Oh, red meat to my skeptical soul, another piece of research showing that a heavily-promoted cost containment tactic may not do much, in this case the bundled payment initiative by CMS, which runs the Medicare program. (NEJM Article) The technical name for the project is Bundled Payments for Care Improvement and it is separate from the joint replacement bundled payment effort, which woefully for glass-half-full types like me, showed some reduction in spending. The BPCI covered certain medical conditions and this evaluation used five common ones–congestive heart failure, sepsis, heart attack, pneumonia and COPD. Claims from 2012 to 2015 were used to compare hospitals participating in the bundled payments with those that didn’t and to assess utilization and costs before and after acceptance of bundled payments. Participation was voluntary and hospitals had several models to select from, with varying degrees of risk. Between 70 and 125 hospitals participated in each of the conditions that were used in the study. The participating hospitals were more likely to be urban, non-profit, teaching institutions and to have more beds. The researchers sought to evaluate the effect of the new payment on spending during an episode including the initial hospitalization and 90 days thereafter, ER use, length of stay, readmissions and mortality.
Prior to the intervention, the average Medicare payment for the conditions per episode was $24,280 at the participating hospitals; under bundled payments it was $23,993. At the control facilities, the average before bundled payments was $23,901 and after the start (remember this group didn’t participate so they were getting paid the usual way) it was $23,503. So the non-participating hospitals actually had a bigger decline in spending, although not statistically significant. The same results held for looking at each condition individually, although there may have been some savings in regard to pneumonia. In a piece of sort-of good news, there did not appear to be a change in patient health status, which would suggest hospitals that participated did not start avoiding more complex patients when they got bundled payments. There was no decline in length of stay, ER use or in readmissions and mortality was indistinguishable, with exceptions for better mortality in heart attack patients and worse in COPD ones. Not great, or even good, results. Participation in this initiative was voluntary, so you would assume only highly motivated hospitals signed up, which makes the results more disappointing. Okay, let me say a kind word. This is very early in an evaluation. Really you should want to wait for five years to see how behavior and outcomes have changed. The behavior change that is intended to be prompted by these programs is not easy, especially in large organizations, so you have to give it time to see what the impacts really are. But early on, my heart leaps in joy, as intellectual policymakers’ promises and hopes are dashed.