Risk-adjusted changes in spending associated with hospital participation in bundled payments for care improvement for four medical conditions, 2011-16. Source: British Medical Journal, June 2020.

In the longest and largest study to date, hospital participation in Medicare’s bundled payments for four medical conditions was associated with a 1-2% savings over three years, with no change in mortality. The savings were primarily achieved through reduced spending on skilled nursing care and greater spending on home health in the 90 days after hospital admission for acute myocardial infarction, congestive heart failure, chronic obstructive pulmonary disease, or pneumonia.

Reporting in the British Medical Journal, a team of investigators led by Joshua Rolnick, Joshua Liao, Ezekiel Emanuel, and Amol Navathe compared 226 participating hospitals to 700 non-participating hospitals, matched on propensity scores. They used difference-in-differences models for each outcome, categorizing episodes of care as bundled if they occurred after a hospital initiated a BPCI contract for a given medical condition. The analysis is elegant and worth a full read. A few thoughts:

  1. Your reaction to these findings, from a policy perspective, might depend on what you had expected from bundled payments. On one hand, you might be encouraged that this is the first study to find savings from bundled payments for medical conditions. On the other hand, you might be disappointed if you had hoped that bundled payments would produce greater savings; in unadjusted analyses, these savings amounted to $172 (in 2016 dollars) per episode on total costs that exceeded $18,700. If you were concerned that bundled payments would reduce quality of care, you might be gratified to learn that savings were produced with no increase in 90-day mortality rates.
  2. The savings achieved for medical conditions over three years were similar to the 1.6% achieved for lower extremity joint replacement, as shown by the same group of investigators. That is a remarkably consistent result. And as with joint replacement, the source of the savings was a decrease in skilled nursing facility (SNF) care – although in the case of joint replacement the savings came from fewer admissions to SNFs, while in medical conditions the savings came from a shorter length of stay in SNFs (but not fewer admissions).
  3. It appears that it takes some time to achieve savings from bundling medical conditions. Compared to the first year of participation, the magnitude of savings was larger during the second and third years, suggesting that hospitals learned as they redesigned practices to implement bundled payments. So there may be cause for optimism that we haven’t yet reached a ceiling on savings from bundled payments.
  4. Although it may seem a technical point, how you define an episode actually matters in the analysis. In a companion blog post, the investigators note that the savings could be artificially washed away by preferentially attributing episodes to participating hospitals, a principle used in in Medicare’s “BPCI Advanced” next generation bundled payment program. Policymakers should take note for future program design.

The study, “Spending and quality after three years of Medicare’s bundled payments for medical conditions: quasi-experimental difference-in-differences study” was published in the BMJ on June 17, 2020. Authors include Joshua A. Rolnick, Joshua M. Liao, Ezekiel J. Emanuel, Qian Huang, Xinshuo Ma, Eric Z. Shan, Claire Dinh, Jingsan Zhu, Erkuan Wang, Deborah Cousins, and Amol S. Navathe.