“If the system permits exceptions and pays for the more expensive procedure because a doctor orders it and says it is really necessary then we’re back where we started. Doctors are going to want the higher reimbursement associated with the more expensive procedure.”
I think this is where ACOs might work. If you calculate the size of the population that would benefit from the more expensive surgery, you could budget accordingly. Let the physicians within the appropriate specialty divide up the cases accordingly.
Not only does this solve the heterogeneity problem with respect to comparative effectiveness based payments, but it also marries value-based insurance design, consumer-directed care, and ACOs. That is, the system pays capitated rates to integrated provider groups for the expected cost (taking account of patient heterogeneity) of the cheapest procedures that work well. Doctors agree to provide the procedure they think best, having already been compensated for it (value-based). Patients who want a procedure more expensive than recommended would pay the marginal cost (consumer-directed).
This puts a heavy burden on quality monitoring. Providers could offer the cheapest procedure to everyone, even knowing it isn’t optimal for everyone. Could patients be protected from that form of “gaming”? This is hard to solve.
Later: No, it isn’t that hard. If we know how to compute the expected cost taking the heterogeneity into account then we know when providers are over-prescribing the cheapest procedures. That is, one need only monitor the rate of procedure type and penalize for more than trivial deviations from expectation. Providers would be encourage to save by preventing the need for procedures, not on the distribution over procedures for a given condition. Recognize that this is, in effect, rationing of procedures: 10% of this, 15% of that, 75% of this other one. The difference is that it is done somewhat implicitly through financial incentives. Kind of an elaborate game just to end up back to pretty explicit rationing. Still, worth thinking about for more than that 10 minutes I have so far.