When developing and selecting methods to account for social risk factors in Medicare quality measurement and payment applications, understanding the type of incentive design is important in evaluating the potential benefits and challenges of various accounting methods. The incentive design will interact with the method used to account for social risk factor(s) and produce certain potential harms. Selecting the appropriate method (or, methods) to account for social risk factors will require weighing these potential harms. Given that the Medicare payment landscape is evolving and the Centers for Medicare & Medicaid Services (CMS) is moving toward more comprehensive population-based alternative payment models (APMs), the committee developed methods that could apply to any Medicare quality measurement and/or payment program, not just the existing ones. The chapter begins with a brief review of the current Medicare payment landscape, with a focus on capitated payments to Medicare Advantage (also known as Medicare Part C) and Medicare Part D plans and on value-based payment (VBP) programs that tie payment to performance in traditional Medicare and Medicare Advantage, and the planned developments. The chapter proceeds with describing the potential benefits and harms of the status quo (not accounting for social risk factors) and compares them to the potential benefits and harms of accounting for social risk factors generally. The chapter then proposes alternative methods for accounting for social risk factors. The chapter closes with guidance on an approach to applying the methods to achieve simultaneous goals of reducing disparities in access, quality, and outcomes; quality improvement and efficient care delivery for all patients; fair and accurate public reporting; and compensating providers fairly.
The Patient Protection and Affordable Care Act of 2010 (ACA), and subsequent legislation such as the Improving Medicare Post-Acute Care Transformations (IMPACT) Act of 2014 and the Medicare and CHIP Reauthorization Act of 2015 (MACRA) require CMS to implement VBP programs. VBP aims to align payment and care delivery goals to improve health care quality and outcomes, while also controlling costs (Rosenthal, 2008). Together these reforms shift focus from delivery of and payment for individual services to a system that focuses on population health management and holds providers accountable for both quality and cost (McGinnis, 2016; Rajkumar et al., 2014).
In addition to congressionally mandated requirements to implement VBP programs, in 2015, Secretary of Health and Human Services (HHS) Sylvia Burwell announced a goal for CMS to have 30 percent of Medicare payments in APMs by the end of 2016 and 50 percent by the end of 2018, as well as to have 85 percent of Medicare payments tied to quality or value by 2016 and 90 percent by 2018 (Burwell, 2015). As described in the committee’s first report, CMS currently administers eight VBP programs and has two in planning (NASEM, 2016a). Additionally, CMS is continually developing and reorganizing more VBP programs, and the Center for Medicare & Medicaid Innovation (CMMI) also tests innovative payment models.
CMS payment models cover a spectrum of approaches from traditional fee-for-service to population-based payment models. As described in Appendix C1 and in the committee’s first report (see Appendix A), VBP models fall into two broad categories, which the committee roughly categorizes as financial incentives and APMs (NASEM, 2016a). Financial incentives such as pay-for-performance programs link financial bonuses and/or penalties to quality or value (NASEM, 2016a). APMs include episode-based payments and population-based (global) payments, shifting greater financial risk to providers to hold them accountable for the quality and efficiency of care they provide, as well as health outcomes achieved (NASEM, 2016a). Additionally, although not considered entirely VBP models nor do they classify strictly as financial incentives or APMs, Medicare Advantage and Part D have design features that tie quality and cost performance to payment, and thus are relevant for purposes of accounting for social risk factors in payment. They also include risk sharing that necessitates consideration of risk adjustment for the capitation amount or global spending target or may include VBP mechanisms such as bonus payments. Moreover, the study sponsor, the Office of the Assistant Secretary for Planning and Evaluation of HHS, included Medicare Advantage and Medicare Part D as relevant payment models in its presentation to the committee at the first meeting
(Epstein, 2015). CMS VBP programs and their specific incentive designs are described briefly below.
Current Financial Incentive Programs
Penalties for Poor Performance
Hospital-Acquired Condition Payment Reduction Program Implemented beginning fiscal year (FY) 2015, the Hospital-Acquired Condition Payment Reduction Program reduces payments to acute care hospitals paid under the Inpatient Prospective Payment System based on their performance on select hospital-acquired condition quality measures, including the Agency for Healthcare Research and Quality Patient Safety Indicator 90 and the Centers for Disease Control and Prevention (CDC) National Healthcare Safety Network infection measures. The bottom 25 percent worst performing hospitals receive a payment reduction of 1 percent for all discharges in those hospitals.
Hospital Readmissions Reduction Program Begun in 2012, the Hospital Readmissions Reduction Program (HRRP) is a penalty program for acute care hospitals paid under the Inpatient Prospective Payment System. The HRRP requires CMS to reduce a share of the base operating payments to hospitals that have excess readmissions (CMS, 2014b). For FY 2013 and FY 2014, CMS calculated excess readmissions for three conditions: acute myocardial infarction, heart failure, and pneumonia. In FY 2014, CMS refined the measures to account for planned readmissions, and in FY 2015, the program was expanded to include excess readmissions from chronic obstructive pulmonary disease and total hip arthroplasty or total knee arthroplasty. The program further expanded to include coronary artery bypass graft surgery for FY 2017 (HHS, 2014; NASEM, 2016a). For FY 2013, the maximum reduction was 1 percent of the hospital’s base operating payment; for FY 2014, the maximum reduction was 2 percent; and in FY 2015, the maximum reduction was 3 percent (CMS, 2014b). For FY 2016, the maximum reduction remains 3 percent (HHS, 2014; NASEM, 2016a).
Rewards and Penalties for Performance
Hospital Value-Based Purchasing Program The Hospital Value-Based Purchasing Program is the only Medicare hospital incentive program that offers both rewards and sanctions. Beginning FY 2013, acute care hospitals paid under the Inpatient Prospective Payment System became eligible for rewards and penalties based on performance on quality, patient experience, and efficiency (Medicare spending per beneficiary). Incentives could total up to 1 percent in FY 2013 and increase in 0.25 percent increments annually to
2 percent in FY 2017 and subsequent years (MLN, 2013). The program is a budget neutral program, so total incentive payments must equal the total payment reductions (penalties).
Physician Value-Based Modifier Program Required by the ACA and established by CMS beginning in 2015, the Physician Value-Based Modifier is a budget-neutral, pay-for-performance program (CMS, n.d.-d). In this program, physicians can receive incentive payments or penalties based on performance on quality, costs, and patient experiences of care. The program divides physicians into two categories based on whether physicians meet minimum reporting requirements using the Physician Quality Reporting System (category 1) or not (category 2). In category 1, physicians are eligible to receive either upward or downward adjustments based on their performance on quality and costs. Physicians in category 2 are subject to a modifier payment set at a fixed downward adjustment (1 percent in 2015 and 2 percent in 2016). Because the program is budget-neutral, total upward adjustments for category 1 must equal total downward adjustments for categories 1 and 2 combined.
Current Alternative Payment Models
APM with Downside Risk
End-Stage Renal Disease Quality Improvement Program The Medicare Improvements for Patients and Providers Act (MIPPA) of 2008 authorized the End Stage Renal Disease (ESRD) Quality Improvement Program. MIPPA requires CMS to reduce payments to outpatient dialysis facilities treating ESRD patients based on the clinical measures that assess a facility’s performance and reporting measures (i.e., whether facilities have met reporting requirements) (CMS, 2015a). Beginning in 2012, CMS reduced the bundled payment rate to ESRD facilities with poor performance by up to 2 percent. To determine penalties, CMS first calculates both an achievement and improvement score for each clinical measure (except the CDC National Healthcare Safety Network Bloodstream Infection in Hemodialysis Outpatients measure, which receives only an achievement score) (CMS, 2014a). Facilities that meet a minimum total performance score receive full payment, while those that fall under this threshold are subject to a reduction between 0.5 percent and 2.0 percent (CMS, 2014a, n.d.-a).
