Nope that doesn't work because there's still no way to reliably control for the confounding factors. We don't even know what all the confounding factors are. Patients aren't randomly assigned to providers.
Sure, we don't have that information today. But, if we want to collect relevant information we could easily have a subset of doctors with random patient assignment. IMO, that's simply an implementation detail required if we want good information.
PS: US healthcare spending is over 3 trillion per year, rationally even minor improvements are worth large investments.
Sure, we don't have that information today. But, if we want to collect relevant information we could easily have a subset of doctors with random patient assignment. IMO, that's simply an implementation detail required if we want good information.
PS: US healthcare spending is over 3 trillion per year, rationally even minor improvements are worth large investments.