The Premier Pay-for-Performance Demonstration project in Medicare shows a tapering trend of quality improvement both for participating and non-participating hospitals. Perhaps secular trends, not P4P, are responsible for this.
The National Demonstration Project is a national pay-for-performance program considered to be Medicare’s flagship test of whether financial incentives delivered at the hospital level can produce improvements in process measures related to quality of care. In this demonstration, 266 hospitals that were members of Premier’s “Perspective” hospital performance benchmarking service agreed to allow the collection of performance data on quality measures that would place those hospitals subject to performance-based compensation.
Phase one of the demonstration (2003-2006) involved rewarding hospitals for performing in the top ten percent (2 percent bonus) or twentieth percent (1 percent bonus) of demonstration hospitals for outcome measures related to heart attack, heart failure, pneumonia, bypass surgery, and hip and knee replacement. Phase two modified incentive schemes to allow for three methods of financial rewards. Under phase two, hospitals received rewards for (a) benchmark attainment relative to other demonstration hospitals, (b) high achievement by attaining the top twenty percent of performance for demonstration hospitals, and (c) improvement compensation by achieving the median score of demonstration hospitals in the current year that ranked in the top twenty percent of demonstration hospitals for quality improvement.
This study compared a sample of 250 demonstration hospitals to 250 matched non-demonstration hospitals for quality outcome measures of heart attack, heart failure, and pneumonia across phase one and phase two of the P4P initiative. The authors addressed the questions of, “did demonstration hospitals improve more in phase two (with incentives added for improvement) than in phase one?” and “did demonstration hospitals with initially lower performance improve more than those with initially moderate performance during phase two?”
In phase one, performance measures for quality among demonstration hospitals improved more than that of matched comparison hospitals. However, the demonstration hospitals experienced a weakening of quality improvement relative to matched comparison hospitals in phase two. Also, demonstration hospitals starting in the lowest quality quartile showed much more quality improvement than their matched comparison hospitals in phase one.
This did not happen in phase two though both demonstration and non-demonstration hospitals continued to improve. For phase two, incorporating incentives for improvement showed no evidence of quality improvement among the lowest performing hospitals. This may have been due to the fact that hospitals starting in the third quartile of quality for phase 2 had the most potential for financial gain, while those in the lowest quartile had the least to gain financially from quality improvement. For hospitals in the lowest quartile even a high level of quality improvement would not put them above the median of the top twenty percent of demonstration hospitals exhibiting improvement, thereby making them ineligible for rewards.
This study suggests that financial incentives are not the silver bullet when targeting lower performing hospitals for quality improvement. This is not because lower performing demonstration hospitals did not improve dramatically between 2004 and 2009 for composite scores of heart attack, heart failure, and pneumonia, but because their matched comparison hospitals did so as well. In fact, the disparities in quality between high performers and low performers in both groups were greatly diminished. The data suggest that the motivation for improvement involves a secular trend not accounted for in this analysis.
The next step is to pinpoint the real cause for quality improvement. Past pay-for-performance studies involving financial incentives have shown quality improvement for composite measures but struggle to associate monetary compensation with improvements at the patient level. This is probably due to uncertainty with how financial compensation is used to facilitate process improvements that reach individual patients. Possibilities range from rewarding individual, high-achieving providers to investments in staff, training, and infrastructure.
Patrick Fitzgerald, MPH