There is probably already some adjustment for risk - anyone can see that a smoker in his 60s will have higher mortality than a non-smoker in his 30s. The probl

There is probably already some adjustment for risk - anyone can see that a smoker in his 60s will have higher mortality than a non-smoker in his 30s. The problem might be that the statistician doing the risk adjustment doesn't have 'skin in the game', so his/her assessment of risk is worse than the surgeon's. (Although if that were the case I would expect the statistician to be just as likely to overestimate the risk than underestimate it, yet I never hear about surgeons favouring the difficult cases because they think they can improve on the expected mortality rate in such cases).