​

The Medicare Shared Savings Program (MSSP) is the nation’s single largest initiative to move the US health care system from volume to value. In only six years, the MSSP has grown to include 561 participating accountable care organizations (ACOs) serving 10.5 million assigned beneficiaries – nearly one-third of traditional Medicare beneficiaries with Part A and Part B coverage.

Under the MSSP, ACOs serve a defined population of Medicare beneficiaries with an annual spending target called a “benchmark” and a series of quality thresholds. ACOs that spend less than the benchmark share the savings with the federal government. ACOs have delivered superior quality to Medicare patients and have achieved savings relative to their spending targets and to other health care providers that do not participate in the MSSP.

On August 9, the administration issued a proposed rule that makes significant changes to the MSSP. It is a complex rule with many technical policy changes, some of which would improve the stability and predictability of the program. But the central proposed policy change would dramatically accelerate the requirement that ACOs take on downside financial risk.

We agree that ACOs should take on greater risk over time. However, we are concerned that the speed of the proposed transition, combined with proposed reductions in shared savings rates, could stall the ACO movement. There is wide agreement that the proposed rules would reduce the number of ACOs but no agreement on the scale of reductions or the impact on total savings over 10 years, which could be significant.

One of the assumptions of the administration’s proposal to accelerate risk is that the MSSP has not saved money for Medicare, while other Centers for Medicare & Medicaid Services (CMS) initiatives that require downside financial risk have generated savings. But both the new research presented below and the recently released 2017 MSSP results from CMS tell a different story. According to CMS, MSSP ACOs saved $1.1 billion in 2017, with net savings of $314 million after accounting for shared savings payments earned by ACOs (See Exhibit 1). The ACOs that joined the program in 2012 through 2014 accounted for the majority of savings. This illustrates that ACO performance improves over time and that ACOs need time before clinical restructuring can generate program savings. Also of note: One-sided MSSP Track 1 ACOs generated more 2017 savings per beneficiary than those bearing risk under MSSP Tracks 2 and 3.

Exhibit 1: Benchmark-Based Savings/Loss To Medicare From MSSP ACOs

Source: Authors calculations based on CMS 2017 Shared Savings Program ACO PUF File

Despite the positive 2017 results, gauging MSSP performance based on calculations using administratively derived spending targets (benchmarks) is simply not an accurate way to measure overall program savings. In fact, the published academic research on MSSP performance points to much higher savings than are suggested by the benchmarks.

In this post, we discuss the challenges of using benchmarks to calculate ACO program savings, identify peer-reviewed studies that estimate ACO savings based on widely accepted statistical methods, and introduce a new study commissioned by the National Association of ACOs (NAACOS) that uses research methods similar to those peer-reviewed studies. The new study shows MSSP savings in 2013 – 2015 of more than $1.8 billion, nearly double the amount reported by CMS. Finally, we discuss the implications of the proposed MSSP rule and offer suggestions that would make the rule more attractive to current and potential future ACOs.

Calculating Medicare ACO Savings: The Benchmark Bias

New payment models that move away from traditional fee for service require a mechanism for measuring the performance of participating health care providers and structuring rewards or penalties based on that performance. In the MSSP, CMS calculates an initial risk-adjusted spending benchmark for each ACO based on its historical spending for a group of attributed Medicare beneficiaries; it then trends this benchmark forward to the current program year based on the national average growth in Medicare spending per beneficiary. (In 2017 CMS began to phase in benchmarks updated with regional trend rates).

If an ACO’s spending in the program year is less than the benchmark and the ACO meets MSSP quality thresholds, it earns a shared savings payment that is generally 50 percent of the calculated savings. ACOs must achieve a savings rate of at least 2 percent – up to 3.9 percent for the smallest ACOs – before they are eligible for bonus payments. Therefore, not all ACOs that save money receive bonus payments.

CMS calculates total MSSP savings as the sum of total savings for ACOs with spending below the benchmark plus the sum of spending above the benchmark for ACOs that exceeded it. Using this method, CMS estimated MSSP savings of $954 million between 2013 and 2015. During this period, ACOs that saved money earned $1.3 billion in shared savings payments. CMS concluded that on a net basis, the program increased Medicare spending by $344 million between 2013 and 2015.

