Previous Submission

Early Results from Medicare’s Bundled Payment Initiative

Next Submission

Decreased post-acute utilization delivers savings to NYU under a bundled payment pilot for joint replacement

As alluded to by Prof Chandra in class today, U.S. insurers are increasingly experimenting with bundled payments. Within Medicare, the Center for Medicare and Medicaid Innovation (CMMI) is testing bundled payments in a variety of settings. The Bundled Payments for Care Improvement (BPCI) initiative, the largest experiment, is a rather complicated collection of several different voluntary bundled payment arrangements. The most popular arrangement (known as Model 2) involves a retrospective bundled payment that covers an inpatient admission, associated post-acute care (such as inpatient or home-based rehab), and all condition- or disease-related services incurred up to 90 days following hospital discharge. Provider organizations–typically acute care hospitals or large physician groups– participating in BPCI Model 2 can choose among 48 different clinical episodes that are defined by specific diagnosis related groups (or DRGs). DRGs are the way Medicare has been paying for inpatient (or hospital-based) care since the 1980s. Under DRG payment, Medicare gives hospitals a lump sum to care for a patient based on their diagnosis at the time of discharge (eg Pneumonia or Hip Replacement). DRGs are, in essence, a “bundle” themselves, but just for inpatient care. The innovation under BPCI Model 2, therefore, is the addition of 90-days of condition- or disease-related expenses to the DRG “bundle.” Since the BPCI bundle is retrospective, Medicare continues to pay under traditional its traditional fee-for-service arrangements, but eventually reconciles actual expenditures against a target price (based on historical payments for similar services). If the actual amount is under the target, Medicare writes the provider organization a check for the difference. If the actual amount is over the target, Medicare sends the provider organization a bill.

NYU Langone Medical Center was one of the earliest participants in the BPCI program, joining in 2011. Just like our good friends in Germany and Sweden, NYU chose to experiment with lower extremity joint replacements (hips and knees). The team from NYU recently published some interesting results in the Journal of Arthroplasty (additional coverage can be found here). To evaluate the impact of the BPCI program, they compared cost, process, and quality outcomes between 721 Medicare patients undergoing joint replacement during the first year of the program with 785 patients undergoing the same procedures in year three. The results were striking:

  • Cost: Medicare payments decreased by $6,708 ($34,249 to $27,541) per patient episode. This was driven almost exclusively by a reduction in post-acute inpatient rehabilitation costs. Discharges to inpatient rehabilitation facilities dropped by nearly half (44% to 28%) driving savings of $5,486 per. By year three of the program, Medicare was spending only $742 on inpatient rehabilitation per patient (down from $6,228 at the beginning of the program). Since NYU was liable for rehabilitation costs, BPCI introduced a new financial incentive to reduce expensive inpatient rehabilitation. The articles do not specify the target for the bundle, but it can be assumed that NYU retained at least a portion of these savings generated for Medicare. Inpatient rehabilitation facilities, on the other hand, lost a ton of revenue.
  • Process: The average hospital length of stay (aLOS) decreased from 3.58 days to 2.96 days. Since the DRG already “bundled” inpatient care, the BPCI program did not really contain additional incentives to reduce aLOS. However, spurred by participation in the BPCI program, the NYU team launched a care process improvement program that employed early recovery pathways and other initiatives similar to what we saw in Germany and Sweden. As we have discussed, reducing aLOS increases utilization and throughput, both of which improve earnings for the hospital.
  • Quality: The average number of readmissions at 30 days decreased from 7% to 5%, from 11% to 6.1% at 60 days, and from 13% to 7.7% at 90 days. This is despite the fact that fewer and fewer patients were being sent to rehabilitation facilities. More robust quality metrics–such as patient function, quality of life–are important to assess, but were not provided. Since NYU was liable for all costs incurred up to 90 days following discharge, BPCI introduced a new financial incentive to thwart readmissions. As part of the care redesign work noted above, NYU instituted post-discharge protocols and hired care managers to follow-up with patients after the left the hospital.

No matter how you slice it, this program seems to be a home run. Medicare (and US taxpayers) saved money. NYU probably made a bunch of money (though we don’t know how much they spent on the care redesign). Patients went home sooner and had to head back to the hospital less often. The only losers are inpatient rehabilitation centers.

3 thoughts on “Early Results from Medicare’s Bundled Payment Initiative

  1. This is a great article and very encouraging for the viability of bundled payments. You make a great point in identifying inpatient rehab centers as the losers in this new scheme, and it will be really interesting to see how they respond to being cut out of the care process for these patients. Ideally, they would reorganize around the care they deliver for conditions that really do require and benefit from inpatient rehab, resulting in more IPU-like entities that deliver higher quality focused care. I’ll also be curious to see if there will be growth in the number of outpatient rehab facilities and home physical therapy practices, with pressure to shift to lower cost settings of care. I also completely agree that patient reported outcomes are an important missing piece of the puzzle here. Hopefully we’ll start to see more of that as the BPCI experiments continue.

  2. I agree, I would really like to see patient-reported outcomes to ensure that quality didn’t suffer in this project. Assuming quality didn’t decline, though, I think the most instructional part of this case is the relationship between the hospital and the rehab center. As we saw in Denmark and Sweden, inpatient rehab for knee replacements has clearly been ripe for cost savings. And when the rehab center is independent from the hospital and the hospital gets the check, it’s easy for them to cut out rehab services. Such a maneuver, however, would be much more difficult if hospital and rehab services were under the same umbrella. NYU Langone, for example, made relatively little headway in cutting costs from internal processes in this case (only $1222 for what was a roughly $28,763 surgery).

    There are undoubtedly be other post-acute services like rehab that can be cut using bundled payments, but I expect it will be hard to replicate this case’s cost savings for processes that are controlled under the same umbrella. After all—like you said—we already had that. They were called DRGs.

  3. This is great news! One question that emerges, across all these changes in payment or delivery, is whether there isn’t some kind of Hawthorne Effect present that brings out providers’ “best selves” in the course of a difficult transition. It will be interesting to see how durable the outcome results are, as I think they would be first to slide when the excitement passes.

Leave a comment