I said earlier that we might not always want to trust the people with the best track record when we go off the beaten path. Sometimes the tools that work in one environment aren’t the best to use when the environment changes and what we really need are experts in developing tools.
A related post on OB points to another reason why trusting track records isn’t always best.
Call it the best-kept secret in Massachusetts medicine: Health insurance companies pay a handful of hospitals far more for the same work even when there is no evidence that the higher-priced care produces healthier patients.
We might naively use this as evidence that less prestigious hospitals actually offer better care. However, another possible interpretation is that the toughest cases go to the most prestigious hospitals and, despite the higher standard of care, they end up with a higher mortality rate.
But who goes to more prestigious hospitals? In general it is the people who can afford to pay the higher costs of healthcare. Those people are likely to be wealthier than those who go to less affluent hospitals, and thus likely to have better health in general.
That leads to two possibilities: first, the only wealthy people who end up in hospital are REALLY sick because minor illness has been caught and treated at the GP level; or, secondly, people in expensive hospitals are likely to be less sick than those at cheaper hospitals. I’d be interested to know which way the selection bias usually runs. Without that knowledge it’s impossible to interpret the inital data and tell which hospitals are a better bet.