3 ms·

It is hard, and once you've defined quality metrics, it's even harder to write a contract that captures them and get parties to agree on them The latter issues

It is hard, and once you've defined quality metrics, it's even harder to write a contract that captures them and get parties to agree on them The latter issues might be resolved somewhat by the Amazon group; the former is still a concern, but I think there are opportunities to do meaningfully better things This is just an idea, and I've not done much research on it, so take with a grain of salt: I think patient focused quality measures might be better. For example, look at cohorts of patients with similar characteristics (like in case control studies) who are experiencing a similar health event (knee surgery) and compare providers based on the outcomes these types of patients have. This is something the JV could do, is patient focused, and avoids all the challenges you mention with metrics measuring / managing individual providers (which isn't just hard to measure but demoralizing to docs)

nradov 9y ago
That's exactly the problem. There's no reliable way to construct a cohort of "similar" patients. We don't even know all the variables other than provider quality which impact the outcome of knee surgery. Do we have to consider age? Genetics? Sex? Fitness level? Co-morbid conditions? Blood Type? Personality? Something else? That data simply doesn't exist, and won't exist in our lifetimes. With large enough populations some of those confounding variables even out. But that doesn't help when trying to evaluate individual providers who treat relatively small numbers of patients per year. Of course in any area all of the local doctors know from subjective experience and word-of-mouth which other doctors are skilled and which are quacks. But none of them will say so publicly, and you can't quantify those opinions.
- aaavl2821 9y ago
  People do case control studies all the time, the factors used depend based on disease but there is plenty of literature to support selection of matching factors I think it would make sense to start at the hospital level; provider level probably isn't feasible as you say. Hospitals see enough patients where the sample sizes are pretty big, and they are the main cost driver in healthcare, so rating hospitals by cost / quality could work

nradov 9y ago

That's exactly the problem. There's no reliable way to construct a cohort of "similar" patients. We don't even know all the variables other than provider quality which impact the outcome of knee surgery. Do we have to consider age? Genetics? Sex? Fitness level? Co-morbid conditions? Blood Type? Personality? Something else? That data simply doesn't exist, and won't exist in our lifetimes. With large enough populations some of those confounding variables even out. But that doesn't help when trying to evaluate individual providers who treat relatively small numbers of patients per year. Of course in any area all of the local doctors know from subjective experience and word-of-mouth which other doctors are skilled and which are quacks. But none of them will say so publicly, and you can't quantify those opinions.

aaavl2821 9y ago
People do case control studies all the time, the factors used depend based on disease but there is plenty of literature to support selection of matching factors I think it would make sense to start at the hospital level; provider level probably isn't feasible as you say. Hospitals see enough patients where the sample sizes are pretty big, and they are the main cost driver in healthcare, so rating hospitals by cost / quality could work