Can Pay-for-Performance Take Nursing Home Care to the Next Level?

“We don't want to make pay for performance only about payment---it's about doing the right thing.” Leslie V. Norwalk, Deputy Administrator for the Centers for Medicare & Medicaid Services, April 19, 2007¹

More than 3 million Americans rely on services provided by a nursing home at some point during the year, and among them 1.5 million stay long enough to consider the nursing home their main residence. These individuals, and their circle of family members, friends, and relatives, count on nursing homes to provide care that is of high quality. However, as late as 2006, one in five nursing homes nationwide was cited for serious deficiencies—deficiencies that caused actual harm or placed residents in immediate jeopardy.²

The task of ensuring high-quality nursing home care falls largely to the Centers for Medicare & Medicaid Services (CMS), which is the largest single purchaser of nursing home services (about $64 billion per year.)³ In coordination with state surveying activities, CMS establishes quality of care standards in nursing homes as conditions of participation in the Medicare and Medicaid programs. Recently, CMS issued its 2007 Action Plan to address nursing home quality.³ The list of activities included mostly ongoing items such as updating their web-based report card “Nursing Home Compare” and expanding the certification and surveying activities required of nursing homes.³ However, the last item on the action plan was new, under the heading “value-based purchasing”, CMS listed its plans to pilot test pay-for-performance in nursing homes.

Pay-for performance is a reimbursement approach designed to reward health care providers for achieving high levels of performance and providing the best value from health care investments. This approach is a modification to the fee schedules of flat rates per service, where reimbursement is the highest when the most services are rendered, regardless of improved patient outcomes. In 2001, the Institute of Medicine identified the flat rate fee schedule as a barrier to quality improvement and recommended that health care purchasers, like CMS, adopt reimbursement policies linked to quality improvement.⁴ This recommendation has become the foundation for more than 100 pay-for-performance programs initiated across a range of health care settings. These programs vary considerably in implementation as each organization grapples with inadequate data systems, stakeholder politics among the perceived losers and winners, inexperience with financial risk management, and financial constraints. For instance, in Minnesota the local nursing home industry lobbied the state legislature to reject a comprehensive pay-for-performance program that incorporated efficiency ratings and instead adopt a simpler model.⁵ Furthermore, the promise of pay-for-performance has yet to be determined, as the impact may not be realized for many years on patient health, provider behavior, and institution viability. There are no long-term studies of pay-for-performance in any setting.

CMS, particularly through the Medicare program, has been encouraged to lead the way in the pay-for-performance movement, and is currently testing several models. The main model that served as the basis for the nursing home demonstration is a pilot test set in the community-setting, the “Physician Group Practice Demonstration.” That study of pay-for-performance in ten physician group practices is nearly completed. The results for the first year of performance have recently been announced for measures related to quality of care for diabetes mellitus (e.g., annual foot exams, influenza immunizations, and HgbA1c measurement every 6 months.)⁶ All ten groups achieved the performance benchmarks on at least seven of the ten quality measures, and two received perfect scores. However, only two groups earned the performance payments (totaling $9.5 million), and they were not the top scorers. The other eight groups did not earn performance payments because they did not generate savings to the Medicare program despite incurring upfront costs to improve their care management systems, totaling millions of dollars in some cases.⁷

Pay-for-performance, as CMS is implementing it, requires budget neutrality. Thus, financing for the incentive payments must come from savings to the Medicare program, regardless of the quality of performance. In the case of the CMS nursing home pilot project, called the “Nursing Home Value-Based Purchasing Demonstration,” the hypothetical savings for the incentive payments would come from reductions in hospitalizations and subsequent skilled nursing facility stays. Savings will be calculated as the difference in the expected Medicare spending for the pay-for-performance homes relative to that for the controls. Any Medicare savings will be divided among the high performing homes at the end of the year after Medicare retains the first 20%. If no savings are generated, no performance payments will be made, regardless of the nursing homes' performance.⁸

Additionally, the CMS nursing home pilot project calls for up to five states to provide 50 nursing homes to test the pay-for-performance program and another 50 to serve as a control group. Although participation will be voluntary, each home will be randomly assigned to the experimental or control group. The demonstration would include all Medicare-eligible beneficiaries residing in nursing homes and will last three years, during which the quality of nursing home performance will be assessed according to the following measures: staffing levels, resident outcomes, survey deficiencies, and appropriate hospitalizations. Incentive payments will be awarded to the top 20% of homes on absolute performance and the top 20% of homes on improved performance, assuming that they generate savings to Medicare in the process. The control group will receive the usual per diem rates.

Assumptions that cost-savings will result from performance-based payment policies---especially within 1 year of the policy change---show a fundamental shift away from the original conceptualization of how reimbursement methods may improve the quality of care. While the Institute of Medicine promoted pay-for-performance, it also cautioned that “even if a change ultimately reduces cost, the process of evolving from the current system to the system of the 21^st century may incur significant cost…health care organizations themselves will need to invest in change.”⁴ In contrast to the CMS model, all of the states with current pay-for-performance programs in nursing homes (Georgia, Iowa, Kansas, Minnesota, Ohio and Oklahoma) set aside new money for the incentives payments, often in the range of several million dollars. The state of Virginia recently estimated that incentive payments might cost the state up to $16 million.⁹

