Contributor: The Future of Accountability for Total Cost of Care

Michael McWilliams, Alice Chen, and Michael Chernew recently articulated the next frontier in total cost of care accountability: Eventually, there will not be a pure fee-for-service (FFS) comparison. In several counties in the country, this may already be the case. The combination of Medicare Shared Savings Program (MSSP) penetration and Medicare Advantage (MA) penetration is high enough that there is no longer a significant population of Medicare beneficiaries for whom neither health care providers nor MA plans are accountable for the total cost of care. Arriving at this desirable goal has profound implications.

MA rates combine lagging historical spending in traditional Medicare, policy choices altering that history (qualifying counties, Stars bonuses), and a bidding process. MSSP uses an annual determination of county-level spending in traditional Medicare to determine regional benchmarks. Direct contracting uses a more complicated version of the MSSP determination to do the same. On our current trajectory, this means that eventually, MSSP will both suppress MA rates and MSSP rates. This trajectory will eventually change the incentive from reducing the total cost of care to consolidation among health care providers to avoid this negative feedback loop. Yet a future where there is so much participation in value programs in traditional Medicare that pure FFS is functionally extinct is desired by the advocates of the transition to value.

In their paper, McWilliams, Chen, and Chernew propose solving this by moving to administratively set rates. These projections replace the retroactive experience of an FFS comparison group and provide certainty to Medicare on future costs with actual cost below the projection deemed savings in which the taxpayer should continue to share but primarily accrue to accountable care organizations (ACOs). The challenge with administratively set rates is how to make them relatable to health care providers. We need to incentivize greater health at a lower cost in a way that generates positive behavior changes at the practice level.

The first benchmarking methodology of MSSP only recognized what I term improvement value. Did an ACO reduce its health care spending compared to previous years? ACOs who already had lower costs were told everyone could improve and that their past efforts had no value. ACOs with higher costs made savings relatively easily, especially in areas of the country where the use of national update factors created positive arbitrage. It is no coincidence that early success in MSSP was geographically concentrated. This arbitrage was even more important than starting from high costs. As CMS refined MSSP, the program has recognized what I term relative value. Does a beneficiary’s use of one primary care practice over another result in lower costs? Relative value flips the script where efficient ACOs are given a head start by getting a bonus to the benchmark valuing their past efforts. Inefficient ACOs have their benchmark lowered and are told it is easier to lower costs from a high baseline. At the same time, CMS removed most of the national trend vs regional trend arbitrage from MSSP by shifting to the aforementioned county cost comparisons.

MA is a 100% relative value calculation where the cost experience of MA beneficiaries is not included in the benchmark at all. In our experience at Aledade, our primary care practices place great emphasis on relative value, while commercial payers understandably place greater emphasis on improvement value. MA, with its 100% relative value rates, is the exception as the government limits profit margins available to MA plans. As we look toward a future where pure FFS can no longer define either improvement or relative value, we still need to keep in mind the differences they have on health care provider behavior.

Much has been made of the slowdown in MSSP enrollment that occurred in 2019 and 2020. To us, the reasoning is obvious: disruption. While the changes in Pathways to Success were largely championed by the ACO community, CMS was so late getting them out that it had to hold a special midyear enrollment period in 2019. Then COVID-19 hit, and there was no 2020 enrollment period for 2021 starts. As I believe the reason for the slow down was primarily disruption, I am skeptical that the solution for reinvigorating growth in the program is more disruption. I would therefore address the intermediate concerns raised by McWilliams et al with the simplest modifications to the MSSP possible to add the least amount of disruption possible.

Problem 1: Small Practices Unwilling or Unable to Go to Risk

Most of the gain in value-based health care is from creating a positive financial incentive to lower the total cost of care. CMS created revenue-based risk first in Track 1+ and then in the Basic track within Pathways to Success. Currently, a low-revenue ACO can spend 10 years in the Basic track with only 8% of its Medicare revenue at risk. I believe it is unlikely that these levels of risk are what are holding practices back. I believe what is needed is mandatory virtual ACO participation. This model is the shallowest of shallow ends of the pool with no risk and no formal governance requirements. Yet it ensures everyone gets in the pool, no more sitting on the deck.

