My Ten Wishes for BPCI 2.0

Posted by: Larry Sobal on Thursday, August 10, 2017
My Ten Wishes for BPCI 2.0


Advanced BPCI? BPCI 2.0.? Better BPCI? Who knows what Centers for Medicare and Medicaid Services (CMS) will call the “new” version of the Bundled Payment Care Initiative program that is supposed to launch in 2018. All we know at this time is that CMS is expected to make an announcement about the next iteration of BPCI soon.

Before I look forward and share my hopes for the next phase of BPCI, which I’ll call BPCI 2.0, let’s look back to see how BPCI 1.0 played out.

In BPCI 1.0, which launched in August, 2011, interested organizations could choose to participate in any of 48 bundles (15 of which were for cardiovascular conditions) in 4 models (Model 1-4), differentiated largely by scope of providers and services included in the bundle.

By October 1, 2013, some BPCI 1.0 participants entered into “Awardee Agreements” with CMS, at which point they began bearing financial risk. In November 2013, CMS offered a second Open Period for current Awardees to add additional Episode Initiators or clinical episodes in BPCI 1.0 Model 3. The second cohort of participants began their performance period on January 1, 2014. CMS offered a third Open Period in February 2014 seeking additional organizations to participate in BPCI1.0 Model 3. The Open Period ended on April 18, 2014, and resulted in many new participants joining the BPCI 1.0 initiative through the summer and fall of 2014. Additional cohorts began on January 1, April 1, and July 1, 2015.

In December 2014, a timeline for transition to Phase 2 of BPCI 1.0 was instituted. According to the timeline, every episode initiating organization, regardless of whether the Episode Initiator was directly bearing risk (as an Awardee) or was participating under an Awardee Convener, had to transition at least one Clinical Episode to Phase 2 by July 1, 2015, in order to remain in BPCI 1.0. The transition of all clinical episodes for all participants into Phase 2 was completed on September 30, 2015, at which point Phase 1 of BPCI 1.0 ended. Phase 2 was previously scheduled to end after each participant completed a three-year period of performance for each clinical episode entered into Phase 2. The BPCI 1.0 initiative is being extended until September 30, 2018, for all BPCI 1.0 Model 2, 3, and 4 Awardees that choose to sign an amendment extending their period of performance for all clinical episodes for up to 2 years.

At the beginning, more than 6,500 Medicare providers had signed up to participate in BPCI 1.0, making it one of the largest demonstrations in Medicare history. However, 13 of the original 24 Awardees in BPCI 1.0 Model 1 had terminated their agreements by July 15, 2015, for a variety of reasons, among them a feeling that the rules were not clear, were changed each year, or that the prospects of return on investment simply weren’t attractive.

Overall, there were mixed results reported from BPCI 1.0, which causes me to reflect on how I feel BPCI 2.0 could be designed significantly better. Here is what I would like to see.

First, keep it simple. The original BPCI initiative tested four different models, rolled out in different phases, based on episodes of care that involved an inpatient hospital stay. One model focused on care provided during the hospital stay, while the other three models included post-acute care provided once the patient was released from the hospital. This was way too complicated to keep up with, let alone be able to draw accurate conclusions from. I recommend that BPCI 2.0 have only two models, one associated with a specific patient diagnosis or procedure and the other associated with post-acute care.

Second, I suggest one model not be triggered by an admission or DRG because BPCI 1.0 seemed to have a perverse incentive to over-utilize hospitals and ignored the rapidly growing, and often significantly less expensive, alternative options such as outpatient surgery centers. For example, in the town where I live, I can now receive a variety of joint replacements without ever stepping onto a hospital campus. Why would we not want to include those types of options and reward providers for appropriately using lower cost options?

Only 25 percent of those that showed interest and volunteered in the original BPCI 1.0 program actually moved forward once downside risk was introduced.

