Medicare APMs at a Crossroads

ABSTRACT

Reaching the goals set by the Health Care Payment and Learning Action Network requires an unyielding and unrelenting focus on encouraging providers to adopt advanced alternative payment models (APMs). Many of these models will continue to be voluntary because they either are in early stages or have not yet proven their effectiveness. The models that have proven their effectiveness should become permanent, comprising the new way that providers are paid in the Medicare program. Either way, getting today’s high performers into those programs and keeping them engaged to continue to innovate and set new benchmarks is as important as attracting and improving the performance of poorer performers. That will require a shift in Medicare’s policy on pricing and evaluating APMs.

Am J Manag Care. 2021;27(9):e290-e292. https://doi.org/10.37765/ajmc.2021.88624

Takeaway Points

Alternative payment models (APMs) are the essential ingredient to delivery system reform. Medicare has launched many APMs, but doubts about their effectiveness have been raised. However, that is because of the way in which they are implemented, not because the models themselves are flawed. One important flaw is the way target prices are set. Overall, we recommend specific changes that can apply to private-sector APMs as well:

• stop changing the goal posts to reduce program uncertainty,
• include an explicit margin to make continued investments in process improvement possible, and
• consider using approved clinical models to set base costs of care.

When the Nobel Prize was awarded to Daniel Kahneman in 2002, in large part because of his development of prospect theory with Amos Tversky,¹ most of the world that had been trained in classical economics realized that behavioral economics mattered, perhaps more so than classical economics. And yet, in our world of health care incentives, exemplified by alternative payment models (APMs), the application of behavioral economics to program design seems lacking. This is even more true with advanced APMs that include unmeasurable downside risk. Contrary to the colloquial belief that providers must have “skin in the game,” such incentive design has a negative and predictable result on human behavior. So why do such models persist?

Advanced APM designers would greatly benefit from taking into account the ample economics literature on the difference between risk and uncertainty.² This is especially true when considering the phenomenon of “loss aversion,” in which the tendency to prefer avoiding losses overwhelms acquiring equivalent gains. Combined, these two explain in large part the hesitancy of providers to participate in downside-risk APMs. Yet, in order for the nation to continue its move from volume- to value-based payments, we must attract more providers to these APMs, especially physician groups. And to do that, we must follow evidence-based guidelines on incentives,³ just as we might for a course of treatment.

Throughout 2020, as the COVID-19 pandemic ebbed and flowed repeatedly, the limitations of fee-for-service (FFS) payments and site-of-service restrictions became very apparent in the Medicare program; this prompted appropriate calls for greater velocity in substituting FFS payment with other forms of payment. This is especially important for primary and specialty care practices and has led to the launch of a focused effort on the part of the Health Care Payment and Learning Action Network (HCP LAN).⁴ Simultaneously, other groups of APM experts have come together to better understand the limitations of existing programs and to make recommendations.

One such group was convened by Leavitt Partners, a health care consultancy. It focused on ways in which Medicare could make existing episodes-of-care programs more sustainable and build on the lessons learned to expand the scope and reach. Overall, the recommendations focus on how the current Bundled Payments for Care Improvement (BPCI)/BPCI Advanced (BPCI-A) programs—the federal government’s most mature forays into episode-based payment—could become a permanent way to pay for hospitalizations stemming from acute conditions, as well as on how Medicare can expand bundled payments to condition-based episodes. Although the recommendations include specific technical points for each of the 2 main prongs for program expansion, a few cut across and have even broader application to any advanced APM.

A Sustainable Pricing Policy

As the saying goes, “No margin, no mission.” When providers are given a mission to deliver higher-value care, they need a margin to keep making improvements. A pricing policy that is solely based on the “ratchet effect”⁵ does not make much sense for the program sponsor because it eventually leads the better-performing providers to stop trying to improve. This is especially true in downside-risk arrangements in which the floor of savings and the beginning of losses come into contact. It holds true as well for both mandatory and voluntary payment programs because, in both, the program sponsor should want the better-performing providers to keep reaching toward the efficient frontier,⁶ eventually resetting the best-in-class benchmark. As such, we recommend that APMs include an explicit margin for providers and make specific proposals for how that could be implemented.

However, even with an explicit margin, the primary challenge with creating a baseline target price is the basis for that baseline. The sole reliance on historical claims data, especially for condition-based episodes, reflects historical practice patterns and not an idealized clinical model for the delivery of high-value care to a patient, focused on generating the best possible outcomes. Building the benchmark up from the components of that clinical model would create a clinically defensible baseline and avoid perennializing existing, often suboptimal, care patterns, which can have the perverse effect of overly rewarding poor performers at the expense of high performers.

Although a clinically sound benchmark and an explicit margin create a solid base on which an advanced APM should be built, that foundation can quickly be destroyed with a combination of 2 variables: the trend rate and the discount. To date, Medicare has set the up-front discount that it applies to bundles uniformly, irrespective of the participant’s historic performance relative to peers. This has the effect of making it harder for higher-value providers, relative to lower-value providers, to get incremental savings. We therefore recommend a more nuanced approach that would reduce the discount for high-value providers. Of course, these higher-value providers and their counterparts are also affected by the direction and slope of trend rates and by when those rates are set. When they are set retrospectively, trend rates create significant uncertainty in program outcomes and can quickly lead participants to drop. When the rates are set prospectively, the uncertainty is eliminated, but that policy can lead the program sponsor to be overaggressive and set strong negative trends, attempting to capture as much of the savings as possible. This increases the risk of losses for the better performers and reduces the potential for their participation. A balanced approach—one in which trends are set reasonably up front, but with a limited ability to adjust them retrospectively—is a better and recommended approach.

