Paying for Nursing Home Quality: An Elusive But Important Goal.

Abstract

On a given day, almost 1.3 million individuals receive care from approximately 15,000 nursing homes nationwide (Kaiser Family Foundation, 2020). Nursing home residents have high levels of medical complexity, have physical and cognitive impairments, and often lack financial resources. As such, these are among the frailest and most vulnerable individuals in our health-care system. In 2020, the United States spent $196.8 billion on nursing home care, the majority of which came from public sources (Hartman et al., 2022). Yet, questions persist about the effectiveness of this investment (Halifax and Harrington, 2022). Nursing home quality of care continues to be an important public policy issue (Ouslander and Grabowski, 2020), despite prolonged public outcry (Institute of Medicine, 1986, Institute of Medicine, 2001; Mendleson, 1974; Vladeck and Twentieth Century Fund, 1980) and government reports (Office of Inspector General, 2014; U.S. Government Accountability Office, 2018; U.S. Senate, 1974). Often, the number of nurses per resident is low, with 75% of nursing homes failing to meet Centers for Medicare & Medicaid Services' (CMS’) expected registered nurse staffing levels based on residents' acuity (Geng et al., 2019). Staffing is also highly variable from day to day (Geng et al., 2019; Mukamel et al., 2022), and the average annual staff turnover rate exceeds 120% (Gandhi et al., 2021). Residents may develop new health problems after nursing home admission due to unsafe care practices. Amenities that are common within a nursing home—including the food, activities, and public spaces—are too often substandard (Miller et al., 2014). Feelings of isolation and loneliness are consistently reported in the states that collect information on quality of life and care experiences (Roberts and Ishler, 2018; Shippee et al., 2015). These quality issues predate the COVID-19 pandemic, but all have been magnified since March 2020 (Levere et al., 2021; McGarry and Grabowski, 2021; Montgomery et al., 2020). Although Medicaid and Medicare pay for most nursing home care in the United States, very little of this public payment is linked to quality, either through bonuses or penalties. In other parts of the health-care system, an increasing number of payers has implemented value-based payment models that hold participating providers accountable for overall spending and quality (Barnett, Wilcock, et al., 2019; Joynt Maddox et al., 2018; McWilliams et al., 2016). A key question for policy-makers is whether and how to increase the accountability of nursing homes for the quality of care they provide under Medicaid and Medicare. Although Medicaid and Medicare pay for most nursing home care in the United States, very little of this public payment is linked to quality, either through bonuses or penalties. In 2022, a National Academies of Science, Engineering, and Medicine (NASEM) report concluded that quality is substandard in many U.S. nursing homes. The report offered a series of recommendations to transform how the United States delivers, regulates, finances, and pays for nursing home services. One of the payment recommendations focused directly on testing expanded value-based payments for nursing home care. Specifically, recommendation 4E states: This paper first reviews the U.S. experience with value-based payments for nursing home services. Unfortunately, most of the models tested to date have not shown improvements in the quality of nursing home care. The paper then considers how policy-makers can implement the NASEM recommendation to ensure a greater likelihood of success relative to prior efforts. The first documented attempt to improve nursing home quality through incentive payments was a randomized experiment conducted in 36 San Diego–area nursing homes over the period of 1981 to 1983 (Weissert et al., 1983). The experimental group received three separate bonus Medicaid payments for admitting heavy-care patients, meeting outcome goals on some patients, and discharging and maintaining some patients in the community. The treatment group was found to have better outcomes at a lower cost, admit more residents with functional disabilities, and shorten lengths of stay (Norton, 1992). Despite these promising early results, over two decades passed until there was a major federal effort to test value-based payments in nursing homes. This delay could have been due to the government's focus on using prospective payment approaches to control expenditures, CMS prioritization of public quality reports that aimed to provide market-based quality incentives by influencing consumer choice, industry's resistance to such models, or public focus on auditing and enforcement of care standards. Nevertheless, CMS introduced the three-year, three-state Nursing Home Value Based Payment demonstration in 2009. Nursing homes in New York State that agreed to participate were randomized into intervention (n = 72) and control (n = 78) groups, while propensity-score matching identified comparator facilities for participating facilities in Arizona (n = 38) and Wisconsin (n = 61). The model assigned nursing homes a quality score for a combination of nurse staffing, Minimum Data Set–based quality measures, survey citations, and hospitalization rates. The design required budget neutrality. If the treatment group collectively saved money relative to the control group, only the participating nursing homes with the highest-quality scores received a payout. Little quality difference was found across the three years of the demonstration, and limited savings were found during some years in Arizona and Wisconsin (Grabowski et al., 2017). However, a mixed-methods evaluation found that these savings were likely due to regression to the mean. Ultimately, the demonstration was not found to be successful at improving quality or reducing Medicare spending, partially attributable to a series of design issues. As authorized in the 2014 Protecting Access to Medicare Act, the federal government next introduced the Skilled Nursing Facility Value Based Payment (SNF VBP) model in 2018. Unlike the earlier demonstration, this model focused on short-stay residents and only examined a single outcome measure, hospital readmissions. Program payment represented uniform 2% withholding that could be paid back to nursing homes based on their performance on the readmissions measure. Research on the SNF VBP found that