Problem 2: Higher-Cost Health Care Providers Not Participating

When CMS introduced a positive incentive for health care providers generating relative value, it introduced a penalty for higher-than-average-cost health care providers. There is a feeling of fairness here. However, I suggest that it does not further the goal of reducing wasteful health care spending. In order to believe this is a money-saving policy, CMS would need to believe that the forgone savings from high-cost providers choosing not to participate are less than the penalties paid by those high-cost providers who do participate. Given the decline in participation, it seems unlikely that this balance is currently working in CMS’ favor. CMS should remove the penalty for regional efficiency and add the recommendations of McWilliams et al to encourage alternative payment model (APM) participation through other policies.

Problem 3: Lack of Options in MSSP

Recently, I, with others, outlined a strategy to use MSSP as an innovation platform. The essential packages to add to MSSP are primary care capitation, incentives to increase the number and equity of primary care relationships, value-based benefit design, and a “Next Generation” ACO package. The Next Generation ACO package would include a discount instead of shared savings (though a higher one than in Next Generation to ensure savings), the use of TIN/NPI combinations, and Medicare facilitated affiliations affected downstream payments. None of these packages would include a different benchmarking methodology as our goal is to minimize disruption to increase participation.

What About Risk Adjustment?

MSSP already has the strongest risk intensity protections of any program. I agree that a better risk model that is less sensitive to the effort spent on the inputs to the model is needed. I note that MSSP has taken the only meaningful step to improve the model in the last decade with its eligibility-specific normalization factors. It is not perfect, but MSSP already does it better than anyone else, so no fundamental disruption is needed while we wait for the development of a better risk model. MedPAC and others have advocated for change in MA. Such changes are beyond the scope of this editorial.

What About MA?

MA has always been a 100% relative value opportunity for MA plans. For the subset of health care providers taking accountability for total cost of care under MA plans, that accountability is similar to what it is in MSSP. MA plans calculate a historical medical loss ratio (MLR) for a group of health care providers, and the group then negotiates for a few percentage points to be added to their historical MLR, functionally equivalent to the MSSP regional bonus. I note that a majority of MA beneficiaries receive health care services paid on a purely FFS basis. If APMs are defined as paying health care providers differently, there is at least much growth in APMs needed in MA as in traditional Medicare. I see opportunity in how MA contracting has evolved to be similar to MSSP to bring them even closer together and encourage even greater participation. Risk adjustment challenges in MA are well documented and need to be addressed beyond the APM context.

Getting Beyond FFS

First and foremost, to get beyond FFS, we need to grow accountability of total cost of care. The two biggest opportunities are to increase participation in MSSP and increase health care provider accountability for total cost of care in MA plans. By stabilizing MSSP (not disrupting it yet again) and adding tools for more innovation under its umbrella, we can restart MSSP growth by making it simple for any health care provider who is taking accountability for total cost of care for one Medicare population to do so for all Medicare populations.

What Happens When We Get There?

Administratively set benchmarks can reward both improvement and relative value. The starting point can be bonuses for efficiency just as it is currently. McWilliams et al propose just using an ACO’s MSSP benchmark as the starting point when the transition is made, thereby locking in the MSSP bonus and penalty if that is not removed. Improvement value comes in the form of beating the administrative updates. The hard part is how do we get the correct administrative update for New York and the right update for Lafayette, LA, and the right update for Los Angeles, CA. We know that updates that were universal for the nation in MSSP created arbitrage across the country, rewarding some health care providers with savings not driven by their efforts and making savings nearly impossible for other health care providers despite their efforts. CMS solved this problem in MSSP by using local FFS comparison trends. In a local market with multiple ACOs and multiple MA plans, how do we administratively set rates that are localized enough so that they do not drive consolidation? This balance is the challenge of the next few years as we look forward to the day when pure FFS goes away, and we truly have made the transition to value.