APM with Upside Gainsharing and Downside Risk
Medicare Shared Savings Program The Medicare Shared Savings Program (MSSP) is a key provision of the ACA that establishes accountable care
organizations (ACOs), a payment and delivery system model that aims to motivate better care coordination, better quality care, and more efficient care through payment reforms (CMS, 2015d). CMS is phasing in the program with two tracks: a one-sided model (shared savings only) and a two-sided model (shared savings and losses). Before each performance year, CMS calculates a risk-adjusted, historical benchmark for per-beneficiary costs. At the end of each performance period, CMS compares the actual spending of each MSSP ACO to the calculated benchmark. Organizations that meet a minimum saving threshold qualify for shared savings, while those that meet a minimum loss threshold must share losses.
Other Current Value-Based Payment Models and Mechanisms
Medicare Advantage/Part C Bonus Payments
Medicare Advantage is the insurance program that covers Part A (inpatient care) and Part B (outpatient care) benefits, typically offers Part D prescription drug coverage, and may also offer additional benefits and services for additional cost (MedPAC, 2015a). For beneficiaries enrolled in MA plans (30 percent of all Medicare beneficiaries in 2015), CMS pays MA plans an annual capitated rate based on national and regional historical benchmarks that are risk adjusted using the CMS hierarchical condition category model. MA plans that achieve higher quality ratings under Medicare’s Five Star Quality Rating System are also eligible for quality bonus payments. In 2016, these bonus payments equal 5 percent of the county-level rate per beneficiary.
Medicare Part D is a pharmaceutical drug reimbursement program administered by CMS and run by Medicare-approved private insurance plans. CMS pays these plans in several ways, including direct subsidies, low-income subsidies for cost sharing and premiums (costs above the direct subsidy an enrollee otherwise pays for out of pocket), and two risk-sharing mechanisms: individual reinsurance and risk corridor adjustments. Through individual reinsurance, Medicare subsidizes 80 percent of drug spending above an out-of-pocket threshold (enrollee costs including the deductible and cost sharing, also known as the catastrophic cap), while the insurance plan pays 15 percent and the enrollee pays 5 percent (Medicare.gov, n.d.; MedPAC, 2014). Risk corridor adjustment limits plans’ potential gains or losses by financing costs that are higher than expected and recouping profits deemed excessive (MedPAC, 2015b). CMS calculates risk corridor adjustments at the end of each benefit year, comparing the plan’s actual costs to
its bid. Up to 5 percent above or below the bid, plans keep all profits and losses. Between 5 and 10 percent above or below the bid, plan share half of savings and losses with Medicare. Above or below 10 percent, Medicare covers 80 percent of the risk and plans are at risk of 20 percent.
Future and Developing Value-Based Payment Programs
Home Health Value-Based Purchasing
In its calendar year 2016 Home Health Prospective Payment Final Rule, CMS proposed a home health value-based purchasing model that would subject home health agencies to upward or downward payment adjustments based on quality and efficiency measures (HHS, 2015). CMS randomly selected nine states (Arizona, Florida, Iowa, Maryland, Massachusetts, Nebraska, North Carolina, Tennessee, and Washington) representing nine regional groups for model participation beginning January 1, 2016. CMS requires all home health agencies within the states to participate. Beginning in 2016, CMS assesses and reports performance. CMS proposed implementing payment adjustments beginning in 2018, with proposed maximum adjustments increasing incrementally from 3 percent in 2018, 5 percent in 2019, 6 percent in 2020, 7 percent in 2021, and 8 percent in 2022 (HHS, 2015). CMS proposed payment adjustment scoring using both achievement and improvement scores (HHS, 2015).
Skilled Nursing Facility Value-Based Purchasing
The Protecting Access to Medicare Act of 2014 authorizes a skilled nursing facility incentive program and also specifies details about quality measures, scoring performance, the performance standards and periods, and public reporting (CMS, n.d.-c). Beginning in 2016, CMS will measure performance on the Skilled Nursing Facility 30-Day All-Cause Readmission Measure (CMS, n.d.-c). CMS will also send skilled nursing facilities feedback reports on their performance beginning in the summer of 2016 and quarterly thereafter, and CMS will publish post-performance data publicly on Nursing Home Compare starting in October 2016. Beginning in 2018 (FY 2019), Skilled Nursing Facilities (SNFs), defined broadly as any institution that primarily provides skilled nursing or rehabilitative services, will receive incentive payments based on the quality of care they provide. CMS submitted a report to Congress detailing their implementation plan and has proposed several incentive design options, including paying for attainment, paying for improvement, and a hybrid attainment and improvement model (HHS, 2012).
The Medicare and the Children’s Health Insurance Program Reauthorization Act
Among other provisions, MACRA streamlines current public reporting programs and incentivizes the development and uptake of VBP models through establishment of a new Merit-Based Incentive Payment System (MIPS) and incentive payments for participation in certain APMs (Conway et al., 2015). In 2019, quality incentive programs including the Physician Value-Based Modifier, the Physician Quality Reporting System, and the Medicare Electronic Health Records Incentive Program (also known as the Meaningful Use program) will end. MIPS will combine these separate programs into a single initiative. MIPS requires the Secretary of HHS to develop a composite performance score that combines performance on each of four weighted categories: quality, resource use, meaningful use, and clinical practice improvement activities (CMS, 2015b). Based on this score, providers may receive an upward or downward adjustment, or no adjustment. Maximum adjustments will be 4 percent in 2019, 5 percent in 2020, 7 percent in 2021, and 9 percent from 2022 forward (CMS, 2015b). From 2019 to 2024, the highest performers will also receive an additional payment adjustment. The program is budget neutral, so total upward adjustments must equal total downward adjustments.
MACRA encourages provider participation in APMs through incentive payments. Qualifying participants are excluded from MIPS payment adjustments and instead receive a lump sum equaling 5 percent of the preceding year’s estimated total Part B expenditures (CMS, n.d.-b; Conway et al., 2015). To qualify for these payments, in 2019 and 2020, qualifying participants must have 25 percent of their payments or patients through an eligible payment entity (CMS, n.d.-b). In 2021 and 2022, the threshold increases to 50 percent of payments or patients and in 2023 and subsequent years, the threshold rises to 75 percent. In early 2016, CMS identified 10 APMs, including MSSP (described above) and several innovative models such as Next Generation ACOs and Bundled Payment Care Improvement, described in the following section (CMS, 2016e).
Select Innovative Payment Models
CMMI designs and tests innovative payment and care deliver models. Three such payment models that tie payment to quality and efficiency of care delivered to Medicare beneficiaries, and thus for which accounting for social risk factors may be relevant, are described below.
Bundled Payments for Care Improvement
The Bundled Payments for Care Improvement (BPCI) initiative incentivizes coordinated, efficient, and high quality care across clinicians and care settings by linking payments for clinical services related to a single episode of care (Press et al., 2016). BPCI tests four types of bundled payments (Press et al., 2016). In model 1, the episode of care includes all diagnostic-related groups for the duration of an inpatient stay at an acute care hospital (CMS, 2016a). The first cohort of model 1 began in April 2013 and concluded in March 2016; the remaining participants conclude in December 2016. In models 2, 3, and 4, participating providers choose the episode of care for one or more of 48 conditions as well as the duration of the episode (hospitalization and related readmissions only, hospitalization and postacute care up to 90 days, or postacute care up to 90 days only) (Press et al., 2016). The elected duration determines the model (CMS, 2016a). Although the payment methodology varies somewhat by model, in each, Medicare compares actual costs to a target bundled rate. Providers whose actual costs are under the target can keep savings, while those with costs over the target must compensate Medicare for the difference (Froimson et al., 2013). As of April 1, 2016, more than 1,500 health care providers were participating in BCPI Phase 2, including 681 skilled nursing facilities, 385 acute care hospitals, 283 physician group practices, 99 home health agencies, 9 inpatient rehabilitation facilities, and 1 long-term care hospital (CMS, 2016a).
