Comment by toast0
> Minimizing patient days spent in hospital was also mentioned as a metric. Sure that may be good for some reasons (eg avoiding iatrogenesis) but incentivizing it could lead to patients being sent home too early instead of receiving proper care.
You usually combine a minimize days in hospital goal with a minimize readmittance goal. And combine with supervision to ensure low readmittance isn't due to patients being admitted to the morgue instead. Ideally, some longer term measure of patient outcomes. But as you mentioned, you also need to account for the mix of patients.
You’re describing the vicious cycle of Goodhart’s law. The one measuring the metric adds epicycle after epicycle while the one gaming the metric develops ever more subtle strategies for subverting it.