OECD/WHO/EOHSP report on P4P

European Observatory of Health Systems and Policies

Paying for Performance in Health Care: Implications for Health System Performance and Accountability

Edited by Cheryl Cashin, Y-Ling Chi, Peter C. Smith, Michael Borowitz and Sarah Thomson
OECD, WHO, October 6, 2014

Forward from the OECD (excerpts):

The problem is that not enough is known about whether and how P4P actually increases value for money in health systems. The evidence that P4P improves health outcomes, or even quality of processes of care, is limited at best.

(This) volume analyses the experience of P4P programmes in 10 OECD countries, selected to reflect the wide range of health system contexts and challenges across the OECD.

The findings of the volume in many ways mirror the findings of the few rigorous systematic reviews of P4P programmes, and the opinions of many leading commentators. Pay for performance does not lead to “breakthrough” quality improvements, and performance measures and other key building blocks of P4P programmes remain highly inadequate.

This volume will not provide answers to questions such as whether or not P4P works, which performance measures are most appropriate, or what is the right level of financial incentive to get results. Instead - and more importantly for real health financing policy in complicated contexts - are the insights about how P4P might be used to strengthen health system governance and strategic health purchasing to continue the shift taking place in many countries from paying for performance to paying for value.

Mark Pearson, Head of Health Division, Directorate of Employment, Labour and Social Affairs
Organisation for Economic Co-operation and Development

Chapter thirteen:

United States: California integrated healthcare association physician incentive programme

By Meredith Rosenthal

One of the first, and perhaps the largest, private pay for performance (P4P) initiatives of this era was launched by the Integrated Healthcare Association (IHA) in 2001 with eight health plans representing ten million members in California. The IHA programme is of particular interest not only because of its size, but also because it has been sustained for more than a decade and has been independently evaluated.

Results of the programme:

Performance related to specific indicators

More generally, IHA’s own monitoring reports give a mixed picture of performance improvement over time. Performance measures included in the IHA P4P programme have improved modestly and unevenly across measures, with no evidence of “breakthroughs” in quality improvement.

Programme monitoring and evaluation

Two controlled studies provide the strongest evidence of impact of the IHA initiative. These studies find that not all targeted clinical process measures of quality improved. Among the measures that could be analysed, only cervical cancer screening improved differentially among the IHA participants, and improvement was modest at best.


While there has been no systemic analysis of the impact of the IHA programme on equity, several empirical clues suggest that P4P may not have distributed its benefits equally… (I)interviews with physician group leaders revealed some concerns that the P4P programme has caused groups to avoid patients whose health of health behaviour would negatively affect the group’s performance.

Cost and savings

While no formal analyses have been reported, it is unlikely that improvements in clinical quality, health information technology, and patient experience (to the extent they have occurred) would generate saving for payers.

From the Conclusions

Another possible explanation for the weak results may be the continued expansion of the measure set and the difficulty physician organizations face in making investments in quality improvement when the targets are continuously moving. There is an obvious tension here with the desire to include a comprehensive set of measures to avoid “teaching to the test,” a narrow focus that causes providers to concentrate on a small subset of tasks at the expense of unrewarded domains, and to incorporate the best available measurement science over time.


Designing Smarter Pay-for-Performance Programs

By Aaron McKethan, PhD; Ashish K. Jha, MD, MPH
JAMA, November 6, 2014

The idea behind pay for performance is simple. Because individuals and organizations respond to incentives, physicians whose patients achieve desirable outcomes should be paid more as an incentive to improve their performance. Yet the results of pay-for-performance programs have been largely disappointing. One argument is that neither the right set of incentives nor the right set of metrics has been identified. Another explanation, which has received far less attention, is that the right set of patients has not been identified for targeted efforts.

To the extent that higher-risk patients can be reliably identified prospectively, this information can inform the design of smarter, more targeted pay-for-performance programs. Specifically, a targeted pay-for-performance program would have, at its core, a prediction model that would identify patients who are at elevated risk of failing to meet a meaningful clinical goal or of having a bad outcome. Predictive models are not just risk-adjustment models already in use by payers to create a level playing field. Predictive models can take into account any factor that is likely to affect a patient’s chance of a poor outcome.

There is little doubt that the effectiveness of these programs will be driven, in large part, by the ability to prospectively identify at-risk patients. However, given the failure of recent efforts to meaningfully improve outcomes, testing targeted pay for performance may be worth the effort.



By Don McCanne, MD

Pay for performance (P4P) continues to be promoted as a means of improving quality while reducing costs. This 338 page OECD/WHO report adds to the abundance of the policy literature that shows that P4P does not achieve these goals, and may actually impair equity.

The policy community never gives up on a bad idea. In this JAMA article (access is free), McKethan and Jha suggest that we improve P4P by applying it only to prospectively-identified at-risk patients. Not only would that be a good study, but it could also result in P4P rewards that are five times the current levels. What? Greatly increase the complexity and uncertainty by testing only at-risk patients, if you could even identify them? And then depend on provider greed to drive the program? Come on!

One thoroughly tested model that would greatly reduce wasteful spending while improving quality by redirecting the savings to more appropriate care is the single payer model - a national health program. We can let the policy people go out in the alley and play their P4P games while we get serious about improving Medicare and providing it to everyone.