Next Generation ACOs Next Generation ACOs build on experience from earlier ACO models such as the MSSP described above and Pioneer ACOs.1 These Next Generation ACOs offer a range of payment mechanisms from fee-for-service to capitation (referred to in the model as all-inclusive population-based payments), which allow participating organizations to take on substantially more financial risk—up to 100 percent (CMS, 2016d; HHS, 2016). This provides the participating organizations with the potential to share a greater proportion of savings, although this also puts the organiza-
1 The Pioneer ACO Model is a CMMI accountable care initiative with higher levels of savings and risk compared to MSSP, which also allows eligible participants to elect to move from fee-for-service to a population-based payment model (prospective per beneficiary per month payment) in the third year of participation (CMS, 2016d). CMS also requires Pioneer ACOs to cover at least 15,000 beneficiaries (5,000 for rural ACOs) and encourages them to negotiate VBP arrangements with other payers by the second year of participation. See https://innovation.cms.gov/initiatives/Pioneer-ACO-Model/Pioneer-ACO-FAQs.html (accessed May 19, 2016).
tions at greater financial risk for losses. Also unlike previous models, CMS will calculate a prospective (rather than retrospective) benchmark, and participating organizations receive a prospective budget (i.e., before the performance year) (HHS, 2016). The Next Generation ACOs also include a set of delivery system tools to enhance beneficiary engagement. These include potential reward payments to beneficiaries for receiving care through the ACO and affiliated providers and increased access to care coordination services, such as access to telehealth, postdischarge home visits, and skilled nursing facility services, among others (CMS, 2016d). In 2016, 21 organizations are participating in the model (HHS, 2016).
Comprehensive Primary Care Plus The Comprehensive Primary Care Plus (CPC+) model is a regionally based, advanced primary care medical home model comprising multipayer payment reform and delivery system reforms that build on the existing Comprehensive Primary Care initiative (Ayanian and Hamel, 2016; Dale et al., 2016). With respect to care delivery, participants meet a series of incremental requirements to achieve five functions: access and continuity, care management stratified by patient risk, preventive care and planned care for chronic conditions, patient and caregiver engagement, and coordinated and comprehensive care (CMS, 2016c; Sessums et al., 2016). For practices with more experience delivering advanced primary care, CPC+ has a separate track that requires these providers to provide additional services, such as identifying psychosocial needs of patients with complex needs and providing resources and other supports to meet those needs (Sessums et al., 2016). To facilitate this care delivery, CPC+ aligns payment, claims and feedback provision, and quality measures across commercial and public payers in a given region (Sessums et al., 2016). CPC+ also includes several payment mechanisms including a prospective monthly care management fee, performance-based incentive payments, and, for track 2 (experienced) models, an upfront comprehensive primary care payment for evaluation and management (CMS, 2016c). CMS also aims to aggregate cost and utility data across all payers as well as to convene health information technology vendors to facilitate providing data and tools to participants to inform practice redesign and quality improvement (Sessums et al., 2016). CMS expects to select up to 5,000 practices in 20 regions to begin a 5-year model in January 2017 (CMS, 2016b; Sessums et al., 2016).
Although adjustment for social risk factors could have important benefits, any proposal to account for social risk factors in Medicare payment programs will entail its own advantages and disadvantages that need to be
carefully considered. Current Medicare VBP programs that do not account for social risk factors could result in tangible harms to populations with social risk factors and to the providers who serve them (Friedberg et al., 2010; Gilman et al., 2015; Joynt and Jha, 2013). Patients with social risk factors may require more intensive care and greater costs to overcome barriers they face to achieving the same health outcomes as patients with fewer risks. By not accounting for the greater cost of caring for these patients, existing payment systems may contribute to disparities in access and quality of care (Joynt and Rosenthal, 2012; Woolhandler and Himmelstein, 2015).
Under current APMs, physicians and hospitals that disproportionately care for socially at-risk populations receive payments that may undervalue the resources and effort required to provide high-quality care for these individuals (Chien et al., 2007). Similarly, it may be difficult for even dedicated providers who disproportionately care for socially at-risk populations (including safety-net providers, minority-serving institutions, critical access hospitals, and community health centers) to gain (or not lose) revenue under quality incentive schemes (e.g., pay-for-performance), because it can be more costly to help patients with social risk factors achieve quality benchmarks (Joynt et al., 2014).
When providers who disproportionately serve patients with social risk factors lose revenue, quality of care and access for patients could decline (Chien et al., 2007; Cunningham et al., 2008; Grealy, 2014; Ryan, 2013; Volpp et al., 2006). In the short term, these providers may be required to limit staffing or reduce the variety of services provided to patients with social risk factors (Lindrooth et al., 2006). Over the longer term, revenue shortfalls could contribute to financial distress for providers and to the closure of hospitals, clinics, and physician offices in underserved communities (Kane et al., 2012; Lipstein and Dunagan, 2014). These closures, in turn, would make it difficult for patients with social risk factors to access care in their communities, contributing to delays in use of clinically beneficial treatments (Bazzoli et al., 2012; Buchmueller et al., 2006; Walker et al., 2011).
Similarly, payments to insurance plans that do not account for social risk factors could lead insurers to avoid covering underserved populations. For example, as described in the previous section, Medicare Advantage plans receive a risk-adjusted annual capitated rate and receive bonuses for achieving quality benchmarks based on performance measures risk adjusted for clinical, behavioral, and some social risk factors under the Five-Star Quality Rating System. However, even after adjustment, plans that have a large number of individuals with social risk factors find it more difficult to achieve the benchmarks because these individuals have lower adherence and greater difficulty managing illnesses, making it difficult for the insurer to obtain star ratings comparable to other plans (Young et al., 2014). In response, plans could decide to withdraw from insurance markets in which
populations with social risk factors reside (Chien et al., 2007). When insurers leave markets, this has the effect of diminishing choice and competition, leading to potentially higher premiums for beneficiaries (Gaynor and Town, 2011). VBP could also reduce incentives for hospitals to care for socially at-risk populations if penalties are larger than hospitals’ margins to care for these patients (Joynt and Jha, 2013).
Finally, under the status quo, plans and providers that serve mixed populations that include individuals with both low and high levels of social risk factors may find that they have incentives to improve care only for patients with low levels of social risk factors (Casalino et al., 2007). For example, to reach a target rate for hemoglobin A1C control among diabetic patients, a physician practice may find it is less costly to focus on improving care for patients that have access to better quality diets and who are more easily able to attend regular checkup visits. As a result, those patients with greater social risk factors may not receive effective interventions available to patients with better social and economic resources.
In summary, the status quo has disadvantages that include incentives for providers and insurers to avoid serving patients with social risk factors, underpayment to providers who disproportionately serve socially at-risk populations, and underinvestment in quality of care. While proposals that do account for social risk factors would likely diminish these harms, there are also some potential ways in which accounting for social risk factors could incrementally introduce new harms.
First, incentives to improve the quality of care for patients with social risk factors could be diluted under some approaches that adjust for social risk factors. Setting lower benchmarks for patients with social risk factors relative to those without social risk factors can, in some circumstances, diminish provider incentives to exceed the established benchmarks. This could be problematic in settings where providers are capable of delivering the same standard of care for patients with and without social risk factors.
Second, any method for accounting for social risk factors that sets lower-quality improvement benchmarks for patients with social risk factors or otherwise holds providers and insurers to different standards for these populations can have a negative symbolic value. While certainly not intended, these adjustments may create the perception that patients with social risk factors are entitled to a lower quality of care. These perceptions are particularly acute because of a well-documented history of exclusion and inequitable treatment in health care settings of racial and ethnic minorities and low-income populations (HealthyPeople.gov, 2016; IOM, 2003). Even if these concerns are unfounded, perceptions of inequitable treatment can further erode trust in the health care system among patients with social risk factors.
