SNF Value-Based Purchasing Demonstration Evaluation Report

On the Medicare side of life, CMS launched a demonstration in 2009 concerning Value-Based Purchasing in three states – Arizona, New York and Wisconsin.  The initiative was designed to test whether improvements in quality of care for Medicare beneficiaries could be achieved.  Last August, L&M Policy Research and Harvard University released a report (click here) that evaluated the demonstration.  AHCA’s Elise Smith, Senior Vice President of Finance Policy and Legal Affairs, provided the following summary of the report:

The Nursing Home Value Based Purchasing (NHVBP) Demonstration is part of the Centers for Medicare & Medicaid Services’ (CMS) initiative to improve the quality of care for Medicare beneficiaries in nursing homes.  The demonstration was launched in July 2009 and implemented by Abt Associates, Inc. (Abt), under contract with CMS.  The three-year, demonstration tested the concept of value-based purchasing in nursing home settings in three states – Arizona, New York and Wisconsin.

For each state, Abt had ranked facilities by performance scores, which determined the annual distribution of payments.  Reflecting the overall budget neutrality requirement for the demonstration, payments in each state were contingent on treatment facilities generating cost savings relative to the performance of a comparison group in each state through the reduction of avoidable hospitalizations and other costs.

L&M Policy Research (L&M) and Harvard, the  consultants contracted to conduct an evaluation of the NHVBP demonstration,  concluded that the demonstration did not directly lower Medicare spending and improve quality for nursing home residents. 

CMS directed that the evaluation address the following questions:
•         How does the pay-for-performance concept work within the nursing home setting?
•         How has the demonstration impacted nursing home quality, cost, service delivery, resident outcomes, organizational structure, and financial status?
•         How do participating homes compare to non-participating homes?

Results
Over the entire three-year demonstration period, savings were realized in Arizona (Year 1) and Wisconsin (Years 1 and 2); no savings were generated in Arizona (Years 2 and 3), New York (Years 1-3), and Wisconsin (Year 3). Thus, only three of the nine NHVBP state-year evaluation periods resulted in payments to the top performing nursing homes. The Year 1 savings in Arizona were relatively modest, while the Year 1 and Year 2 savings in Wisconsin were more.

However, in its evaluation of quality relatively few outcomes suggested major pre-post performance differences across the treatment and comparison groups. Based on its analysis, L&M concluded that quality was unchanged due to the NHVBP demonstration.

•         Officials in all three states explained that many decisions and actions, even those in quality domains targeted by the demonstration, were most likely attributable to the increasing pressures, independent of demonstration incentives, to contain costs and improve quality in response to health care reform.

•         L&M heard very little to suggest that nursing homes responded to the NHVBP demonstration incentives through direct interventions.

In combining the quantitative and qualitative results, L&M concluded that the NHVBP demonstration did not directly lower Medicare spending and improve quality for nursing home residents. Two important questions emanate from this conclusion.

•         First, how did Arizona (Year 1) and Wisconsin (Year 1 and 2) generate savings if nursing homes generally did not explicitly act in response to the NHVBP demonstration?

•         And second, why did the treatment facilities appear to not respond to the payment incentives under the NHVBP demonstration?

L&M surmised that the answer to the first question might relate to the design of the NHVBP demonstration. New York was the only state in which facilities that applied to participate were randomized across the treatment and comparison groups. Thus, the observed savings in Arizona and Wisconsin may reflect differences in facilities that comprised comparison groups selected by propensity scores in these respective states. Indeed, the difference in base-year spending for long-stay residents between the treatment and comparison facilities was much larger in Arizona and Wisconsin than in New York.

Regarding the second question, nursing homes may have altered their behaviors under the NHVBP demonstration for a variety of reasons.

•         First, the demonstration had a very complex payment and reward system and nursing homes may not have understood how their efforts towards improving quality would result in a better performance score and ultimately a reward payment.

•         Second, because CMS had a savings threshold of 2.3 percent and an 80 percent sharing rule, the payouts under the demonstration may have been too small to incentivize major changes in quality.

•         Third, because a payout was only made if the treatment nursing homes generated savings relative to the comparison homes in that state-year period, many nursing homes may have decided not to act in direct response to the NHVBP because their likelihood of a payout depended on other nursing homes in the state also generating savings.

•         Fourth, due to the use of administrative data to determine savings and performance, the sharing of performance reports and payouts to top performing nursing homes took nearly 18 months. This may have lowered the salience of any potential rewards to treatment facilities.

•         Fifth, many nursing homes may have lacked the infrastructure and expertise to engage in quality improvement innovation on their own.

Finally, rather than being incented to change practices because of the possibility of a payout, many facilities saw the demonstration as a reinforcement of actions they were already planning to take or had already begun implementing. Most nursing homes did not change their actions because of the demonstration; rather, some hoped to be rewarded for things that they were already doing or thought their involvement in the demonstration would just be an opportunity to learn from other nursing homes, or prepare for what is to come from CMS moving forward.
Although the NHVBP demonstration was found to have a minimal direct effect on quality, L&M concluded that this result may say more about the specific design features of the demonstration rather than the actual potential of nursing home pay-for-performance.

L&M Recommendation: 

If the Medicare program chooses to move forward with the pay-for-performance concept in the nursing home setting, it should consider changes to optimize the response to payment incentives to improve quality. Modifications to the design of any future NHVBP program might include:
•         Simplified payment and reward rules;
•         Increased payout pools;
•         Relaxation/elimination of budget neutrality restrictions such that the likelihood of payout does not hinge on the efforts of other participating facilities;
•         Offering more immediate payouts;
•         Real time feedback on performance and quality activity results; and providing increased education and guidance on best practices to providers.