> some system of having different physicians assess ratings than the ones performing the procedures - maybe in some kind of double-blind fashion to prevent any kind of coordination strategy.

That would simply result in an unwritten rule where everyone gives everyone else good ratings.

The "objective metrics" part is key. Medical records are generally written in a way that weaves a standard narrative of the doctor as either a hero/savior or a valiant defender of a lost cause.

Doctors only get investigated when derogatory facts accumulate to an extent that this "standard narrative" collapses. It's not really an objective or scientific standard.

The article itself alludes to a 'EuroSCORE' that purports to rate the risk of mortality beforehand. It seems like the mortality ratings should be risk-adjusted. A double-blind system as you suggest could work like this: First, the rating system would be opt-in on the part of the participating surgeons. It seems likely that patients would prefer rated surgeons, so there would be an incentive to participate but it would not be mandatory. Those who belonged to the rating body would be required to do anonymous risk assessment on some number of cases per month. They wouldn't know who the patients are or who is treating them, they would only get anonymized records. Their risk assessments would be calibrated in the short run against other assessments, and in the long run against outcomes. These calibrated risk assessments would be used to adjust the score of the treating surgeon on the outcome, so that they would be penalized less for mortality in patients with a very high assessed risk.