Third, adjustment for social risk factors will not necessarily help patients find providers who will deliver the best quality of care for them. For example, star ratings are intended to guide patients to providers who provide an excellent average quality of care to patients. When only one single summary measure of quality is created for a provider or plan (such as a star rating), unadjusted results convey information about providers’ unadjusted performance for their whole patient population, which varies across reporting units. Adding social risk factors to existing risk-adjustment methods may provide more accurate information about the relative performance across reporting units if they were faced with an average patient. However, neither summary score provides information about which provider is better for a patient based on his or her level of social risk factors unless all providers are equally good or bad with all patients. This may be especially true when patients with social risk factors comprise a small number of patients in a practice. Only stratification by social risk factors will reveal such insights.
Finally, some methods of accounting for social risk factors could obscure differences due to poor quality care, such as failure to tailor care or provide culturally competent care, which may result in uneven relative allocation of rewards relative to effort.
Conclusion 4: It is possible to improve on the status quo with regard to the effect of value-based payment on patients with social risk factors. However, it is also important to minimize potential harms to these patients and to monitor the effect of any specific approach to accounting for social risk factors to ensure the absence of any unanticipated adverse effects on health disparities.
Any approach chosen to account for social risk factors should aim to minimize the potential harms described in the previous section. In particular, accounting for social risk factors, especially adjustment, is not intended to obscure disparities that do exist. Disparities should be brought to light, and the payment system should be sure to include sufficient incentive for quality improvement for both socially at-risk populations and for patients overall. Hence, the use of these factors in quality measurement and payment schemes should not disincentivize providers from doing all they can to overcome the influence of these factors on outcomes. Incentivizing providers to find strategies to overcome barriers to better outcomes in socially at-risk populations is critical to the reduction of health disparities. At the same time, incentivizing quality improvement and efficient care for all patients is
an important goal of including social risk factors in VBP. Finally, achieving good outcomes (or improving outcomes over time) may be more difficult for providers caring for patients with social risk factors precisely because the influence of some social risk factors on health care outcomes is beyond provider control. Similarly, achieving good outcomes may also be more costly for providers caring for patients with social risk factors owing to additional costs required to tailor care appropriately or because these patients have fewer resources outside the health systems available to contribute to outcomes. Accounting for these factors in performance measurement and payment mechanisms under VBP would level the playing field by ensuring that provider compensation is commensurate with the true quality of care they deliver (i.e., fair and accurate). Thus, the committee’s review of methods to account for social risk factors in Medicare VBP programs takes as the point of departure that the goal of Medicare payment and reporting systems are reducing disparities in health care access, affordability, quality, and outcomes; quality improvement and efficient care delivery for all patients; fair and accurate public reporting; and compensating providers for the services they provide.
Observed differences in quality by social risk factors may reflect a combination of drivers, including
- mechanisms that occur during the patient–provider encounter (e.g., inadequate tailoring of care to account for social risk factors, discrimination and bias);
- provider characteristics such as having fewer financial resources (e.g., lower margins, historically lower reimbursement rates) and having fewer and lower-quality clinical/health care resources (e.g., fewer technological resources and lower information technology capacity, fewer and less qualified clinicians);
- differences in patient preferences; and
- barriers to access and financial constraints for disadvantaged persons (NASEM, 2016b).
In practice these mechanisms may occur simultaneously and also interact; it is difficult if not impossible to decompose observed differences into these components quantitatively. The committee therefore proposes approaches that do not require disentangling the mechanisms of these multiple pathways for social risk factors.2 The fact that some units (e.g., providers)
2 These mechanisms describe direct effects of social risk factors on performance indicators used in VBP. Some effects may be mediated by health status and therefore at least partly accounted for in clinical adjustments. At the same time, social risk factors may also capture unmeasured differences in clinical risk and are likely to have independent effects on performance
do well with socially at-risk populations does not imply that it is equally easy to do so on average, and such population differences may also affect the relationship between provider quality and observed provider scores. The standard for taking such factors into account should not be that it is impossible to provide optimal care, but that it is more difficult on average. Taking such factors into account need not “adjust away” disparities. Lower levels of performance for any group should not be considered sufficient or qualify a provider to receive maximum rewards. However, a provider that does not achieve performance on par with top performers (i.e., optimal care) could still be eligible for some reward because, for example, it improved substantially relative to its own benchmark.
Conclusion 5: Characteristics of a public reporting and payment system that could accomplish the goals of reducing disparities in access, quality, and outcomes; quality improvement and efficient care delivery for all patients; fair and accurate public reporting; and compensating providers fairly include
- Transparency and accountability for overall performance and performance with respect to socially at-risk members of the population;
- Accurate performance measurement—with high reliability and without bias (systematic error) related to differences in populations served;
- Incentives for improvement overall and for socially at-risk groups, both within reporting units (i.e., the provider setting that is being evaluated—hospitals, health plans, etc.) and between reporting units.
The committee reviewed literature on a range of methods to account for social risk factors in public reporting and payment systems for which inclusion of social risk factors may be appropriate, with the aim to be more inclusive. These methods are described briefly in the following text and in more detail in Table C4-1.
- Finding: The committee identified four categories—(A) public reporting; (B) adjustment of performance measure scores; (C) direct adjustment of payments; and (D) restructuring payment incentive design—encompassing ten methods to account for social risk
indicators used in VBP. Evidence described in Appendix C3 includes documented associations of social risk factors on performance indicators used in VBP above and beyond effects of social risk factors on health status. The committee’s approaches do not require disentangling pathways mediated through health status.
factors in that could be used to address policy goals of reducing disparities in access, quality, and outcomes; quality improvement and efficient care delivery for all patients; fair and accurate public reporting; and compensating providers fairly.
Public reporting seeks to make overall quality visible—to consumers, providers, payers, and regulators (IOM, 2007). It may lead to quality improvement via reputation incentives, and particularly when linked to behavioral nudges, by increasing market share (i.e., influencing choice of provider) for higher-quality reporting units (IOM, 2007). Public reporting methods that could be used to account for social risk factors include (1) stratification by patient characteristics within reporting units (i.e., for population subgroups by social risk factors) and (2) stratification by reporting unit characteristics (i.e., comparisons to peers, such as those with a similar share of low-income patients). (Methods are described in more detail in Table C4-1.) If publicly reported performance is stratified by indicators of social risk, public reporting can also be important for monitoring disparities, particularly when applied together with risk and/or payment adjustment.
Adjusting performance measure scores seeks to “level the playing field,” to estimate true reporting-unit quality—that which would occur if all units had the population average patient. As described in Appendix C2, social risk factors can be considered confounders of true performance if they are beyond provider control and unevenly distributed across units and thereby distort (bias) comparisons. Adjustment is a means to account for social risk factors statistically in an effort to more accurately measure true performance. Methods include
- risk adjustment for mean within-provider differences (e.g., to account for the average disparity between population subgroups with high and low levels of social risk factors);
- risk adjustment of performance data for within- and between-provider differences (e.g., to account for all patient-level differences in performance associated with social risk factors); and
- adding quality measures tailored (and only meaningful) to socially at-risk groups in addition to overall performance. Applicable statistical methods may include linear or logistic regression with or without mixed effects, doubly robust estimation, and direct and indirect standardization (Elliott et al., 2001, 2009a,b; Lyratzopoulos et al., 2012; Zaslavsky and Jha, 2015; Zaslavsky et al., 2001).
Any effects of risk adjustment on payment are indirect and require consideration of the particular form of the payment function.
TABLE C4-1 Methods to Account for Social Risk Factors in Medicare VBP Programs
|A. Public Reporting Methods
Stratification by itself does not inﬂuence payment, but reporting may inﬂuence choice or provider, leverage reputational incentives, and/or be important for monitoring disparities in conjunction with methods B and/or C.
|1. Stratification by patient characteristics within reporting unitsa||Present performance data for population subgroups.b||Social risk factors can be represented by discrete strata (e.g., high and low income, Black and white patients).
Works best if there are only a few key dimensions and few interactions among dimensions or reporting will become complex.
