Yet Again, Value-Based Purchasing Did Not Improve Quality

The federal government’s funding of a value-based purchasing (VBP) demonstration project in the Medicare Advantage (MA) program did not improve quality of care, as measured by the plans’ five-star quality ratings. The findings from this demonstration are the most recent evidence that paying health care providers more to provide better care, or to improve their performance – the essential approaches of VBP – does not actually improve quality of care for program beneficiaries or reduce costs for Medicare.  Like the MA demonstration, earlier VBP demonstrations for hospitals and skilled nursing facilities also found no improvements in quality.

Medicare Advantage’s Pay-for-Performance Demonstration

The Medicare Advantage Bonus Payment Demonstration paid bonuses to MA Plans between 2012 and 2014, based on their performance on more than 30 measures across five domains (“preventive care and staying healthy, management of chronic conditions, health plan responsiveness and care, customer satisfaction, and telephone customer service”), as determined by the Centers for Medicare & Medicaid Services’ five-star quality rating system for MA plans.[1]  In contrast to most VBP demonstrations, this demonstration made relatively large bonus payments, 3-10% of plan payments, compared to the more typical 1-2% of payments.  An MA plan was eligible for double bonuses in a county if the county “had lower than average fee-for-service Medicare costs in 2012, if it had a Medicare Advantage penetration rate of 25 percent or greater as of December 2009, and if it was designated as an urban floor county in 2004.”[2]   The three-year demonstration was estimated to have paid an additional $3.43 billion to MA plans, just for the double bonuses.[3]  As with many VBP demonstrations, low-quality plans had their revenues reduced, while high quality plans saw increases in revenues.”[4]

A study of the Demonstration found, “the receipt of double bonuses was not associated with improved quality,” as determined by plans’ star ratings.[5]  However, MA plans “increased plan offerings in counties that were eligible for double bonuses.”[6]  The study concluded, “evidence from this study suggests that double bonuses simply acted as transfer payments to high quality plans in double bonus counties, not as a stronger quality incentive as initially intended.”[7]

VBP Demonstrations Have Not Improved Outcomes in Other Health Care Settings

VBP programs in other health care settings have similarly found no improvement in quality of care for program beneficiaries.

Hospitals

Beginning in 2003 and for its first three years, Medicare’s Premier Hospital Quality Incentive Demonstration provided financial incentives only to the highest-performing hospitals.  The redesign in 2006 was intended to provide incentive payments, in addition, to poorly performing hospitals that improved.  Under the redesign, hospitals could receive incentive payments for improving in measures related to three clinical diagnoses (heart attack, heart failure, and pneumonia) as long as (1) they were in the top 20% of demonstration hospitals for quality improvement and (2) their scores placed them above the median of demonstration hospitals.

A study of the results of the redesign, based on an analysis of a matched sample of 468 hospitals, found no evidence that the demonstration’s goal was achieved.[8]  In fact, “demonstration hospitals improved less in phase 2 than in phase 1,” when their chances of being eligible for incentive payments for providing the best care had been, in reality, non-existent.  The study found that hospitals with the strongest financial incentives under the demonstration’s “improvement” goals were hospitals “whose initial quality was just above the median.”[9]

The researchers offered several explanations for hospitals’ performance, including the difficulty of sustaining improvement, the complexity and limitations in the program’s design, and hospitals’ lack of enthusiasm for VBP.[10]

The researchers described the demonstrations’ results as “not encouraging” as a method of improving quality of care in poorly performing hospitals and suggested, “incentives must be designed carefully if they are to stimulate improvement among lower performers.”[11]

Skilled Nursing Facilities

The poor results of the VBP demonstration in Skilled Nursing Facilities (SNFs) were similar, although the demonstration’s design was quite different.  First, the three-year SNF demonstration was budget-neutral.  Second, it evaluated facilities on four broad domains: nurse staffing (30%), quality outcomes (20%), survey deficiencies (20%), and potentially avoidable hospitalization rates (30%).[12]  Nursing facilities were eligible to receive a bonus if they were in the top 20% overall in performance or, if the facility was on the improvement track, if they were in the top 20% in overall improvement and also in the 40^th percentile in performance.[13]  The amount of money eligible for incentive payments was limited to 5% of total Medicare expenditures, with only 80% payable to facilities and the remaining 20% retained by CMS.  A further limitation in payments was that performance was not rewarded unless there were Medicare savings in the state.  (“If there were no savings in a state relative to the comparison group, then no incentive payment was made to any nursing home in that state regardless of performance.”[14])

An analysis of the SNF demonstration, which continued between 2009 and 2012, found that VBP “did not directly lower Medicare spending and improve quality for nursing home residents.”[15]  The complexity of the VBP system and the limited incentive payments available for participating SNFs could help explain the failure of the demonstration to either lower spending or improve quality.

Additional Concerns about VBP

In addition to its other challenges, VBP exacerbates socio-economic, racial, and ethnic health disparities.  By shifting funding from poorly-performing health care providers to better-performing providers, VBP can have the unintended consequence of further reducing the quality of care at health care providers that serve poor and minority populations.

An analysis of the hospital VBP found that “hospitals caring for more disadvantaged patients did in fact fare worse in the first year of HVBP” and that “payments for improvement in the second year of HVBP did not eliminate disparities.”[16]  Hospitals serving poor patients and a large number of patients from racial and ethnic minorities tended to do worse financially under VBP.

Conclusion

In 2007, the Center for Medicare Advocacy expressed skepticism about VBP as a method of improving health care quality for Medicare beneficiaries, based on early research that found little evidence that VBP (earlier, called Pay for Performance) improved care and some evidence of unintended negative consequences (including health care providers’ denial of care to severely ill patients, provider gaming, and findings that changes reflected better documentation, rather than real changes in care for patients).[17]

More recent VBP demonstrations continue to find no evidence of care improvement.  They indicate that the design of VBP programs is critical and suggest that in order to have any chance of improving quality of care, VBP programs need to choose the correct measures of performance and appropriate payment levels (not too big, not too small).  However, the lack of success with the VBP demonstrations to date, whether the design includes large payments (MA plans) or is budget-neutral (SNFs), calls into question the validity of VBP as a principle for meaningful health care reform.

December 16, 2015 – T. Edelman