Incentives in Health: Different Prescriptions for Physicians and Patients

Financial incentives abound in health care. They are found in the ways physicians are paid and in the ways health insurance coverage, copayments and deductibles are structured for patients. The effects of these incentives are often understood through conventional economic principles, with the assumption that individuals are self-interest maximizers who respond directly to changes in incentives. In contrast, behavioral economics imports insights from psychology and recognizes that people often fail to respond to incentives as rationally as they might. In some cases people lack information, but in others they just seem to act contrary to their own known interests, like when they overeat, don’t take medication that will improve their health, or fail to wear seat belts in cars.

Although sometimes caricatured as focusing exclusively on human irrationality, behavioral economists recognize that rationality is dependent on the person and the situation they are in. People tend to be more economically rational when they are highly educated, have good information about alternatives, make similar decisions repeatedly, receive clear and rapid feedback on the consequences of those decisions, and have the time and emotional distance to make decisions in a deliberative and dispassionate fashion. People tend to be less perfectly rational when they are less educated, lack relevant expertise, do not receive feedback about the decisions they make, and make decisions rapidly and under the influence of strong emotions. This dichotomy is relevant to informing our thinking about the role of incentives in medicine, particularly when it comes to thinking about how to reduce widespread overutilization of low-value services.

There is ample evidence that doctors, who are highly educated, tend to repeat similar decisions, and get immediate and clear feedback about those decisions, are exquisitely sensitive to the incentives they face. Physicians tend to recommend the tests and treatments they will financially benefit from. For example, oncologists who are reimbursed based on the chemotherapy drugs they provide administer more, and more expensive, drugs.¹ Physicians paid for specific procedures tend to recommend more of those procedures, on average, than those paid on a capitated basis.² Clearly, if the US health care system is going to improve the value of health care spending, it will have to do so in part by exploiting this very rationality of doctors, by abandoning provider payment systems that contribute to excess utilization of high cost, low value, services.

Patients utilizing health care services face a situation starkly different from that of physicians. Patients tend to face idiosyncratic health issues, get relatively little useful feedback about the quality of medical decisions, and often make decisions when sick and, as a result, in a heightened emotional state. Perhaps not surprisingly, therefore, at least partial irrationality abounds in the behavior of patients. We cannot expect a patient who is suffering and vulnerable to weigh the costs and benefits of alternative tests or treatments rationally. These decisions are hard enough when patients are feeling well, as can be seen from studies of health insurance decisions, which reveal that many patients often fail to choose insurance policies that provide the most extensive coverage at lowest cost.³ This is not to say that patients are entirely irrational; for example, as classical economic theory predicts, once patients meet their deductible, their readiness to utilize health care services increases.⁴

What are the implications for health policy of the greater degree of rationality displayed by physicians relative to patients?

First, attempts to address issues of overutilization of low value services should focus mainly on physicians rather than patients. We should reduce provider payment for services of low value. Given the fiscal consequences of poorly controlled health care spending, it makes no sense to pay just as much for a service or procedure that isn’t really necessary as one that is life saving. This approach need not affect physicians adversely; we could couple lower payment for low value services with higher payment for high value services in situations where there is underutilization of those high value services.

At the same time, however, we need to be careful in the implementation of physician incentives. Broadly based incentive alternatives, like capitated payment versus fee for service, are blunt. Both can perversely create unwanted results—too much care in the case of fee for service medicine, and too little in the case of capitation–with little regard to value. Although incentive schemes can be made more complicated to try to motivate desired and narrower patterns of practice, the more complicated they are the less effective they are likely to be.⁵ Moreover, research in psychology and behavioral economics shows that monetary incentives can crowd out non-material motivations such as those inherent in the desire to conform to professional values.⁶ Financial incentives are not meant to replace intrinsic professionalism, but no system of reimbursement is incentive neutral, so we need to develop and test systems that are generally successful at achieving intended consequences without creating new problems.

Second, although patients may be less ideal targets for incentive programs in general, our knowledge of who responds well to incentives and when suggests areas in which they might be more or less constructively applied.

One straightforward implication of the greater rationality shown by people facing repeated decisions is that we should think differently about the role of patient cost sharing in the setting of chronic illnesses than in the setting of acute illnesses. Patients with acute conditions are likely to confront unfamiliar and emotional decisions and are therefore not appropriate targets of cost-sharing incentive programs that require a dispassionate evaluation of costs and benefits. In contrast, higher cost sharing for low-value services can make more sense for patients with chronic conditions, who are more likely to face repeated similar decisions repeatedly. Although studies of reduced cost sharing for medications based on the value of medications find disappointingly small improvement in adherence at a relatively high cost (few patients who would not have otherwise taken the medication take it, but many who would have taken it at the original cost receive a discount),⁷ studies of increased cost sharing consistently observe large reductions in utilization.⁸

A second implication of limited rationality on the part of patients is that we need to think carefully about how incentive programs that target patient health behavior are designed. Section 2705 of the ACA expands, to 30% of total premiums, the proportion of employee health insurance premiums that can be conditioned on health outcomes measured by biometric measures such as LDL cholesterol, blood pressure, BMI, and smoking status. A key question for these provisions – about which there is quite limited data – is how much these behaviors can be modified through premium-based health incentives. Employees who smoke or are obese, who will end up paying the highest rates, tend to be the poorest, leading to a situation in which the most disadvantaged individuals and families will pay the highest insurance premiums. In our own research we have found that the impact of incentives for behavior change can be greatly enhanced by implementing them in a fashion that takes account of and leverages irrationalities in decision making; in other words, while adjusting the prices of insurance will likely have some impact on behavior the effects achieved could potentially be much bigger if the underlying psychology of decision making is taken into account in the design.⁹ If premiums tied to outcomes succeed in making people healthier at high rates, the health benefits may justify the increase in regressivity, but programs will need to be designed carefully to increase the likelihood that they will have a positive impact on behavior.¹⁰

There is an expression in economics: “There are no bad people, only bad incentives.” Incentives have been and will continue to be an inevitable part of health care financing and can be used to increase the value our health care spending provides in improving health. But, to be most likely to achieve intended without unintended consequences, incentive design and implementation need to take account of how patients and physicians respond to these incentives and how they respond differently in different circumstances.