct payments to incentivize collaboration with public health and social service agencies and community-based organizations) may be more effective at incentivizing high-quality care for socially at-risk populations. Accounting for social risk factors is necessary but insufficient by itself to achieve health equity.
The committee urges policymakers to remember that quality measurement and payment policies affect the lives of real patients. In the case of accounting for social risk factors, changes to the current VBP system would especially influence the lives of patients with social risk factors who have historically experienced barriers to accessing high quality health care. Together, accounting for social risk factors in quality measurement and payment in combination with complementary approaches—other payment reforms, implementing strategies to improve the quality of care for patients with social risk factors and to reduce disparities, and further research to identify what drives observed differences in quality and outcomes for patients with social risk factors—may achieve the policy goals of reducing disparities in access, quality, and outcomes, and quality improvement and efficient care delivery for all patients, and thereby promote health equity.
Filice, C. E., and K. E. Joynt. 2016. Examining race and ethnicity information in Medicare administrative data. Medical Care. doi: 10.1097/MLR.0000000000000608.
NASEM (National Academies of Sciences, Engineering, and Medicine). 2016a. Accounting for social risk factors in Medicare payment: Criteria, factors, and methods. Washington, DC: The National Academies Press.
NASEM. 2016b. Accounting for social risk factors in Medicare payment: Data. Washington, DC: The National Academies Press.
NASEM. 2016c. Systems practices for the care of socially at-risk populations. Washington, DC: The National Academies Press.
Pope, G. C., J. Kautter, R. P. Ellis, A. S. Ash, J. Z. Ayanian, M. J. Ingber, J. M. Levy, and J. Robst. 2004. Risk adjustment of Medicare capitation payments using the CMS-HCC model. Health Care Financing Review 25(4):119-141.
Samson, L. W., K. Finegold, A. Ahmed, M. Jensen, C. E. Filice, and K. E. Joynt. 2016. Examining measures of income and poverty in Medicare administrative data. Medical Care. doi: 10.1097/MLR.0000000000000606.
Snyder, J. E., M. Jensen, N. X. Nguyen, C. E. Filice, and K. E. Joynt. 2016. Defining rurality in Medicare administrative data. Medical Care. doi: 10.1097/MLR.0000000000000607.