This article first appeared in ajmc.com. It is co-authored with Sean Cavanaugh, MPH, and Travis Broome, MPH, MBA.
The Medicare Shared Savings Program (MSSP) has seen its growth slow, but CMS has an opportunity to act on proposals that would address benchmarking and more fairly allocate savings to accountable care organizations in the program.
In August, Administrator Chiquita Brooks-LaSure and other leaders from CMS shared their priorities for the next 10 years. In particular, they called out the importance of reducing inequities in health care, striving to get all Medicare and Medicaid beneficiaries in accountable care relationships, and designing financial incentives that will encourage participation in new payment and service delivery models.
Unfortunately, one of the programs with the most documented success for helping to achieve these aims, the Medicare Shared Savings Program (MSSP) or, more commonly, accountable care organizations (ACOs), has seen its growth slow partially as a result of regulatory changes that delayed one new applicant cycle, as well as the cancellation of another cycle due to the COVID-19 pandemic. If the new CMS leadership is looking for levers to pull to fulfill the goals it is setting related to reducing health disparities and promoting value-based care, then it needs to improve the operations and incentives of the MSSP, as well as address some unintended consequences of earlier policy decisions that have slowed its progress.
The good news is that CMS has an opportunity before it to act—by pushing forward with 2 provisions discussed in the proposed rule for the 2022 Medicare physician fee schedule that address challenges with benchmarking in MSSP that deprive ACOs of a portion of the savings they generate.
First, CMS must address a problem called the “rural glitch.” An ACO’s benchmark is established by taking the historical costs of its aligned beneficiaries and trending those costs into the performance year based upon local cost increases. Unfortunately, CMS includes an ACO’s own beneficiaries in calculating the local cost increase. So, an ACO that reduces costs in the performance year (relative to other providers) is also reducing the rate of local cost increase and bringing down its own benchmark. And, perversely, an ACO that increases costs actually drives up the local cost trend and raises its own benchmark. In rural counties, where an ACO may have 30% of all Medicare beneficiaries, this effect can be significant. The rural glitch violates the principles articulated by CMS leadership because it means (1) fewer rural beneficiaries will benefit from the better care at lower cost being achieved in the MSSP and (2) fewer rural providers will participate in the MSSP.
The rural glitch also affects fairness across ACOs. The suppression of the savings rates of ACOs varies widely based on market share. The greater the market share of an ACO, the more an ACO lowers its regional trend. Over 5 years, an ACO with 20% market share will earn over $4,000,000 less in shared savings than one with 5% market share, everything else held constant. Fortunately, a simple algebraic calculation outlined by CMS in the proposed rule can be used to back out the effect of an individual ACO on the local trend. By using data already collected by CMS on ACO trend and market share, CMS can create a measure of regional inflation that excludes the effect of the ACO itself.
Second, the Hierarchical Condition Categories (HCC) risk adjustment in Medicare is fundamentally broken. As MedPAC and others have documented, the HCC system rewards Medicare Advantage plans for documenting a greater number of diagnoses, which has led to significantly higher payments to plans for many years compared to Medicare. From the very inception of MSSP, CMS has been wary of rewarding upcoding in the MSSP, so CMS only recognizes up to 3% growth in HCC risk scores across a 5-year period for an ACO. This is much lower than the increase in HCC risk scores of Medicare Advantage plans, which increase by a compounded annual growth rate of 1.4% per year. CMS has also not capped the risk growth of the population to which an ACO is compared. This means that if an ACO and its comparison population both experience 6% growth in costs and 6% growth in risk, the ACO will have risk-adjusted cost increase of 3% but the comparison population will have risk-adjusted cost increase of 0%, simply because one is capped and the other is not.
The original purpose of this policy was to deter ACOs from upcoding, not to unfairly deprive ACOs of the savings they are generating. This is not a theoretical problem. In 2021, 13.6% of Medicare beneficiaries assignable to the MSSP live in a county with a risk ratio above 1.03, based on our analysis of Medicare data. However, the most vulnerable Medicare beneficiaries, those who are dually eligible for Medicaid, are twice as likely to live in a “capped-out” county as beneficiaries who only have Medicare. CMS could fix this problem by changing the policy to allow an ACO’s risk adjustment to vary up to 3% relative to the changes in risk in its region (instead of an absolute 3% cap) or, second best, simply cap regional risk growth at 3% growth to parallel the policy applied.
In both cases, CMS has recognized the problem and discussed viable solutions. But, disappointingly, CMS has also signaled that it will wait for future rulemaking to fix these problems. Why wait? The current rulemaking process for the 2022 Medicare physician fee schedule provides the perfect opportunity to right these wrongs. Additionally, as CMS continually improves MSSP benchmarking, it can consider using the MSSP as the chassis upon which to test other models such as primary care capitation and specialist models.
CMS proposes improvements for most Medicare payment systems each and every year. There are now more than 10 million Medicare beneficiaries in the MSSP, with ACOs accountable for over $111 billion in total Medicare spending. We encourage CMS to show the same rigor and diligence in making annual improvements in the MSSP that it does for hospital, physician, durable medical equipment, and other payment systems.