Incentives for Quality Clinical Care

There is very little that people agree on in healthcare reform. If you’ve paid any attention to the controversy surrounding the Affordable Care Act (ACA, also known as Obamacare), you know that the policies governing our healthcare system are not only hopelessly complex, but also very contentious. Disagreements stem from fundamental ideological gaps, different theories of implementation, and competing financial interests. The fact that very few of the policymakers and advisors have any substantive experience in the healthcare industry only complicates matters further.

Despite these differences, there does seem to be something of a consensus on one systemic problem in American healthcare: the existing compensation scheme for physicians encourages overuse of medical services, which contributes to the burgeoning costs of healthcare in the United States. The ACA’s solution is to shift away from the existing method of paying providers to one that seems a whole lot more logical. Enthusiasm has exploded for these “pay-for-performance” programs over the past decade, with the number of programs growing from dozens to hundreds. However, this new direction, although seemingly a more logical way of reimbursing physicians, may not be as impactful as pundits and scholars predict. In fact, a body of research across a number of fields suggests it is precisely the wrong direction of reform.

In our “fee-for-service” system, physicians are paid based on the quantity and type of services they provide. The more tests and scans your doctor orders, the bigger the paycheck he or she takes home. From an economic standpoint, this method of determining payments is almost blasphemous. Professionals ought to be compensated based on the value of the good or service they deliver to a consumer. Although in reality many other factors go into determining the size of the paycheck for any job, that simple economic principle generally holds in many industries. Healthcare is an exception. Doctors can make a lot of money in the U.S. without providing high quality care; all they have to do is order a lot of tests.

Obviously, this reimbursement scheme encourages providers to order more services as a means of maximizing their income. Not only is this a flawed incentive structure that fails to incentivize good care in any way, but it also drives overutilization, or the ordering of more (and more expensive) services by physicians than is actually necessary for the patient. Overutilization is often cited as a primary reason why the United States spends so much more on healthcare per capita than any other country, and has the same or worse health outcomes than other developed nations.

One of the goals of recent healthcare reform efforts has been to replace this system of misaligned incentives with one that promotes accountability, efficiency, and quality. The ACA has a number of provisions that attempt to do this. There has been bipartisan support for this push away from the fee-for-service model to a model that promotes value by rewarding high quality, cost-effective care. In fact, this restructuring of incentives was cited as a point of agreement between President Obama and Governor Romney during the 2012 election cycle, as both candidates agreed that this was a necessary step to lowering costs and improving quality. These new schemes, which are becoming more and more popular, fall under the umbrella model of pay for performance, in which physicians are incentivized to produce good health outcomes for patients by delivering quality care. Doctors will no longer be reimbursed solely based on quantity of services. Their paychecks will increasingly depend on the value they create for patients.

On the surface, this seems great. In fact, it makes you question why it wasn’t always this way. Basic economic principles dictate that you’ll get more of what you incentivize. The fee-for-service system incentivized the use of medical services, which helped make American healthcare the mostly costly healthcare system in the world. The new pay for performance model incentivizes physicians to deliver costeffective care that helps patients to recover as quickly as possible. It follows that we should start to receive better quality care at lower costs. But will we?

No one knows for sure if these models will prove effective in the long run, although recent studies have shown mixed evidence, most of which does not bode well for the new schemes. It may be too early to tell, but multiple recent studies, which find that health outcomes didn’t improve very much and costs didn’t go down very much after implementing pay for performance models, give us reason to worry. A recent metaanalysis of all the published literature evaluating pay-for-performance programs in hospitals reported bleak results, concluding that most of these programs have little to no effect on clinical care quality, patient outcomes, or cost. A number of other recent reviews have also found insufficient evidence to prove that these new programs benefit patients. The biggest support for pay for performance models in the United States came from an oft-cited joint public-private study investigating how effective this incentive structure could be in controlling costs and improving patient outcomes over six years. Although early results were very promising, helping to fuel the hype around pay-for-performance, many studies have since found that any improvements in patient outcomes were short-lived, and largely evaporated by the fifth year. It certainly seems that these new payment schemes may not be as great as we thought.

It turns out that there are decades of research from sociology and emerging evidence from behavioral economics that might explain why pay-for-performance isn’t working as well as we expected. To understand this, though, we have to briefly digress to a famous social science experiment concerning the “candle problem.” The psychologist Karl Duncker designed the “candle problem” in the 1940s, and it continues to be one of the most popular cognitive performance tasks. The test asks the participant to affix a candle to a wall (so the candle wax doesn’t drop on the table below), and provides participants a book of matches and a box of thumbtacks with which to do this. Many people first try to use a thumbtack to tack the candle on the wall while others try to melt part of the candle and adhere it to the wall. Neither strategy works. The solution requires that participants overcome “functional fixedness” to realize that the box the thumbtacks are in can also be used to hold the candle, and that box can be tacked to the wall. It’s not a very difficult problem to solve, but it does take some time and a little creativity.

Why is this relevant to how we pay doctors? The candle problem has been used to conduct some of the most rigorous and insightful studies into the power of incentives. Several studies have shown that when you offer monetary incentives based on how quickly a participant can solve the candle problem, the participants take longer to solve the problem than do participants who are not offered any incentives. This counterintuitive result goes against our basic understanding of economic principles and human behavior. However, when you take the thumbtacks out of the box and provide them as separate objects to the participant, eliminating the need to break functional fixedness to solve the problem, then the incentivized group solves the task much faster than the control group. These results are very robust, and have been replicated dozens of times. Social scientists conclude that incentives, although very effective for simple, straightforward tasks, do not help and can even do harm for tasks that require creative thinking and problem-solving skills. Particularly for tasks that are both mentally demanding and intrinsically rewarding, higher incentives often lead to worse performance.

The connection to pay-for-performance incentives in healthcare is clear. Most physicians would probably agree that diagnosing and treating a patient is not an easy thing to do and is often mentally challenging, and I certainly hope that the vast majority of doctors find caring for patients to be personally rewarding. In these ways, the practice of medicine is analogous to the candle problem. And the provision of incentives for clinical care may have similar results as well. Because clinical medicine is a difficult yet satisfying engagement, pay-for-performance incentives might simply not work in healthcare. According to researchers, this is a consequence of human nature. Insights from social science and behavioral economics have been largely missing from the healthcare reform controversy, although they may be among the most relevant pieces of evidence, providing part of the explanation for why studies have consistently reported mixed results about the effectiveness of pay-for-performance incentives.

There’s a larger takeaway here beyond worries about pay for performance. The fact that one of the few proposals that garnered bipartisan support in the healthcare reform controversy is also one of the least evidenced ones is deeply troubling. We trust our policymakers to listen to the research and make educated decisions about our futures, but the only issue on which they can agree and be decisive is also one where research provides only tenuous support at best. Until we can ensure that a political consensus among policymakers arises from a well-substantiated consensus among researchers about relevant evidence, it will be difficult to trust our political institutions to make the right decisions regarding our healthcare system.