The CMS pay-for-performance demonstration project in nursing homes calls for nursing homes to invest in care management systems without a clear way to recoup the investment. Most of the performance measures (e.g., no survey deficiencies) will not generate the payment pool of Medicare savings, and thus high performers may receive no performance payment. Furthermore, it is uncertain whether Medicare savings can be determined in a reliable manner. For instance, savings from residents in managed care plans are excluded and the risk adjustment methodology, which is critical for discouraging cherry-picking, has not been selected.⁸ Additionally, CMS' approach does not account for several unique characteristics of the nursing home setting. First, there is a reciprocating tension between payers. Medicaid is the main payer of daily routine care and some skilled services in nursing homes, while Medicare is the main payer of acute care services needed by nursing home residents. Nursing homes that improve their care management systems to reduce hospitalizations might save money for Medicare but may increase the cost of routine care for Medicaid.¹⁰

Second, it is unclear to what extent nursing homes influence potentially avoidable hospitalizations or whether appropriate versus inappropriate hospitalizations can be distinguished for nursing home residents.¹¹ “Potentially avoidable hospitalizations” is a construct developed for and tested in patients admitted from the ambulatory setting. When these measures are applied to the nursing home setting, inexplicable factors emerge that suggest incongruity. For instance, nursing home residents with Alzheimer Disease have a high likelihood of putatively avoidable hospitalization; this may indicate the need for measurement exceptions when patients have severe cognitive-impairments that impede communication about emerging medical problems^{12, 13} Hospitalizations of nursing home residents have been increasing but the reasons for the increase are unclear.¹⁴ The increase may relate to factors outside the control of the nursing home such as staffing shortages, a more medically complex and functionally-impaired population, or less generous per diem rates.¹⁴

Third, the nursing home industry is faced with a shrinking pool of patients whose need for more intensive health care services has been increasing.¹⁵However, a business case for pay-for-performance can generally be made only by organizations meeting two criteria: 1) their current levels of service are above or near the quality standard so the costs of upgrading systems are low; and 2) the incentive payment offers them more benefit than admitting more patients because they are operating at or near full capacity.¹⁶ As supporting evidence of this premise, the two organizations in the CMS physician group demonstration that earned incentive payments through cost-savings to Medicare were also the two largest. Approximately 52% of all nursing homes operate as part of large chains, making them the mostly likely candidates to benefit from a pay-for-performance program.¹⁷ Finally, there is generally poor correlation between the most common measures of quality in nursing homes and resident outcomes (or even across the different measures of quality). For instance, a nursing home with lower than average rates of mortality may perform poorly in the area of pressure ulcer prevention, or visa versa, in contrast to other settings.^{18, 19} Scant consensus or scientific evidence yet exists on the definition of high quality care in nursing homes or which resident outcomes can be improved with better quality care.

Unfortunately, the potential for pay-for-performance to improve health care quality or reduce costs in nursing homes is still largely speculative. The only empirical work comes from the early 1980s when incentive payments were made to nursing homes in San Diego for: 1) accepting patients needing the most functional assistance, 2) improved patient functional status within 90 days of admission, and 3) prompt discharges of patients who remained out of the facility for at least 90 days.^{20, 21} After adjusting for multiple factors, the investigators found that the incentives induced homes to admit more individuals with severe disabilities and send more individuals home or to lower skilled facilities. In addition, residents in the homes receiving the incentive payments were less likely to be hospitalized or to die than those in the control homes. However, this promising result came at a cost---the average daily cost to Medicaid rose by about 5%, due to the incentive payments and increased administrative burdens.

The consequences of pay-for-performance in a budget neutral environment is evident in the example of St. John's Health System---one of the two sites in the CMS Physician Group Demonstration that achieved a perfect score on quality but did not receive a performance payment.⁷ This multi-specialty group practice of 500 physicians serves over 35 communities in southwest Missouri and northern Arkansas. For the demonstration, St. John's Health System developed an electronic patient registry to prompt routine care management tasks. The process required investments in the existing billing and clinical systems, and nearly a year's worth of time commitments from senior staff, including physicians, nurses, case managers and information technology specialists. After initiating the new patient registry, the practice saw large improvements in screening and therapeutic targets. For instance, the proportion of patients with diabetes who received retinal eye exams increased from 18% to 41%, and those achieving target blood pressures increased from 3% to 53%. However, these improvements did not result in cost-savings to Medicare, so the group practice ended up bearing all of the costs of the investments in health information technology.

Over 20 years ago, the Omnibus Reconcilliation Act of 1987 (OBRA '87) was passed to focus national attention on nursing home quality issues and to reform the standards of care. Since then, many regulations have been passed to improve the processes of care delivered in nursing homes. Despite these efforts, though, quality of care in nursing homes is still a concern. Reforming nursing home care means redesigning the care process to enhance coordination and efficiencies. Research shows that the following areas have high potential for reducing nursing home hospitalizations without adversely affecting residents: 1) increased use of nurse practitioners, physician assistants, and nursing staff; and 2) improvement in the hospital-to-nursing home transition.²² Such quality improvement activities require upfront and sustained outlays before the benefits may be realized. Aligning payment policies with these activities will certainly help in accelerating the adoption of these changes; however, pay-for-performance programs cannot substitute for strategic investments to improve care. In the meantime, the nursing home community/industry would be well advised to think carefully before participating in CMS' Nursing Home Value-Based Purchasing Demonstration, as participants may be at substantial risk of having pay-for-performance turn into performing well without being paid.