Third, there has to be some form of severity adjustment and accounting for outlier scenarios in BPCI 2.0. BPCI 1.0 relied on DRG assignment to distinguish severity, which is a poor way to do so. Without proper severity adjustment being taken into account to determine target pricing, two bad things can happen: providers may be less likely to manage the treatment of complex cases if their reimbursement is subpar, which could essentially block health care access for the sickest patients and; there is not a way to carve out patients who have multiple complications that no longer make them a fair candidate for a bundle. Furthermore, if you have particular drugs or services that are expensive and highly variable, it may make sense to exclude those from the bundles.

Fourth, only 25 percent of those that showed interest and volunteered in the original BPCI 1.0 program actually moved forward once downside risk was introduced. The downside risk of the bundled payments program included the fact that providers would be responsible for any spending above the contracted amount offered from the payers. Not surprisingly, many dropped out. I would like to see BPCI 2.0 incorporate a model that offers an easier path to taking risk so organizations can ease into the world of risk and build the required competencies and infrastructure as they go.

Fifth, success in any bundled payment model requires new levels of collaboration between hospitals, physicians and other providers. Therefore, it is critical that BPCI 2.0 include substantial Stark-related regulatory and legislative waivers to make it easy to innovate without running afoul of fraud and abuse laws, or at least providing attorneys clearer guidance on what is acceptable variation from currently stringent rules that are barriers to innovation.

Sixth, one of the complaints of the original BPCI was that the performance targets were constantly changing. This was because the BPCI program “re-based” the targets every quarter as a function of observed changes in national trends. These shifting targets were then not shared until at least six months after the performance period. That meant you had no idea ahead of time what the target was and whether you had an opportunity to gain or lose on your performance. Nobody wants to be in that kind of game, as it creates too high a level of uncertainty. Therefore, it is critical that there be clear and consistent rules related to benchmark prices and determination of savings, and that targets be established before a performance period begins and your reward or penalty reconciled back to that pre-determined target. Furthermore, the target pricing should include cost of living increases, which would increase predictability for model participants and encourage wider participation.

Seventh, the attribution methodology from BPCI 1.0 must change. Episode attribution has been an imperfect science at best, and is a significant risk for any physician considering participation in future iterations of voluntary bundle programs. So why are we trying to do the impossible? I propose attribution of episodes based solely upon Medicare Part B claims data, which has both the NPI and the TIN information and limits attribution to the TIN level only (i.e. not to a physician level).

Eighth, while we are already in a world where too many metrics are required to be reported, BPCI 1.0 had no thresholds required to receive reconciliation payments. In other words, some could be rewarded for reducing cost relative to the target price, even if quality of care deteriorated. What I propose is that rather than adding new metrics, BPCI 2.0 participants be required to demonstrate designated levels of performance on existing measures, whether that be MACRA, Value-Based Purchasing or another existing program. The point is to thoughtfully tie BPCI financial reward to quality performance but not create an onerous (or new) reporting process.

Ninth, BPCI 2.0 should be voluntary (only).

Tenth, BPCI 2.0 should offer participating physicians status as an Advanced Alternative Payment Model (and therefore a MIPS exemption) under MACRA/QPP.

Am I asking for too much?  Not if CMS wants BPCI 2.0 to work better than 1.0. Let me know if you agree or disagree.


Illustration: Lee Sauer


Larry SobalLarry Sobal is Executive Vice President and a Senior Consultant at MedAxiom. He has a 35-year background as a senior executive in medical group leadership, hospital leadership and health insurance. Larry consults, writes and presents on topics relevant to transforming physician practices and health systems. His weekly blog post comes out on Thursdays and can be accessed at



Great suggestions for the (hopefully) next iteration of BPCI. There is definitely much to be learned from the downsides and barriers of 1.0 that need to be fixed to maximize the true value of this potentially game-changing model.
Posted by: Scott Campbell on August 18, 2017 @10:41:02 am

Leave a Comment

Go Back