While health spending in the United States far surpasses that in other industrialized nations, the quality of care in the US is no better overall, and on several measures it is worse. This stark fact has led to a wave of payment reforms that shift from rewarding volume (as fee for service does) to rewarding quality and efficiency. Such pay-for-performance schemes seem to be common sense and are now widely used by private payers and Medicare. But astonishingly, there’s little evidence that they actually improve quality.
What do we really know about the effectiveness of using financial incentives to improve quality and reduce costs in health care? There is robust evidence that health care providers respond to certain financial incentives: medical students have a higher demand for residencies in more lucrative specialties, physicians are more likely to order tests when they own the equipment, and hospitals seek to expand care for profitable services at the expense of unprofitable services. It would seem that increasing payment for high-quality care (and, conversely, lowering payment for low-quality care) is an obvious way to improve value in health care. But evidence suggests that health care is no different from other settings where similar payment incentives have been tried, such as education and private industry. Not only do these payment policies often fail to motivate the desired behaviors, they may also encourage cheating or other unintended responses.
Overall, evidence of the effectiveness of pay-for-performance in improving health care quality is mixed, without conclusive proof that these programs either succeed or fail. Some evaluations of pay-for-performance programs have found that they can modestly improve adherence to evidence-based practice.
There is little evidence, however, that these programs improve patient outcomes, suggesting that to the extent that health care providers have responded to pay-for-performance programs, that response has been narrowly focused on improving the measures for which they are rewarded — such as making sure patients receive recommended blood tests if they have diabetes or the right cocktail of medications if they are hospitalized with a heart attack. Although these measures are important for patient care, it may take a full reengineering of the health care delivery system to broadly improve patient outcomes.
Despite considerable concern about unintended consequences in these programs, so far the adverse effects have been relatively minor, with little evidence that providers are avoiding high-risk or disadvantaged patients, gaming, or ignoring areas of care that are not financially rewarded in order to improve their apparent performance. The lack of evidence for unintended effects is perhaps not surprising, given the limited evidence of the intended effects of these programs, though concerns remain that as pay-for-performance incentives become stronger, and perhaps more effective, evidence of cheating may surface.
Given the wide adoption of pay-for-performance programs, it’s surprising is that even after extensive research, very little is known about how their design — including what outcomes are rewarded, the optimal size of incentives, and the criteria for payment (e.g., quality achievement or quality improvement) — affects provider behavior. Also, because nearly all of the evidence comes from programs that reward quality, we know almost nothing about whether pay-for-performance can improve efficiency or lower cost.
Experience with pay-for-performance in health care and other settings shows that these programs are hard to design. The best combination of performance measures, organizational level of accountability, criteria for payment, and incentive size is not obvious, and unintended consequences are common.
To be effective, we may need smarter incentives that take advantage of the cognitive biases that skew decision making, such as loss aversion. For example, evidence from a randomized trial in Chicago schools found that student math scores improved if teachers were paid in advance and forced to repay bonuses if an improvement standard was not achieved. Scores did not improve if teachers stood to receive incentives only after scores met improvement standards.
Other insights from psychology and behavioral economics may also have the potential to make pay-for-performance programs more effective. For example, another possible reason for the disappointing response to these program in health care is that targeted provider behavior is more likely to be intrinsically motivated (driven by the desire to reduce suffering, for example) and thus less likely to respond to external incentives such as payment. Examining the effects of pay-for-performance in other sectors underscores the difficulty of using performance pay for intrinsically motivated workers — and the pitfalls of trying.
Recent critiques of productivity pay claim that extrinsic incentives are only effective in situations in which tasks are routinized and narrowly defined, leaving workers with little intrinsic motivation, such as windshield installation. This view is supported by copious evidence from social science that financial rewards for intrinsically valuable activities – including performance in school, sports, and interesting work activities – undermine motivation and can decrease task performance. Perhaps not surprisingly, there is little evidence that pay-for-performance has been effective in U. S. education: A major program in New York City – including 20,000 teachers and $75 million – proved to be a high-profile failure. Dan Ariely and colleagues have also argued that for professionals working in situations where there is uncertainty about the relationship between inputs (such as the choice of diagnostic tools, reperfusion therapies, and discharge planning for patients admitted with acute coronary syndrome) and outputs (such as 30-day mortality), performance contracts cannot be sufficiently detailed to reward optimal practice in all circumstances. As a result, pay-for-performance can divert attention from the big picture and toward a myopic focus on meeting the performance goals that are typically defined in these contracts. Thus, even if we had pay-for-performance programs with smarter designs, it remains unclear whether we could overcome the fundamental problems associated with incentive contracts directed at narrow goals for intrinsically motivated activities.
Pay-for-performance was brought to health care to address a real problem: the suboptimal quality of our health care given our levels of spending. In the face of perverse financial incentives, health care providers’ intrinsic motivation to deliver quality has not been enough to provide sufficiently high-quality, high-value care in the United States. The root of these problems, however, may lie in system failures, not the failures of individual providers. While health care providers want to help the patient in front of them, they may not feel obligated (or have the incentive) to solve system-level problems stemming from factors they feel are outside their control. One potential solution lies in broader health reform, such as global payment for populations rather than piece-rate bonuses for individual patients. Coupled with public quality reporting, global payment reform has the potential to expand the scope of provider accountability, take advantage of providers’ intrinsic motivation, and improve population health. Such efforts may hold more promise for value improvement in US health care than attempts to exploit providers’ extrinsic motivation through tweaks to fee-for-service payment.
This blog first appeared on Harvard Business Review on 10/09/2013.
View our complete listing of Strategic HR blogs.