Requires sufficient sample sizes.
|2. Stratification by reporting unit characteristicsc (e.g., safety-net hospitals)||Present performance data for subsets of reporting units (e.g., reporting quality performance separately for physicians located in Health Provider Shortage Areas).||Requires a meaningful method of classifying providers, hospitals, or health plans according to the population they serve.|
|Potential Advantages||Potential Disadvantages||Special Considerations for Cost|
|All data can be seen.
Disparities can be monitored.
High and low performers for at-risk groups visible.
|Too many dimensions or strata may lead to ambiguity and information overload.||Interpretation of cost differences for at-risk population complicated by demand effects: patients with higher income consume more services all else equal either due to income effects, price effects for un-/underinsured or access constraints for those receiving Medicaid. Likewise, other characteristics of patients with more social risk factors such as low educational attainment, may cause underuse of services relative to need.|
|Comparisons made within peer groups: different types of providers and health plans have different capabilities for attaining and improving performance owing to patient differences and resource constraints.
Stratification at the unit level requires only unit-level data (characterization of the unit rather than each patient contributing to performance data).
|Does not illuminate within reporting unit differences (for example, differences due to quality compared to those due to patient mix), which might also be important.
Reporting units could try to manipulate their patient mix in order to change strata. “Notch” effects are possible, and units near a notch may especially distort their behavior. Finally, the correlation between the mix of social risk factors and resource constraints may be limited.
|B. Adjustment of performance measure scoresd
Seeks to improve measurement and estimate provider quality under similar patient populations. Any effects on payment are indirect and one must consider the particular form of the payment function.
|1. Risk adjustment for mean within-provider differencese||Statistical methods are used to account for (remove) the average disparity between high- and low-social risk factor groups.||Social risk factors can be measured at the patient level.
Mean within-provider differences represent what is typically achievable.
|2. Risk adjustment of performance data for within- and between-provider differences||Statistical methods are used to account for (remove) all differences in performance associated with social risk factors at the patient level.||Social risk factors can be measured at the patient level.
Providers have little control over either social risk factors or their impact on performance.
There is no true difference in the quality of providers seen by those with and without social risk factors.
|3. Add quality measures for performance for at-risk groups in addition to overall measure relates to A1||Tailors performance data to target populations through measures that are only meaningful for target populations.||Adequate sample sizes in at-risk groups.|
|Potential Advantages||Potential Disadvantages||Special Considerations for Cost|
|Scores improve with improvement in care to any group.
Avoids adjusting for “true” between-differences in quality.
May reduce disincentives to avoid patients with social risk factors compared to no adjustment.
Better quality measurement may better focus “nudges” into better plans to reduce between-plan SES disparities.
|Effects on payment may be limited. Depending on payment functions, could reduce incentives to improve.
Under-adjusts if between-provider differences are caused by patient characteristics.
Does not make disparities visible without also using method (A).
Does not allow unit-level adjustors.
|Adjustment will typically increase estimated costs for at risk populations. Unclear interpretation of disparity.|
|May capture full effects of social risk factors on quality measures if caring for at-risk patients reduces quality via resources or some similar mechanism.
Allows unit-level adjustors.
|Depending on payment functions, could reduce incentives to improvement.
Does not make disparities visible without also using method (A).
|Adjustment will typically increase estimated costs for at risk populations. Unclear interpretation of resulting performance estimates.|
|Fewer model assumptions than adjusted models; direct measure of performance for at-risk groups.||Will not be available for all units—how to pay then? How much to pay relative to overall score?
May be strongly correlated with overall performance.
|C. Direct Adjustment of Payments
By themselves do not affect performance measure scores
|1. Risk adjustment in payment formula without adjusting measured performancef
Approach may be based on B(1) or B(2) methods; magnitude might be calibrated to less or more than the indirect effect via measurement.
|Alter the payment threshold or increment of bonus/penalty based on mix of social risk factors —specifically increase the return on investment for improving performance for at-risk populations.||Social risk factors (social risk factors) can be measured at the patient level.
Social risk factors result in differential cost of improvement that needs to be compensated for equal “incentive” or the value of improvement is greater for at-risk populations.
|2. Stratification of benchmarks used for paymentg||First determine payments according to any pay-for-performance approach, including the current one. Second, select reporting unit strata based on social risk factors. Third, multiply payments by factors that result in equal mean payouts for each stratum (as in A2).||Meaningful strata of social risk factors exist (high and low income, safety-net versus other, academic medical center versus community). Not too many strata. Social risk factors and their consequences are beyond the control of provider/health plan.|
|Potential Advantages||Potential Disadvantages||Special Considerations for Cost|
|Resources are allocated in a manner more favorable to institutions serving at-risk populations.
Improvement in care for at-risk populations is differentially rewarded. Magnitude of adjustment can be directly controlled.
|Does not improve the accuracy of publicly reported quality measures.
Providers/health plans can be rewarded despite poor outcomes/performance.
|Adjustment will typically reduce payments associated with at-risk populations through bundled, global or shared savings mechanisms. Such adjustment would freeze in place patterns of use known to be reﬂective of underuse at least for some services.|
|Comparisons are possible across a wider range—stretch goals may be more apparent, while ensuring that resource allocation does not punish institutions that serve at-risk groups. Incentives may strengthen for at-risk groups.||For payment if benchmarks are stratified, the number of social risk factor dimensions would be limited.
Incentives may weaken for groups not at risk.
Reporting units could try to manipulate their patient mix in order to change strata. “Notch” effects are possible, and units near a notch may especially distort their behavior. Finally, the correlation between the mix of social risk factors and resource constraints may be limited.
|Adjustment will typically increase payments associated with at-risk populations.|
|C. Restructure Payment Incentive Design
Measures of social risk factors not explicitly used but implicitly accounted for
|1. Pay for improvement relative to own benchmark (to a greater extent or exclusively), including “growth models”h||Payment formula is based in part or wholly based on percentage improvement relative to prior period rather than absolute level of performance.||Good measurement of prior performance of well-defined unit (how to handle mergers, etc.).|
|Potential Advantages||Potential Disadvantages||Special Considerations for Cost|
|Strongly controls for unobserved social risk factors. Clear incentives for improvement.||Does not explicitly recognize absolute level of performance.
No guarantee that providers disproportionately serving socially at-risk populations are more likely to achieve improvement compared to achievement targets (though they do in some work in progress).
Do you ratchet-improve versus best ever or do you allow rewards for alternation?
How to handle ceiling effects (high-performing units that may have little room to improve during the performance period) is unclear.
Depends on functional form of both the payment formula and the effort required to improve from different baselines.
Measuring improvement is noisier than comparing performance to a fixed (achievement) benchmark—particularly for rare events like mortality.
|Rewards units that have high baseline costs where improvement is more feasible.|
|2. Downweight social risk factor-sensitive measures in payment||In payment formula measures can be weighted differentially to alter their importance to providers.||Some quality measures – like readmissions and intermediate outcomes – are strongly associated with social risk factors.
Some performance measures—like inpatient safety measures—can be expected to have little relationship to social risk factors.
Ideally, the measures not affected by social risk factors signal high quality/value overall.
|3. Add bonus for low disparitiesi||In addition to other rewards and penalties explicitly measure and reward the magnitude of difference between groups.||Adequate sample sizes in low-risk and high-risk groups.|
NOTE: SES = socioeconomic status.
b See, for example, https://www.cms.gov/About-CMS/Agency-Information/OMH/Downloads/National-Level-Results.pdf (accessed June 9, 2016).
d Examples of applicable statistical methods: linear or logistic regression with or without mixed effects, doubly robust estimation, indirect and direct standardization (Elliott et al., 2001, 2009a,b; Lyratzopoulos et al., 2012; Zaslavsky and Jha, 2015; Zaslavsky et al., 2001).
h The Medicare Advantage and Hospital Value-Based Purchasing payment arrangements currently include a measure of improvement (or failure rate reduction) in their payment formula
|Potential Advantages||Potential Disadvantages||Special Considerations for Cost|
|May reallocate resources appropriately without providing incentives to cherry pick patients.||Does not improve quality measurement.
