[colinmhayes]

Outcomes for very sick people being worse is taken into account so providers aren't punished if their success rate is lower for those patients. We have all the data we need to know what the outcomes should be for a provider over the course of a year. Anyway, providers have to precommit to their pool of patients, so it's not like they can just pick the lucky ones.

Capitation is admittedly a much more complicated way to administer health care, but fixing the incentives is absolutely worth the pain we may face in the short term. Which we might not considering how well these programs have been shown to work.

[clairity]

you're not fixing the distortive incentives with this scheme, just transforming them. 'capitation' incentivizes hoarding of patients, most obviously by incentivizing the minimization of treatment times and associated costs but also through outright fraud (by falsifying records, etc.). so you're essentially trying to maximize patient outcomes by capping costs, which is a terrible proxy metric all around.

to actually improve healthcare, we'd need to increase spending per person (by an order of magnitude or more, using more team-based medical care rather than 1-on-1), for not only direct care costs, but also research and development. what we absolutely need none of is deadweight losses, like profit-maximizing medical administration, big pharma, and the whole medical insurance industry. medical 'insurance' has become a complete misnomer, as it no longer serves to mitigate the distasterous effects of low probability, high impact events, but as a socialization of routine medical care, which is not insurance at all.

[RSZC]

Huh, I didn't expect to come across something I have such direct knowledge of.

I used to work with this data directly - I worked with MACPAC to prepare reports for the CBO on the efficacy of pay-per-performance programs, especially as it pertained to the possibility of switching Medicaid from a fee-for-service model to a pay-for-performance model. (6-9 years ago)

I just wanted to chime in on this one quote:

> We have all the data we need to know what the outcomes should be for a provider over the course of a year.

The data here is exceedingly low quality, especially for Medicaid. Medicare is in a better state due to being centrally administered. We do *not* have all the data we need to make pay-for-performance more than an extremely rudimentary approach. Is that rudimentary approach still better than fee-for-service? Maybe - that's a complicated question I let the public health PHDs write very long reports on.