Researchers have determined, however, that using benchmarks to calculate MSSP saving systematically understates true program savings. This occurs for several reasons. ACO benchmarks have been trended forward using a national average per-beneficiary amount, but Medicare spending growth varies substantially across geographic areas due to underlying market factors, and ACOs tend to be located in areas of higher Medicare spending growth. Therefore, an ACO could significantly outperform its regional peers but still lose money based on CMS’s accounting. Exacerbating this problem is the fact that CMS does not adjust the benchmarks to account for the growing burden of illness as continually enrolled beneficiaries age during each three-year contract period, even though this results in higher actual spending.

Benchmarks and other types of spending targets serve a useful role in administering new payment models but produce different results than research-based evaluations. This may confuse stakeholders and lead to misinformed policy decisions. The Center for Medicare and Medicaid Innovation’s (CMMI) Next Generation ACO, Pioneer ACO, and Bundled Payment for Care Improvement initiatives have all been formally evaluated by researchers. However, CMS has never commissioned a formal evaluation of the MSSP.

Calculating Medicare Shared Savings: The Research-Based Approach

Research evaluations of actual payment reforms are essential for guiding future policies governing alternative payment models. Several academic research teams have analyzed MSSP savings, using standard statistical methods to evaluate the financial performance of ACOs by estimating what spending would have been in the absence of the ACO program. Colla and colleagues found that MSSP ACOs reduced spending by 1.3 percent in 2013. A 2016 study by McWilliams and colleagues estimated that the MSSP saved $867 million during 2013 and 2014, resulting in overall savings of $213 million after subtracting shared savings payments earned by the ACOs. More recently, McWilliams and colleagues found total MSSP savings of $704 million in 2015, with net savings to Medicare of $145 million.

In an effort to improve the evidence available to policymakers and other key stakeholders, NAACOS commissioned the firm of Dobson|Davanzo to conduct a statistical analysis of MSSP savings from 2013 through 2015. The Dobson|Davanzo study found total MSSP savings of more than $1.8 billion for this period using statistical methods similar to McWilliams and colleagues – which is nearly double the savings calculated by CMS for 2013 to 2015 based on the benchmarks, as shown in Exhibit 2. The Dobson|Davanzo study used difference-in-differences analysis and a Medicare claims database with 100 percent of ACO-attributed beneficiaries from 2011 – 2015, and more than 80 percent of beneficiaries who were eligible but not attributed to an ACO. Based on that analysis, net savings to Medicare after accounting for shared savings payments earned by the ACOs was $542 million (See Exhibit 3).

Exhibit 2: Total MSSP Program Savings For 2013 – 2015: Comparison of Dobson|Davanzo Estimates with CMS Benchmark Calculations

Source: Dobson|Davanzo study, CMS MSSP ACO 2013-15 results

Exhibit 3: Net MSSP Program Savings for 2013 – 2015 after Shared Savings Bonuses: Comparison of Dobson|Davanzo Estimates with CMS Benchmark Calculations

Source: Dobson|Davanzo study, CMS MSSP ACO 2013-15 results

ACO Benefits Beyond Direct Savings

The MSSP’s direct savings to Medicare are relatively modest compared to overall Medicare spending, but they are also encouraging, especially since individual ACO performance improved with length of time in the program. In addition to direct savings, the MSSP generates substantial indirect savings to Medicare due to the spillover effect of delivery system changes on fee-for-service Medicare spending. In markets where at least 10 percent of the beneficiaries are attributed to ACOs, Medicare spending grew more slowly than the national average. According to CMS’s own estimates, Medicare savings including spillover effects in 2016 were $1.8 – $4.2 billion, or about 0.5 – 1.2 percent of traditional Medicare spending.

As MSSP reduces spending in traditional Medicare, it also reduces spending in Medicare Advantage, where government payments are tied to regional Medicare fee-for-service spending. According to one estimate, the 0.7 percent net spending reduction by the MSSP program in 2014 would lower Medicare Advantage spending by roughly $272 million.

The MSSP program also directly benefits care for Medicare beneficiaries by creating incentives for ACOs to coordinate care and improve quality. According to CMS, ACOs have consistently achieved higher average performance rates compared to medical group practices reporting similar quality measures; ACOs in the program from 2013 through 2016 improved their quality scores by 15 percent on average. Comparable quality improvements have been documented by independent researchers.