facilities serving vulnerable groups were less likely to receive bonus payments and more likely to be penalized in Year 1 of the model (Hefele et al., 2019; Qi et al., 2020). Moreover, the program did not improve readmission rates for the lowest-performing facilities (Burke et al., 2022). The Medicare Payment Advisory Commission (2021) detailed a series of shortcomings to improve the model, some of which are potentially addressed in changes incorporated by CMS in 2021. Nevertheless, this value-based payment program has also largely been unsuccessful to date. Several states have introduced value-based payments through their state Medicaid programs. Between 2001 and 2009, eight state Medicaid agencies adopted pay for performance (P4P) programs in nursing homes. Considerable variability exists in the size of the incentive payments and the quality measures included in the payment models (Werner et al., 2010). Overall, these Medicaid value-based payment models were not found to result in consistent improvements in nursing home quality (Werner et al., 2013). One successful state model was the state of Minnesota's value-based payment model, which was individualized to specific groups of nursing homes. The Performance-Based Incentive Payment Program allowed nursing homes to propose a specific quality improvement program to the state. The models focused on a range of different quality measures, including fall prevention, readmissions, and quality of life. Nursing homes received 85% of expected payments upfront and were at risk for the remaining 15%. An evaluation of the Performance-Based Incentive Payment Program model found improvement in a composite measure of quality that combined the very specific quality programs among participating nursing homes (Arling et al., 2013). The NASEM report adopted the CMS definition of an alternative model. The report states that the APM, “according to the Centers for Medicare & Medicaid Services, is a ‘specific subcategory of value-based purchasing initiatives that require providers to make fundamental changes in the way they provide care’ and that ‘shift financial incentives further away from volume by linking provider payments to both quality and total cost of care results’” (Centers for Medicare and Medicaid Services (CMS), 2015, p. 5). Thus, providers take on substantial financial risks to deliver high-quality care at lower costs, typically across health-care providers and settings. APMs can apply to a specific medical condition, an episode of care, or a patient population (Centers for Medicare and Medicaid Services (CMS), 2015, p. 367). Nursing homes have been affected by several newer APMs, such as bundled payments that link the hospital, the provider, and post-acute care payments to an episode of care (Barnett, Wilcock, et al., 2019), and accountable care organizations, in which provider organizations share incentives for controlling health-care costs and meeting benchmarks for a population of Medicare enrollees (McWilliams et al., 2017). Although these models have not focused specifically on nursing home care, the elimination of post-acute nursing home services has been the major source of savings within these demonstrations, with a recent commentary terming nursing homes “the piggy bank” for these APM payment models (Barnett, Mehrotra, and Grabowski, 2019). Similarly, Medicare Advantage (MA) plans have also targeted post-acute care as a source of savings. Some of this lower utilization might be due to favorable health and socioeconomic selection, but several studies have found that lower post-acute care utilization within MA plans is not associated with any change in claims-based outcomes, such as mortality or readmissions (Huckfeldt et al., 2017; Kumar et al., 2018). Moreover, enrollees in MA plans typically use lower-quality nursing homes (Meyers et al., 2018). One Medicare Advantage model that was specifically designed for long-stay nursing home residents is the Intuitional Special Needs Plan (or I-SNP) model. The I-SNP takes on the financial risks for Medicare expenditures for nursing homes and typically increase on-site clinical care to manage this risk. One recent study suggested that beneficiaries in the I-SNP have fewer hospitalizations and emergency department visits and more skilled nursing facility use relative to traditional Medicare beneficiaries (McGarry and Grabowski, 2019). Few, if any, of these models have focused on improving quality in nursing homes to date. Indeed, these concurrent payment initiatives, co-occurring with shifts in post-acute populations, regulations, and available community alternatives, make isolating the individual effects of a particular payment program difficult. Additionally, none of these models have addressed the NASEM report's recommendations to expand future demonstrations to consider advanced illness care models and the influence of patient social drivers of health on outcomes. A series of important lessons have emerged from the existing value-based payment models that should be considered in the implementation of the NASEM recommendation. First, NASEM recommendation 6A called for the inclusion of a robust measure of resident and family experience on the Care Compare website. In the home health-care setting, results from patient surveys to understand patient care and care issues, communication between providers and care recipients, and overall quality of care have been publicly posted and utilized in the home health star rating calculation for years (Schwartz et al., 2021). These measures should also be incorporated into future nursing home value-based models to ensure these new payment approaches are consistent with the wants and needs of residents and their families. Similarly, the outcomes included in the model should be concordant with the goals of the specific resident. These goals and preferences will differ across residents (Roberts et al., 2018). Further, given the high proportion of residents with Alzheimer's disease and related dementias who receive care in nursing homes, it is important to also capture the caregiver experience to ensure that the experiences of residents with cognitive impairments are incorporated into the model. Moreover, the social drivers of health and available caregiver support will influence the resources needed to obtain important outcomes across facilities. Second, many of the prior models rewarded nursing homes based on a very limited set of quality measures. In many respects, it has