Ideas to improve the measurement of surgeon skill

In Britain, "[a]t least one in three heart surgeons has refused to treat critically ill patients because they are worried it will affect their mortality ratings if things go wrong" (Knapton, 2016).

Ideas for improvement

Because heart surgeons' success rates are published, they have an incentive to avoid treating high-risk patients. And, apparently, the do have a say in which patients they treat. How can we avoid such a situation, while still informing patients about how good individual surgeons are? A few ideas:

Don't allow surgeons to choose patients

It's unclear exactly how surgeons choose patients, but we could try to remove any opportunities they have for influence. This could work to some extent, but I think it would be difficult to remove their influence altogether. For example, surgeons could call in sick on days with high-risk patients, or they could sabotage the scheduling in other ways.

Differ the scores depending on patient prospects

Comparing heart surgery on an old and sick patient to that of a young and otherwise healthy patient seems very crude. At a minimum, we could make a simple model taking into account factors such as age to score different cases differently. Another approach would be to incorporate risk assessments from doctors other than the ones partaking in the surgery. On the more sophisticated side, we could set up prediction markets to make the probabilities more accurate. Betting on people's lives could be difficult to sell, though.

Point auctions

A third way could be to have surgeons "bid" points on patients. There are several ways to implement the details. The end goal is to award surgeons who take on high-risk patients and succeed. Patient risk is measured using the auction prices.