Social risk factor-sensitive measures may be important dimensions of quality that are not picked up by social risk factor-insensitive ones.
|Directly rewards equity.||Requires other components to address improvement or overall level of performance.
Disparity could be reduced by making better off group worse unless steps are taken to avoid this.
|Might reward cost increases for at-risk populations that may or may not be warranted.|
(CMS, 2012). The committee lists this approach here to acknowledge the benefits and risks of such an approach vis à vis accounting for social risk factors in other Medicare payment systems. The committee also notes that increasing the weight given to improvement or altering the particular approach to scoring improvement (e.g., in terms of absolute improvement versus failure rate reduction for Hospital Value-Based Purchasing) is a method open to CMS for obtaining a different balance of incentives in programs that currently incorporate improvement (Casalino et al., 2007; Rosenthal et al., 2004).
i In simple linear scoring, this is equivalent to giving greater than proportional weight to performance with the high-risk (H) group relative to the low-risk (L) group. If FL, FH is fraction in the groups (FL + FH = 1), Y = mean performance in a group (YL, YH), then the proportionally weighted score is FL*YL + FH*YH with a linear penalty on disparity YL – YH, the score is FL*YL + FH*YH – C*(YL – YH) = (FL – C)*YL + (FH + C)*YH (Blustein et al., 2011; Casalino et al., 2007).
VBP incorporates explicit or implicit (as in the case of bundled or global payment including shared savings) rewards or penalties based on performance on quality and/or cost of care. This can be achieved through three underlying conceptual approaches. First, payers could pay for more to those that are doing a better job in the measurement period (i.e., pay for achievement). Second, payers could pay for the mix of patients the reporting unit treats, that is, pay more to those that treat greater numbers of socially at-risk patients under the assumption that they simply need more resources. This approach lacks incentives to improve unless some other system for accountability is superimposed. Third, payers could pay for improvement, that is, pay more to those who improve to a greater degree.
The committee also expands upon how VBP could incorporate measures of social risk factors. Payments could be directly adjusted using social risk factors, or incentive design could be restructured. Direct adjustments of payment explicitly use measures of social risk factors, but by themselves do not affect performance measure scores. Methods include (1) risk adjustment in the payment formula without adjusting measured performance (i.e., applying a different payment threshold or increment for rewards or sanctions based on the reporting unit’s mix of social risk factors), or (2) stratification of benchmarks used for payment (i.e., applying payment multipliers to reporting-unit strata based on social risk factors). Restructured payment incentive designs do not explicitly use measures of social risk factors, but implicitly account for social risk factors. Methods include (1) paying for improvement (rather than attainment), (2) downweighting social-risk factor-sensitive measures in payment (i.e., weighting measures differentially in the payment formula to alter their importance to providers), or (3) adding a bonus for achieving low disparities.
Table C4-1 summarizes the four categories of methods that could be used individually or in combination to account for social risk factors in Medicare value-based purchasing programs. The table also lists the possible methods within each category described briefly above and describes them in more detail along with prerequisites or optimal conditions for implementation, as well as potential advantages and disadvantages. Because considerations for cost performance may differ compared to quality performance, the table also notes special concerns for cost-related incentive programs, including bundled and global payment.
In many cases, methods from multiple categories can be used together. In some cases, multiple methods from a single category can be used in combination. In this respect, each approach has some advantages and disadvantages and a combination of approaches may yield a better result than
any one method alone. As described in the previous section, the committee underscores that the benefits and harms of any single or composite method of accounting for social risk factors should be assessed in reference to the status quo or some other feasible alternative rather than a perfect world in which social risk factors do not confound efforts to improve the quality and efficiency of health care delivery (referred to by some as a “full information” scenario). As illustrations, Table C4-2 compares the potential harms of accounting for social risk factors relative to the status quo. Box C4-1 describes a hypothetical example of stratification by social risk factors and a simple risk adjustment of a performance measure for mean within-provider differences between groups with high and low levels of social risk factors. This example also describes potential advantages and disadvantages of this approach relative to the status quo (no adjustment for social risk factors).
Conclusion 6: To achieve goals of reducing disparities in access, quality, and outcomes; quality improvement and efficient care delivery for all patients; fair and accurate public reporting; and compensating providers fairly, a combination of reporting and accounting in both measures and payment are needed.
Considerations around the trade-offs of various methods of accounting for social risk factors are different for cost-related performance compared to quality performance. Costs in the context of VBP can refer to the costs of improving quality or achieving good outcomes for socially at-risk patient or to the cost of care billed to a payer. As noted earlier, because achiev-
TABLE C4-2 Potential Harms of Accounting for Social Risk Factors Compared to the Status Quo
|Status Quo||Accounting for Social Risk Factors|
|Patient dumping/avoidance||Reduces this risk relative to the status quo|
|Unfair to providers disproportionately serving socially at-risk populations (if factors beyond provider control—and/or the cost of improvement is higher for populations with social risk factors—cause poor performance)||Unfair to providers who provide high-quality care to all patients if truly poor quality causes poorer performance for socially at-risk patients|
|Will reduce quality and access for socially at-risk populations||Reduces this risk relative to the status quo|
|Incentives to improve care might favor focusing on patients with few social risk factors if they are easier to improve||Same unless payment is adjusted upward for socially at-risk populations|
ing high performance on performance indicators used in VBP may require greater investments on the part of health care providers and health plans to overcome barriers socially at-risk populations face, costs to achieve good outcomes and improve care quality for socially at-risk populations are likely to be higher than costs to achieve the same outcomes and improve care quality for more advantaged patients. Because at least some of these costs will be outside of the services that can be billed to payers like CMS, as described in an earlier section, a potential harm of not accounting for social risk factors in a VBP environment is that this increased cost may be a disincentive to care for socially at-risk populations. On the other hand, lower resource use observed in billed costs of care may reflect unmet need or barriers to access rather than the absence of waste. Thus, lower cost is not always better; whereas, higher quality is always better.
Conclusion 7: Strategies to account for social risk factors for measures of cost and efficiency may differ from strategies for quality measurement, because observed lower resource use may reflect unmet need rather than the absence of waste, and thus lower cost is not always better, while higher quality is always better.
As described earlier in the chapter, accounting for social risk factors in Medicare value-based purchasing programs is intended to achieve a balance between incentives for reducing disparities in access, quality, and outcomes; quality improvement and efficient care delivery for all patients; fair and accurate public reporting; and compensating providers fairly. Both the status quo and any new approach to accounting for social risk factors will have uncertain trade-offs in terms of these goals—many unknowable factors including provider and patient beliefs and behavioral responses will affect the results that any new system yields. Monitoring data on a variety of indicators will facilitate assessment of the effects of existing and new programs on potential unintended adverse effects. Such indicators might include enrollment (for health plans), patient complaints, access to and quality of care for socially at-risk populations, and the financial sustainability of providers disproportionately caring for socially at-risk populations.
Conclusion 8: Any specific approach to accounting for social risk factors in Medicare quality and payment programs requires continuous monitoring with respect to the goals of reducing disparities in access, quality, and outcomes; quality improvement and efficient care delivery
for all patients; fair and accurate public reporting; and compensating providers fairly.
Finally, because behavioral and other responses to new systems may change the balance of risks and benefits over time, to take into account these behavioral and other responses, the specific approach to accounting for social risk factors may need to be reassessed.
The committee notes that it is not within its statement of task to recommend whether social risk factors should be accounted for in VBP or how; that decision sits elsewhere. The committee hopes that the conclusions in this report help CMS and the Secretary of HHS make that important decision. In the next report, the committee tackles the question of how to gather the data that could be used to account for social risk factors in Medicare VBP.
Ayanian, J. Z., and M. B. Hamel. 2016. Transforming primary care—we get what we pay for. New England Journal of Medicine 374(15):457.
