A(nother) Tale of Two Papers

At the close of 2022, two papers on the effectiveness of alternative payment models were published, one by JAMA, and one by HealthAffairs. The first studied the effects of the Medicare bundled payment model, contrasting the effect between participating hospitals and participating physician groups. The second studied the effects of an "ACO-type" payment incentive, mostly with larger health organizations.

Both studies compared the performance of participants with non-participants. The JAMA paper focused solely on Medicare beneficiaries, while the the HA paper focused solely on commercially insured populations, splitting the observations between fully insured and self-insured plan members. That's where the similarities stop.

Led by Joshua M. Liao and colleagues, the JAMA paper uses very rigorous and proven methods to compare participants and non-participants in order to draw highly defensible observations. And the upshot is that both participating hospitals and physician groups generated savings compared to non-participants, with physician groups achieving measurable savings in surgical bundles only, while hospitals achieved savings in surgical and medical bundles.

Led by a team from Elevance (fka Anthem), the HealthAffairs paper uses a decent framework for comparing those participating in the health plan's mostly upside-only payment model with those that don't. However, it falls far short of the Difference-in-Difference approach of the JAMA paper. And the upshot in this study is....hard to tell. But from my perspective, the "savings" are mostly due to pharmacy savings, which could have simply been the result of better managing the formulary or any other factors. Unsurprisingly, the authors found that there were no savings in hospital spending and limited savings in outpatient facility care. It's kind of a "meh" paper.

As is more fully described in a recently published report by the Duke-Margolis Institute for Health Policy , in many ACO-type payment models (especially those with weak/no downside), expecting any real savings is illusory. But as the JAMA paper shows, if there is real downside and the payment model doesn't cause a facility to cannibalize revenues, then savings accrue.

Don't get me wrong, total costs of care payment models with downside risk work. It's the weak sauce ones that don't. And they also work a lot better when the providers contracted aren't worried about keeping heads in beds. As we forge into 2023, let's not forget these important lessons from 2022.