been a “Goldilocks problem,” in that some models include too many measures (e.g., Nursing Home Value-Based Payment) and others include too few (e.g., SNF VBP). Another concern is that these models may lead to “teaching to the test,” as providers target measures that are rewarded by the model (e.g., hospital readmissions), at the expense of resources and attention diverted from measures that are not rewarded (Feng et al., 2005; Kash et al., 2007). Recent improvements to ensure accurate data reporting, such as the use of the Payroll-Based Journal system to capture staff hours paid, instead of hours reported, provide improved consistency and reliability (Geng et al., 2019). The IMPACT Act of 2014 tasked CMS with identifying and implementing standardized data elements for inclusion across post-acute care assessments, including nursing homes, to allow valid comparisons of the value of services being provided across the broader long-term care financing and delivering system (Chen et al., 2022). However, further efforts are needed to ensure the integrity of provider-reported assessments and administrative data and to translate these improvements in data integrity into improvements in the quality of care received. Because there are many measurable and actionable components that comprise nursing home quality, determining the “right” set of measures is key. Indeed, evidence from the state Medicaid P4P models suggests a larger weighting of clinical outcomes in the payment formula leads to larger improvements, with small weights leading to no improvement or even a worsening of some clinical outcomes (Konetzka et al., 2018). As a result, it is important to get the “right” set of measures and then determine the appropriate weights to optimize improvements in nursing home quality of care. Third, policy-makers will need to incentivize the growth of successful models. An old line suggests P4P often stands for “pennies for performance” in many models, offering weak incentives to motivate providers to improve quality. Meaningful payouts for quality improvement and performance need to be ensured. Moreover, many voluntary APM models, such as bundled payments, and accountable care organizations have suffered from low enrollment and face potential issues of selection, as the most primed providers volunteer to participate. Although there are obvious political trade-offs with mandatory enrollment, mandatory enrollment is not only essential to guard against differential selection but is also easier to implement, as most nursing homes receive Medicare and/or Medicaid funding, so program participation could be made a condition for receiving federal funding. The existing SNF VBP is limited in many ways, but the mandatory enrollment approach to enrollment provides a stronger foundation for future nursing home payment models. Most importantly, any value-based payment system should encourage health equity and carefully monitor for disparate effects on historically marginalized populations. Once again, results from the existing SNF VBP model have reinforced concerns that these models exacerbate existing disparities. Nursing homes that serve vulnerable populations are less likely to receive bonus payments and more likely to be penalized, thereby punishing facilities that care for a greater share of Medicaid residents and historically marginalized communities, as dual eligibility, racial minority, and ethnic minority statuses are significant predictors of admission to the lowest-rated facilities in Medicare's star rating system (Zuckerman et al., 2019). Such programs create payment deficits for nursing homes serving high-risk populations, which reduce the resources that may be allocated toward patient care and quality improvement. The growth in the use of assisted living facilities as a substitute for traditional nursing home care for those with greater financial resources and lower acuity (Grabowski et al., 2012), alongside the growth in the use of non-institutional care settings, including the push towards home- and community-based services by state Medicaid programs (Siconolfi et al., 2019; Siconolfi et al., 2021), has signaled that any payment changes must be considered within the context of the larger post-acute and long-term care health system to ensure equitable access and delivery of care. Moving forward, CMS should closely monitor the highest and lowest performers of these value-based models—and support any additional expenditures needed to achieve the desired quality goal (Damberg and Elliot, 2021)—to encourage a more equitable payout across providers. The concept of value-based payments for nursing homes is one that has existed for decades. Lessons from prior efforts provide important insights that should be taken into consideration, alongside the recent 2022 NASEM recommendation, as the development of a value-based payment model moves forward and continues to focus on aligning nursing home payments with quality of care. The right payment model will require incorporating elements of the resident experience, getting the mix of quality measures included in the incentive payment correct, incentivizing participation and program retention, and promoting health equity across nursing homes nationwide. The right payment model will require incorporating elements of the resident experience, getting the mix of quality measures included in the incentive payment correct, incentivizing participation and program retention, and promoting health equity across nursing homes nationwide. This supplement is sponsored by the John A. Hartford Foundation. Dr. Saliba is a part time employee of the Veterans Health Administration. In addition, over the past three years, Dr. Saliba has received research support from the Borun Center for Gerontological Research, the National Institutes of Health, Veterans Administration (VA) Health Services Research and Development, the Centers for Medicare & Medicaid Services and the University of Toronto Centre for Aging and Brain Health Innovation, Baycrest. Over the past three years, Dr. Grabowski has received research support from: AARP, Agency for Healthcare Research & Quality, Commonwealth Fund, Donaghue Foundation, GRAIL, John and Laura Arnold Foundation, National Institutes of Health, the Robert Wood Johnson Foundation, and the Warren Alpert Foundation. He has also received payments from the Medicare Payment Advisory Commission, the Analysis Group, Bluestone Physician Services, and Health Care Lawyer PLC. None.