Bazzoli, G. J., W. Lee, H. M. Hsieh, and L. R. Mobley. 2012. The effects of safety net hospital closures and conversions on patient travel distance to hospital services. Health Services Research 47(1Pt1):129-150.
Blustein, J., J. S. Weissman, A. M. Ryan, T. Doran, and R. Hasnain-Wynia. 2011. Analysis raises questions on whether pay-for-performance in medicaid can efficiently reduce racial and ethnic disparities. Health Affairs (Millwood) 30(6):1165-1175.
Buchmueller, T. C., M. Jacobson, and C. Wold. 2006. How far to the hospital?: The effect of hospital closures on access to care. Journal of Health Economics 25(4):740-761.
Burwell, S. M. 2015. Setting value-based payment goals—HHS efforts to improve U.S. health care. New England Journal of Medicine 372(10):897-899.
Casalino, L. P., A. Elster, A. Eisenberg, E. Lewis, J. Montgomery, and D. Ramos. 2007. Will pay-for-performance and quality reporting affect health care disparities? Health Affairs (Millwood) 26(3):w405-w414.
Chien, A. T., M. H. Chin, A. M. Davis, and L. P. Casalino. 2007. Pay for performance, public reporting, and racial disparities in health care: How are programs being designed? Medical Care Research and Review 64(5 Suppl):283s-304s.
CMS (Centers for Medicare & Medicaid Services). 2012. Frequently asked questions: Hospital value-based purchasing system. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/hospital-value-based-purchasing/downloads/HVBPFAQ022812.pdf (accessed November 2, 2015).
CMS. 2014a. End-Stage Renal Disease Quality Incentive Program payment year 2016 program details. https://www.cms.gov/medicare/quality-initiatives-patient-assessment-instruments/esrdqip/downloads/qip-details-py16.pdf (accessed October 26, 2015).
CMS. 2014b. Readmissions reduction program. https://www.cms.gov/medicare/medicare-fee-for-service-payment/acuteinpatientpps/readmissions-reduction-program.html (accessed October 26, 2015).
CMS. 2015a. End-Stage Renal Disease Quality Incentive Program frequently asked questions. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/ESRDQIP/Downloads/ESRDQIPFrequentlyAskedQuestions.pdf (accessed October 26, 2015).
CMS. 2015b. The Medicare Access & CHIP Reauthorization Act of 2015: Path to value. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/Value-Based-Programs/MACRA-MIPS-and-APMs/MACRA-LAN-PPT.pdf (accessed April 19, 2016).
CMS. 2015c. Medicare program; contract year 2016 policy and technical changes to the medicare advantage and the medicare prescription drug benefit programs. Final rule. Federal Register 80(29):7911.
CMS. 2015d. Shared Savings Program. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/sharedsavingsprogram/index.html (accessed October 26, 2015).
CMS. 2016a. Bundled Payments for Care Improvement initiative (BPCI). https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2016-Fact-sheets-items/2016-04-18.html?DLPage=1&DLEntries=10&DLSort=0&DLSortDir=descending (accessed April 18, 2016).
CMS. 2016b. Comprehensive Primary Care Plus. https://innovation.cms.gov/initiatives/comprehensive-primary-care-plus (accessed April 18, 2016).
CMS. 2016c. Comprehensive Primary Care Plus (CPC+) fact sheet. https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2016-Fact-sheets-items/2016-04-11.html (accessed April 18, 2016).
CMS. 2016d. Next generation ACO model: Frequently asked questions. https://innovation.cms.gov/Files/x/nextgenacofaq.pdf (accessed April 20, 2016).
CMS. 2016e. Overview of select alternative payment models. https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2016-Fact-sheets-items/2016-03-03.html (accessed April 18, 2016).
CMS. n.d.-a. End-Stage Renal Disease Quality Incentive Program summary: Payment years 2014-2018. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/ESRDQIP/Downloads/ESRDQIPSummaryPaymentYears2014-2018.pdf (accessed October 26, 2015).
CMS. n.d.-b. MACRA RFI posting “RFI on physician payment reform” (CMS-3321-NC) external FAQ. https://innovation.cms.gov/Files/x/macra-faq.pdf (accessed April 19, 2016).
CMS. n.d.-c. The Skilled Nursing Facility Value-Based Purchasing program (SNFVBP) https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/Value-Based-Programs/Other-VBPs/SNF-VBP.html (accessed April 19, 2016).
CMS. n.d.-d. Summary of 2015 Physician Value-Based Payment Modifier policies. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PhysicianFeedbackProgram/Downloads/CY2015ValueModifierPolicies.pdf (accessed October 26, 2015).
Conway, P., T. Gronniger, H. Pham, K. Goodrich, A. Bassano, and J. Sharp. 2015. MACRA: New opportunities for Medicare providers through innovative payment systems (updated). Health Affairs Blog [blog on the Internet].
Cunningham, P. J., G. J. Bazzoli, and A. Katz. 2008. Caught in the competitive crossfire: Safety-net providers balance margin and mission in a profit-driven health care market. Health Affairs (Millwood) 27(5):w374-w382.
Dale, S. B., A. Ghosh, D. N. Peikes, T. J. Day, F. B. Yoon, E. F. Taylor, K. Swankoski, A. S. O’Malley, P. H. Conway, R. Rajkumar, M. J. Press, L. Sessums, and R. Brown. 2016. Two-year costs and quality in the comprehensive primary care initiative. New England Journal of Medicine 374(24):2345-2356.
Damberg, C. L., M. N. Elliott, and B. A. Ewing. 2015. Pay-for-performance schemes that use patient and provider categories would reduce payment disparities. Health Affairs (Millwood) 34(1):134-142.
Elliott, M. N., R. Swartz, J. Adams, K. L. Spritzer, and R. D. Hays. 2001. Case-mix adjustment of the national CAHPS benchmarking data 1.0: A violation of model assumptions? Health Services Research 36(3):555-573.
Elliott, M. N., A. M. Haviland, D. E. Kanouse, K. Hambarsoomian, and R. D. Hays. 2009a. Adjusting for subgroup differences in extreme response tendency in ratings of health care: Impact on disparity estimates. Health Services Research 44(2 Pt 1):542-561.
Elliott, M. N., A. M. Zaslavsky, E. Goldstein, W. Lehrman, K. Hambarsoomians, M. K. Beckett, and L. Giordano. 2009b. Effects of survey mode, patient mix, and nonresponse on CAHPS hospital survey scores. Health Services Research 44(2 Pt 1):501-518.
Epstein, A. M. 2015. Accounting for socioeconomic status in Medicare payment programs: ASPE’s work under the IMPACT Act. Presented to the Committee on Accounting for Socioeconomic Status in Medicare Payment Programs, Washingon, DC.
Friedberg, M. W., D. G. Safran, K. Coltin, M. Dresser, and E. C. Schneider. 2010. Paying for performance in primary care: Potential impact on practices and disparities. Health Affairs (Millwood) 29(5):926-932.
Froimson, M. I., A. Rana, R. E. White, A. Marshall, S. F. Schutzer, W. L. Healy, P. Naas, G. Daubert, R. Iorio, and B. Parsley. 2013. Bundled Payments for Care Improvement initiative: The next evolution of payment formulations: Aahks bundled payment task force. The Journal of Arthroplasty 28(8):157-165.
Gaynor, M., and R. J. Town. 2011. Competition in health care markets. Cambridge, MA: National Bureau of Economic Research.
Gilman, M., E. K. Adams, J. M. Hockenberry, A. S. Milstein, I. B. Wilson, and E. R. Becker. 2015. Safety-net hospitals more likely than other hospitals to fare poorly under Medicare’s value-based purchasing. Health Affairs (Millwood) 34(3):398-405.
Grealy, M. R. 2014. Measure under consideration (MUC) comments: Letter to the National Quality Forum: Healthcare Leadership Council, December 5, 2014. http://www.hlc.org/wp-content/uploads/2014/06/HLC_Early-Public-Comment-on-Measures-Under-Consideration.pdf (accessed October 30, 2015).
