Skip to main content
White paper

Predictability vs. accuracy in MSSP benchmarks

3 February 2025

Historically, accountable care organizations (ACOs) participating in the Medicare Shared Savings Program (MSSP) have struggled to understand their emerging ACO performance because the final Medicare fee-for-service (FFS) expenditure growth rates (both regional and national) are not available until well after the performance year is over. The calendar year (CY) 2023 Medicare Physician Fee Schedule (MPFS) Final Rule introduced the concept of the Accountable Care Prospective Trend (ACPT) for MSSP agreement periods starting in 2024 or later. While this methodological change increases the predictability of the year-end benchmark, it also introduces the risk that the actual Medicare trends may differ meaningfully from the prospective trend.

What are the key components of the ACPT?

The ACPT development starts with benchmark year 3 United States Per Capita Cost (USPCC) estimates, which are then adjusted to exclude payments for indirect medical education (IME), disproportionate share hospitals (DSHs), and hospice claims—consistent with the expenditures included in the MSSP program. These national values are projected forward separately for both end-stage renal disease (ESRD) and aged/disabled (AD) populations to create a set of ”uniform annualized projected rate of growth” over each year of the five-year performance period.1 The ACPT is given one-third weight (versus a two-thirds weight for the current national-regional blended trend methodology) to the total trend calculation as seen in the formula below:

Trend Factor=(PY ACPT*1/3)+{[(National Trend Factor*National Weight)+(Regional Trend Factor*(1-National Weight)]*2/3}

USPCC trend estimates are typically announced in April in the Medicare Advantage Rate Announcement, but the ACPT for performance year 2024 was not released until November 2024.

Figure 1: Trend assumption comparison

Enrollment Type 2023 to 2024 Medicare FFS USPCC
Trend Estimates From 2025
Medicare Advantage Rate Announcement

Released April 1, 2024
2023 to 2024
ACPT Trends for
Performance Year 2024

In BNMRK reports produced
November 11, 2024
Authors’ Approximate
Best Estimate of
Nationwide FFS Expenditure Trend

As of 2024 Q3 CMS reporting
ESRD 3.9% 3.9% 8%
Aged/Disabled 3.5% 3.6% 9%

Despite being released in late 2024, the 2023 to 2024 ACPT assumption was set approximately in line with the 2023 to 2024 Medicare FFS trend estimates from the 2025 Rate Announcement. These estimates can be contrasted with the actual 2023 to 2024 expenditure trends for the nationwide MSSP assignable population. Based on a review of nationwide assignable expenditures through the 2024 Q3 quarterly expenditure report package, we estimate nationwide expenditure trend will be approximately 9% from 2023 to 2024. Therefore, we expect ACOs subject to the ACPT in 2024 will have approximately 2% lower benchmarks on average than they would have using the previous regional/national blended trend methodology. Note that Figure 1 uses the 2025 Medicare Advantage Rate Announcement trends as a comparison to the 2023 to 2024 ACPT trends because each year’s ACPT trend assumptions are expected to be released at around the same time as the rate announcement. The 2026 Medicare Advantage Advance Notice was released on January 10, 2025, and contained updated 2023 to 2024 FFS USPCC trend estimates of 4.7% for the ESRD population and 4.8% for the non-ESRD population. While these estimates are slightly higher than the 2023 to 2024 trend estimates from the 2025 Rate Announcement, they are still well below our expectation of final 2023 to 2024 FFS trends.2

Another key feature of the ACPT is that the trend assumptions are prospectively set for the ACO’s entire five-year agreement period. For example, even if the 2024 to 2025 nationwide expenditure trend exactly matches the ACPT trend assumption, ACOs in agreement periods beginning in 2024 may still experience a benchmark reduction in performance year (PY) 2025 relative to the previous two-way blended trend because the cumulative ACPT trend from 2023 to 2025 will be below the actual nationwide trend. Therefore, this miss on 2023-2024 trend of the ACPT will have a lasting impact on ACOs beginning in PY2024 if left unchanged. ACOs in this situation may want to consider the pros and cons of rebasing under a new agreement period to avoid having the negative impact of the 2023-2024 ACPT drag down performance for the remainder of the agreement period. Additionally, if national expenditure trends continue to exceed the ACPT for future performance years, then it creates a compounding gap between the trend calculation with ACPT and without ACPT.

How does the ACPT affect ACO performance?

The ACPT can enable ACOs to budget against a known value at the beginning of each performance year. However, the release of the first set of ACPT growth values shows that MSSP benchmarks may be inadequate before the performance year begins. As of the time of writing, our estimates of emerging 2023-2024 national FFS expenditure trends are 8% to 9% for ESRD and 9% to 10% for non-ESRD enrollment types, which is two to three times higher than the ACPT for the corresponding time period for agreement periods starting in 2024.3 Given the magnitude of the expected difference between the ACPT and actual Medicare FFS expenditure trends, we expect the impact to be substantial. The illustrative example in Figure 2 shows the ACPT impact to the benchmark trend for two ACOs: one that is in a region identical to the nationwide average, and another in a region with higher FFS expenditure trends from 2023 to 2024 than the nationwide average. For simplicity, we show these calculations for the AD category only and assume a 20% ACO market share.

