These are difficult days for those of us who have advocated for pay-for-performance (P4P) as a policy tool to improve health care quality. The idea behind P4P has always been simple: physicians and hospitals should be financially rewarded for providing high-quality care and financially penalized for providing low-quality care. Although this idea has been around for some time, it gained national traction over the past decade, as policy makers pushed toward paying for “value” and not just volume.
What began as experiments with P4P in the early 2000s became a major policy focus with the passage of the Affordable Care Act (ACA). Six years after the ACA’s passage, the evidence on P4P in general is largely mixed, and the evidence on Hospital Value-Based Purchasing (VBP), the national hospital P4P program, is discouraging.
The early studies on P4P found that these programs had little effect on quality but provided bonuses primarily to those who were already doing well. A key study found that the largest hospital-based P4P program known as the Premier Hospital Quality Incentive Demonstration, had no effect on patient outcomes. This program provided bonuses and penalties of up to 2% of total Medicare payments to hospitals based on their performance on a series of quality metrics. However, there has been the mounting evidence—even in multiple meta-analyses—that P4P programs were having little effect across a range of clinical services, from quality of ambulatory care to rates of breast cancer screening. Despite this, Congress created multiple P4P programs within the ACA to incentivize better care. Although some, like the Hospital Readmission Reduction Program appear to have had modest effects, the national VBP effort is not.
When VBP was created, it was structured largely on existing P4P models that already had been found to be largely ineffective. To reward or penalize hospitals, the program made 2% of overall Medicare payments contingent upon performance on a complex set of measures, including process measures, mortality rates for targeted conditions, patient experience, and even “efficiency.” Two recent studies have illuminated the VBP’s effects on patient outcomes.
The first comprehensive study, published early 2016, found that 3 years after the initiation of the program, it had no impact on patient outcomes. Before the program began, mortality rates for the targeted conditions (acute myocardial infarction, congestive heart failure, and pneumonia) were collectively declining at approximately 0.13% per quarter (or about half a percentage point per year). After the VBP incentives kicked in, mortality reductions actually slowed, to 0.03% per quarter. Comparisons of hospitals participating in the program with control hospitals or targeted conditions with nonincentivized conditions provided a similar picture. Instead of accelerating improvements in patient outcomes, VBP incentives appeared to have no effect or possibly even a detrimental one.
Value-based purchasing efforts focused on more than just mortality. Another key metric incentivized by the program is patient experience. So has it improved patient experience in US hospitals? The answer is no, according to a recently published study. Before it was established, the proportion of patients rating their hospital highly was increasing by approximately 1.5% per year. After VBP was established, that increase slowed to 0.6% per year. Again, comparing the effects with those at non-VBP hospitals did not improve the picture.
Why is Hospital VBP Failing?
Those of us who have been advocates of P4P have argued that VBP as designed was destined to fail because key principles of what makes good P4P programs have not been met. As I noted in a JAMA Viewpoint 4 years ago, to be serious about P4P, programs must have 3 design features: incentives that are large enough to motivate hospitals to make sizable investments in improving care, a focus on a small number of high-value measures that will motivate clinicians to engage in changing practice, and a simple design that will enable clinical and organizational leaders to know how they are doing.
The current version of VBP represents none of these principles. Alternatively, the Hospital Readmission Reduction Program was designed using some of these principles and appears to have been successful in lowering national readmission rates. The incentives for reducing readmissions are relatively large, readmissions are easy to measure, and most hospitals can track their readmission rate. Admittedly, readmissions (the measure the program is targeting) are easier to alter than meaningful clinical outcomes, but nevertheless, this design structure has likely facilitated the program’s success and could be applied more rigorously to the VBP program.
Although critics and advocates might disagree about the cause of the poor performance of VBP, there is broad agreement that VBP, as currently structured, is not living up to the promise advocates originally envisioned. Given its cost, this national program should be retooled or stopped. Its critics would argue that it is time to stop the VBP program move on, that P4P in general and VBP in particular have failed because these programs fundamentally cannot succeed. They argue that our quality measurement enterprise is woefully inadequate and produces performance measures that are deeply flawed so that hospitals are more likely to respond by gaming the system than by trying to improve underlying care.
Although there is little doubt that the quality measurement enterprise needs work, it is still possible to salvage VBP and other P4P programs. This is a policy issue for the Trump administration, and it will need to decide whether to redesign VBP to improve patient outcomes or to abandon it altogether.
If the administration decides to redesign the program, it should use the principles laid out above. First, start with strong incentives. Why pay high-mortality hospitals 99.6% as much as low-mortality hospitals? Incentives that put 5% or even 10% of a hospital’s Medicare payments at risk would ensure that hospitals pay attention. Second, policy makers should focus on a few measures that matter most to patients. The mortality rate is a good candidate, as are some measures of patient experience or functional status. Ensuring adequate risk adjustment and accounting for differences in underlying patient populations are important design principles to consider when developing metrics so that hospitals are motivated to provide better care, not avoid patients with clinically complex conditions or who are economically less well off. And finally, any program that is likely to be successful has to have a transparent design and implementation process so that hospitals can track how they are doing and make course corrections along the way.
None of these changes will be easy. Should the administration go down the road of rebooting VBP? This is a decision that cannot be made in a vacuum. There are a variety of other national efforts and experiments under way to improve care delivery, including alternative payment models such as accountable care organizations and bundled payments. Value-based purchasing may still have a role to play in the national payment landscape, but only if it can be redesigned to promote better care.