Potential Impact Of The Proposed MSSP Changes On Program Participation

The MSSP is a voluntary program and growth in program participation has greatly exceeded expectations. Most ACOs opted to join MSSP Track 1, which offers the opportunity to earn up to 50 percent of the Medicare savings they generate. Participants in Track 1 do not incur financial penalties if their spending exceeds the CMS benchmark. But these organizations are solely responsible for the cost of establishing and operating their ACO with no additional financial support from CMS. Running an ACO requires new clinical and administrative personnel, data systems and analytics, and investments in developing new clinical protocols and workflows; this requires millions or even tens of millions of funding annually, depending on the size of the ACO.

About one in five ACOs in the MSSP has voluntarily entered a track with downside financial risk. Organizations that are developing accountable care models, especially for the first time, face many uncertainties about the program and their own ability to earn shared savings. They may be hesitant to take on risk.

If enacted as written, the CMS proposal would reduce the number of ACOs participating in the MSSP and also slow the growth of new entrants. It is impossible to accurately predict the magnitude of the reduction. CMS estimates that the MSSP would shrink by 109 ACOs over the next 10 years. We are concerned that the magnitude could be much greater. NAACOS recently surveyed MSSP Track 1 ACOs with terms ending in 2018 about their interest in continuing the program if required to take on two-sided risk. Seventy-two percent of respondents reported they were likely or highly likely to leave the program.

The MSSP has grown rapidly in its first six years at a rate of nearly 100 net new ACOs annually. Even if the growth rate slowed to 50 new ACOs annually over the next 10 years, there would be more than 1,000 ACOs covering more than 20 million beneficiaries by 2028. CMS’s own estimate that the MSSP program would shrink by 109 ACOs over the next 10 years suggests both a reduction in current ACOs and a drastic slowing of new entrants, which would leave the MSSP with only 452 ACOs in 10 years. Given compelling evidence about direct MSSP savings, the fact that ACO performance improves over time, and substantial savings due to spillover effects, policies that lead to a smaller program in ten years rather than a potential doubling of participation in accountable care would be a huge missed opportunity for the nation.

There are three aspects of the proposal that are most likely to adversely impact ACO participation. First is the reduction of the shared savings opportunity from the current 50 percent to 25 percent in the initial year of the new BASIC track. Establishing a successful ACO requires significant investment and potential new entrants would be concerned that recouping their investment would be impossible with such a low sharing rate.

Second is the shortening of time before new ACOs must accept risk from six years to two. This would deter organizations that lack sufficient capital reserves or are simply not ready to take on down-side risk from participating.

Third is the aggregate 3 percent limit over five years on recognizing changes in beneficiaries’ risk scores when calculating the benchmarks. CMS proposes to extend the MSSP agreement period from three to five years, and the burden of illness of Medicare beneficiaries can change rapidly over a five-year time period. While we support the proposed extension of the agreement period, we are concerned that the cap on recognizing changes in the clinical risk of attributed beneficiaries creates yet more financial risk for the ACOs.

A Path Forward

Addressing the three issues described above would be a major step towards promoting both new entrants and continued ACO participation in the MSSP. We suggest the following changes. First, continue the current 50 percent shared savings rate for the entire length of the new BASIC track to ensure that the model remains financially feasible for new entrants. Second, allow all new ACOs to remain in a shared savings model without downside risk for at least three years and allow an additional two years for ACOs that meet specified savings and quality goals in the initial three years. Third, allow the beneficiary risk score to vary by up to 5 percent over the five-year enrollment period.

The study released today by NAACOS as well as earlier studies using similar methodology by Harvard University and others demonstrate clearly that the MSSP has generated direct savings to Medicare. Furthermore, the MSSP has created spillover effects in both fee-for-service Medicare and Medicare Advantage that likely result in billions in savings. This underscores the importance of maintaining a robust Medicare ACO program that continues to grow over time.

We agree with the administration that ACOs need to evolve and take on more financial risk, but the administration’s own estimates suggest that rather than growing, the Medicare ACO program would shrink by 20 percent over the next decade under its proposed rule. That would be a tragedy. The changes we propose would go a long way towards improving incentives for growth of Medicare ACOs while achieving the administration’s goal of a faster movement to risk.

Our nation urgently needs to transform its health care system to deliver better care at lower cost. Accountable care organizations are the best chance we’ve got to get there. Let’s invest in policies that accelerate this movement rather than slow it down.