Figure 2: ACPT impact to benchmark trends

Trend Component Trend Values – ACO A Trend Values – ACO B
ACPT 3.6% 3.6%
National 9% 9%
Regional 9% 14%

Under the previous regional/national blended trend calculation:

ACO A trend factor=(9% *80%)+(9%*(1-80%)]=9.0% and ACO B trend factor=(14% *80%)+(9%*(1-80%)]=13.0%

Under the new ACPT calculation:

ACO A trend factor=9.0%*2/3+3.6%*1/3=7.2% and ACO B trend factor=13.0%*2/3+3.6%*1/3=9.9%

As shown in Figure 2, ACO A (in the region identical to the nationwide average) receives a 1.8% benchmark trend reduction due to the ACPT relative to the benchmark trend calculated with the two-way regional/national blend. ACO B (in the higher-cost region) receives a 3.1% trend reduction due to the ACPT. All applicable ACOs will likely experience a trend reduction due to the ACPT in PY2024, but it will be a larger reduction for ACOs in regions with higher 2023 to 2024 expenditure trends.

The graph in Figure 3 illustrates the potential benchmark trend differences based on the ACO’s regional trend, using the scenario above as a starting point. In Figure 3, the minimum gap (4.5% regional trend) is 0.6% between the two-way and three-way blended trend, and the maximum gap (15.5% regional trend) is 4.5% between the two-way and three-way blended trend.

Figure 3: Impact of regional trend on ACPT adjustment

Figure 3: Impact of regional trend on ACPT adjustment

Note: The blue line represents the regional/national blended trend calculation and the green line represents the ACPT trend calculation. Each trend line maintains the assumptions of the calculations above (20% ACO market share, 9% national trend, and 3.6% ACPT).

What options are available for CMS to mitigate the impact of the ACPT in PY2024?

Balancing the incorporation of prospective trends that add predictability to the benchmarks for ACOs and retrospective trends that take into account emerging data is a difficult task. ACOs are concerned with the PY2024 settlement impacts if the ACPT is applied as described above. We are aware of two key items that can mitigate some of the ACPT impact (for the prevention or reduction of losses only) and that provide the Centers for Medicare and Medicaid Services (CMS) some flexibility in how they apply the ACPT:4

  1. The ACPT has a “guardrail” that compares the settlement result with and without the ACPT. If an ACO will generate losses under the three-way ACPT blended trend, then CMS will recalculate the settlement under the two-way regional/national blended trend methodology (i.e., the methodology in place before the ACPT was introduced). If the loss is smaller using the two-way blended trend, then the two-way blended trend will be used in the final settlement calculation up to a shared savings of $0. The guardrail can reduce a shared loss but cannot result in a shared savings even if the two-way trend would result in a shared savings. Therefore, this mitigates the impact of the ACPT on ACOs with expenditures higher than the benchmark, but does nothing for ACOs that earn lower shared savings due to the ACPT.
  2. CMS reserves the right to reduce the weight of the ACPT (currently set at one-third) ”if it is determined that expenditure growth has differed significantly from projections made at the start of the agreement period due to unforeseen circumstances.” It is difficult to predict whether CMS will utilize this clause for PY2024 (or any future PYs).

Other ACPT considerations

While many ACOs are concerned about the ACPT in 2024 because emerging trends appear to be higher than the ACPT trend assumptions, the ACPT in future years could be higher than the actual nationwide expenditure trend. While recent years have seen higher expenditure trends than the projected USPCC trends in the rate announcement, that has not always been the case. The table in Figure 4 shows a comparison of non-ESRD FFS USPCC trend assumptions in the Medicare Advantage Rate Announcement to the actual nationwide expenditure trends for the non-ESRD population.

Figure 4: Non-ESRD assumptions vs. actual nationwide expenditure trends

2017
to
2018
2018
to
2019
2019
to
2020
2020
to
2021
2021
to
2022
2022
to
2023
2023
to
2024
Rate Announcement Trend Assumption 2.9% 4.0% 5.2% 11.7% 9.4% 8.8% 3.5%
Actual FFS Expenditure Trend 3.8% 4.5% -6.8% 8.0% 4.7% 10.2% 9%*
Difference -0.9% -0.4% 12.0% 3.7% 4.7% -1.4% -5.5%

* 2023 to 2024 actual trend is estimated based on data through 2024 Q3.

Another feature of the ACPT that ACOs should consider is the method CMS will use to apply the ACPT trend. CMS starts with the same percentage trend assumptions for all ACOs, then converts the percentage trend to a risk-adjusted flat dollar growth increment for each ACO. That means that the final ACPT will vary from ACO to ACO, with efficient ACOs (i.e., lower risk-adjusted costs than the nationwide average) receiving higher trends than the starting ACPT, and inefficient ACOs receiving lower trends.

In summary, the incorporation of the ACPT is expected to result in materially different benchmark trends relative to the previous two-way blend methodology and can result in benchmark insufficiency through an entire agreement period if the ACPT factors are significantly lower than actual trends. Please reach out to your Milliman consultant if you would like to continue the conversation about how the ACPT will affect your ACO.


1 CMS (August 2024). MSSP: Shared Savings and Losses, Assignment and Quality Performance Standard Methodology: Specifications of the Accountable Care Prospective Trend (ACPT) and Three-Way Blended Benchmark Update Factor. Retrieved January 24, 2025, from https://www.cms.gov/files/document/medicare-ssp-acpt-specifications.pdf.

2 The 2026 Medicare Advantage Advance Notice does not take into account full CY2024 claims experience in its updated USPCC trend estimates.

3 These nationwide trend estimates are based on a review of nationwide assignable trends in 2024 Q3 QEXPU report packages, adjusted for seasonality and trend based on past years’ relationships between Q3 and final per beneficiary per year (PBPY) expenditure trend values. Actual results will vary, and final 2023 to 2024 nationwide expenditure trends will not be known until the 2024 MSSP settlements are released in fall 2025.

4 CMS (August 2024), Specifications of the Accountable Care Prospective Trend (ACPT) and Three-Way Blended Benchmark Update Factor, op cit.


About the Author(s)

We’re here to help