HealthyPeople.gov. 2016. Foundation health measures: Disparities. https://www.healthypeople.gov/2020/about/foundation-health-measures/Disparities (accessed May 19, 2016).
HHS (Department of Health and Human Services). 2012. Report to Congress: Plan to implement a Medicare Skilled Nursing Facility Value-Based Purchasing program. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/SNFPPS/Downloads/SNF-VBP-RTC.pdf (accessed April 19, 2016).
HHS. 2014. Medicare program; hospital Inpatient Prospective Payment Systems for acute care hospitals and the long-term care hospital Prospective Payment System and fiscal year 2015 rates; quality reporting requirements for specific providers; reasonable compensation equivalents for physician services in excluded hospitals and certain teaching hospitals; provider administrative appeals and judicial review; enforcement provisions for organ transplant centers; and electronic health record (EHR) incentive program. Federal Register 79(163):50094.
HHS. 2015. Medicare and Medicaid programs; CY 2016 home health prospective payment system rate update; home health value-based purchasing model; and home health quality reporting requirements Federal Register 80:39840.
HHS. 2016. New hospitals and health care providers join successful, cutting-edge federal initiative that cuts costs and puts patients at the center of their care. http://www.hhs.gov/about/news/2016/01/11/new-hospitals-and-health-care-providers-join-successful-cutting-edge-federal-initiative.html (accessed April 20, 2016).
IOM (Institute of Medicine). 2003. Unequal treatment: Confronting racial and ethnic disparities in health care (full printed version). Washington, DC: The National Academies Press.
IOM. 2007. Rewarding provider performance: Aligning incentives in medicare (pathways to quality health care series). Washington, DC: The National Academies Press.
Joynt, K. E., and A. K. Jha. 2013. A path forward on medicare readmissions. New England Journal of Medicine 368(13):1175-1177.
Joynt, K. E., and M. B. Rosenthal. 2012. Hospital value-based purchasing: Will Medicare’s new policy exacerbate disparities? Circulation: Cardiovascular Quality and Outcomes 5(2):148-149.
Joynt, K. E., N. Sarma, A. M. Epstein, A. K. Jha, and J. S. Weissman. 2014. Challenges in reducing readmissions: Lessons from leadership and frontline personnel at eight minority-serving hospitals. The Joint Commission Journal on Quality and Patient Safety 40(10):435-435.
Kane, N. M., S. J. Singer, J. R. Clark, K. Eeckloo, and M. Valentine. 2012. Strained local and state government finances among current realities that threaten public hospitals’ profitability. Health Affairs (Millwood) 31(8):1680-1689.
Lindrooth, R. C., G. J. Bazzoli, J. Needleman, and R. Hasnain-Wynia. 2006. The effect of changes in hospital reimbursement on nurse staffing decisions at safety-net and nonsafety-net hospitals. Health Services Research 41(3 Pt 1):701-720.
Lipstein, S. H., and W. C. Dunagan. 2014. The risks of not adjusting performance measures for sociodemographic factors. Annals of Internal Medicine 161(8):594-596.
Lyratzopoulos, G., M. Elliott, J. M. Barbiere, A. Henderson, L. Staetsky, C. Paddison, J. Campbell, and M. Roland. 2012. Understanding ethnic and other socio-demographic differences in patient experience of primary care: Evidence from the English general practice patient survey. BMJ Quality and Safety 21(1):21-29.
Martino, S. C., R. M. Weinick, D. E. Kanouse, J. A. Brown, A. M. Haviland, E. Goldstein, J. L. Adams, K. Hambarsoomian, D. J. Klein, and M. N. Elliott. 2013. Reporting CAHPS and HEDIS data by race/ethnicity for medicare beneficiaries. Health Services Research 48(2 Pt 1):417-434.
McGinnis, J. M. 2016. Income, life expectancy, and community health: Underscoring the opportunity. Journal of the American Medical Association 315(16):1709-1710.
MedPAC (Medicare Payment Advisory Commission). 2013. Report to the Congress. Medicare and the health care delivery system. Washington, DC: MedPAC.
MedPAC. 2014. Explainer: Risk sharing mechanisms in Part D. http://www.medpac.gov/blog/october-2014/october-2014/2014/10/15/explainer-risk-sharing-mechanisms-in-part-d (accessed October 26, 2015).
MedPAC. 2015a. Medicare Advantage program payment system. In Payment basics. Washington, DC: MedPAC.
MedPAC. 2015b. Part D payment system. In Payment basics. Washington, DC: MedPAC.
MLN (Medicare Learning Network). 2013. Hospital Value-Based Purchasing program. https://www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN/MLNProducts/downloads/Hospital_VBPurchasing_Fact_Sheet_ICN907664.pdf (accessed October 26, 2015).
NASEM (The National Academies of Sciences, Engineering, and Medicine). 2016a. Accounting for social risk factors in Medicare payment: Identifying social risk factors. Washington, DC: The National Academies Press.
NASEM. 2016b. Systems practices for the care of socially at-risk populations. Washington, DC: The National Academies Press.
NQF (National Quality Forum). 2014. Risk adjustment for socioeconomic status or other sociodemographic factors. Washington, DC: National Quality Forum.
Press, M. J., R. Rajkumar, and P. H. Conway. 2016. Medicare’s new bundled payments: Design, strategy, and evolution. Journal of the American Medical Association 315(2):131-132.
Price, R. A., A. M. Haviland, K. Hambarsoomian, J. W. Dembosky, S. Gaillot, R. Weech-Maldonado, M. V. Williams, and M. N. Elliott. 2015. Do experiences with Medicare managed care vary according to the proportion of same-race/ethnicity/language individuals enrolled in one’s contract? Health Services Research 50(5):1649-1687.
Rajkumar, R., P. H. Conway, and M. Tavenner. 2014. CMS—engaging multiple payers in payment reform. Journal of the American Medical Association 311(19):1967-1968.
Rosenthal, M. B. 2008. Beyond pay for performance—emerging models of provider-payment reform. New England Journal of Medicine 359(12):1197-1200.
Rosenthal, M. B., R. Fernandopulle, H. R. Song, and B. Landon. 2004. Paying for quality: Providers’ incentives for quality improvement. Health Affairs (Millwood) 23(2):127-141.
Ryan, A. M. 2013. Will value-based purchasing increase disparities in care? New England Journal of Medicine 369(26):2472-2474.
Sessums, L. L., S. J. McHugh, and R. Rajkumar. 2016. Medicare’s vision for advanced primary care: New directions for care delivery and payment. Journal of the American Medical Association 315(24):2665-2666.
Volpp, K. G., A. J. Epstein, and S. V. Williams. 2006. The effect of market reform on racial differences in hospital mortality. Journal of General Internal Medicine 21(11):1198-1202.
Walker, K. O., R. Clarke, G. Ryan, and A. F. Brown. 2011. Effect of closure of a local safety-net hospital on primary care physicians’ perceptions of their role in patient care. The Annals of Family Medicine 9(6):496-503.
Woolhandler, S., and D. U. Himmelstein. 2015. Collateral damage: Pay-for-performance initiatives and safety-net hospitals. Annals of Internal Medicine 163(6):473-474.
Young, G. J., N. M. Rickles, C. H. Chou, and E. Raver. 2014. Socioeconomic characteristics of enrollees appear to influence performance scores for Medicare Part D contractors. Health Affairs (Millwood) 33(1):140-146.
Zaslavsky, A. M., and A. K. Jha. 2015. The role of socioeconomic status in hospital outcomes measures. Annals of Internal Medicine 162(9):669-670.
Zaslavsky, A. M., L. B. Zaborski, L. Ding, J. A. Shaul, M. J. Cioffi, and P. D. Cleary. 2001. Adjusting performance measures to ensure equitable plan comparisons. Health Care Financing Review 22(3):109-126.
This page intentionally left blank.