David A. Hyman and Charles Silver advocate Results-Based Compensation Agreements between patients and doctors.

Another common complaint about RCBAs is that they would encourage providers to “cherry pick” by treating patients with good success odds and excluding patients who, being seriously ill, are poor risks…

We agree that RBCAs would encourage providers to sort cases. This is one of the principal benefits of RBCAs, namely, their tendency to encourage agents to balance costs, risks, and benefits when assessing the desirability of possible actions. If asked to accept an RCBA for performing [coronary bypass surgery], a cardiac surgeon would rationally consider many factors, including the likelihood that the patient would die. If the patient’s survival odds were dismal, the surgeon would reject the offer, sending a clear signal that the small likelihood of success failed to justify the investment of medical resources.

RBCAs thus frame in cold numerical terms the reality that some medical interventions are inefficient and should not be performed. In a world laden with RBCAs, doctors would get better and better at predicting outcomes. Consequently, they would more often send the message that the cost of health care exceeds the likely gain. This would be a radical departure from current practice, where a procedure that has even a small upside potential qualifies as “medically necessary” and is likely to be performed.

My fear of cherry-picking relates to the idea for compensating physicians on the basis of their aggregate results. What Hyman and Silver are proposing is that patients negotiate with providers on the spot in individual cases. So, if my doctor and I think that the procedure I require is low-risk and straightforward, I might offer a low reward. If we think it is going to be delicate and dicey, the reward might for success might be much higher.

Thanks to Alex Tabarrok for the pointer. The paper is from 2001.