A Sustainable Evaluation

One of the challenges for Medicare in adjusting trend rates is assembling a reasonable peer group of nonparticipating providers whose performance can be compared with that of the participants. This has proven to be very difficult, and increasingly so as APMs have proliferated. Expanding bundled payments to condition-based episodes will further complicate the process. That is why we propose having real control groups with providers randomly assigned as nonparticipants. Although that will not fully account for potential spillover effects,⁷ it will reduce the reliance on propensity matching as the sole solution. In addition, the effect of APMs should be evaluated over longer periods of time to make sure that there isn’t a regression back to prior performance.

But evaluating the effect of a new APM should go beyond establishing a rigorous control group. Participants should be motivated to significantly improve patient outcomes and to be held accountable for doing so. That means moving away from an excessive reliance on process measures and toward patient-reported outcome measures (PROMs). Using PROMs is not a new concept, but it has not been put into practice. One reason is the challenge of data collection and adequate sample sizes, which is why we recommend that Medicare take the bold step of leading a coalition of public and private sector payers to collect and measure PROMs at scale and nationally.

As part of that effort, Medicare should also lead the way in baking the reduction of social and racial inequities into improved outcomes. Providers cannot be responsible for addressing all the social needs that are barriers to positive clinical outcomes, but they can add community and clinical social workers to their practices and create more formal links with community-based organizations. Medicaid agencies are already encouraging this practice and so should Medicare through its advanced APMs, provided that it applies the recommendations on a sustainable pricing policy.

Making Change Permanent While Continuing to Innovate

There is no stronger signal on the resolve to move away from FFS than making APMs a permanent part of Medicare payment and fundamentally changing the way in which FFS operates today. That’s why we recommend making BPCI permanent and mandatory, converting the current payment of acute inpatient stays to a single bundled payment that would cover the inpatient stay and associated professional services, as well as all costs of care for 30 days after discharge. This change in payment policy, which could be phased in over the next 5 years (initially excluding those participating in BPCI-A), could be accompanied by the elimination of the 30-day readmission penalty because the penalty would be baked into the global payment.

Importantly, this would free up the Center for Medicare & Medicaid Innovation to pursue its mission to test new and innovative payment models, which is what we propose with the introduction of bundles for condition-based episodes. Providers also need the freedom to innovate, which is often hampered by rules and regulations that have been designed to manage an FFS payment system. That’s why we also propose that those in advanced APMs—whether those APMs are the new normal for Medicare payment or being tested—can apply for global waivers that would include all the individual waivers that they must apply for today.

Conclusions

Reaching the goals set by the HCP LAN requires an unyielding and unrelenting focus on encouraging providers to adopt advanced APMs. Many of these models will continue to be voluntary because they either are in early stages or have not yet proven their effectiveness. The models that have proven their effectiveness should become permanent, comprising the new way that providers are paid in the Medicare program. Either way, getting today’s high performers into those programs and keeping them engaged to continue to innovate and set new benchmarks is as important as attracting and improving the performance of poorer performers. That will require a shift in Medicare’s policy on pricing and evaluating APMs. At this important crossroads, we hope that Medicare will consider our recommendations.

Author Affiliations: SignifyHealth LLC (FdB), Norwalk, CT; America’s Physician Groups (VR), Los Angeles, CA; The University of Texas at Austin Dell Medical School (KB), Austin, TX.

Source of Funding: None.

Author Disclosures: Dr Bozic is second vice president on the American Academy of Orthopaedic Surgeons Board of Directors; has consulted for Harvard Business School, Carrum Health, and CMS; has received grants from the Agency for Healthcare Research and Quality; receives UpToDate chapter royalties from Wolters Kluwer; and is on the editorial board of Slack Incorporated. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (FdB, VR, KB); drafting of the manuscript (FdB, VR, KB); and critical revision of the manuscript for important intellectual content (FdB, VR, KB).

Address Correspondence to: Francois de Brantes, MS, MBA, SignifyHealth LLC, 800 Connecticut Ave, Norwalk, CT 06854. Email: fdebrantes@gmail.com.

REFERENCES

1. Kahneman D, Tversky A. Prospect theory: an analysis of decision under risk. Econometrica. 1979;47(2):263-291. doi:10.2307/1914185

2. Tversky A, Fox CR. Weighing risk and uncertainty. Psychol Rev. 1995;102(2):269-283. doi:10.1037/0033-295X.102.2.269

3. de Brantes F, Rosenthal M, Struijs J. Ensuring the adoption of the health care warranty: a well-defined model to resolve issues with risk and uncertainty. NEJM Catalyst. July 31, 2020. Accessed March 25, 2021. https://catalyst.nejm.org/doi/full/10.1056/CAT.20.0344

4. Healthcare Resiliency Framework. Health Care Payment Learning & Action Network. Accessed March 25, 2021. https://hcp-lan.org/resiliency-collaborative/framework/

5. Weitzman M. The “ratchet principle” and performance incentives. Bell J Econ. 1980;11(1):302-308.

6. CorredoiraRA, Chilingerian JA, Kimberly JR. Analyzing performance in addiction treatment: an application of data envelopment analysis to the state of Maryland system. J Subst Abuse Treat. 2011;41(1):1-13. doi:10.1016/j.jsat.2011.01.006

7. Navathe AS, Boyle CW, Emanuel EJ. Alternative payment models—victims of their own success? JAMA. 2020;324(3):237-238. doi:10.1001/jama.2020.4133