On August 8, 2018, the Centers for Medicare and Medicaid Services (CMS) released a proposed rule that will significantly change the Medicare Shared Savings Program (MSSP) if enacted. This paper is the second in a series of white papers Milliman is writing on the proposed rule.
The proposal, titled “Pathways to Success,” includes changes to the financial benchmark methodology that measures an accountable care organization’s (ACO’s) gross savings or losses under the MSSP. There are four key elements where changes have been proposed: agreement period length, regional fee for service (FFS) adjustment, risk adjustment, and trend. In this brief, we discuss the proposed changes and important implications for Medicare ACOs.
Agreement period length
CMS has proposed to lengthen the agreement period from the current three performance years to five performance years. Consistent with the current approach, the benchmark will be rebased (i.e., recalculated using updated experience data) for each agreement period and be based upon the ACO’s experience in the historical benchmark period, which is the three years prior to the agreement period. CMS has not proposed any changes to the weighting of benchmark years (BYs) used to set each ACO’s historical benchmark, which will continue to vary between the first agreement period (10%/30%/60% for BY1/BY2/BY3, respectively) and later agreement periods (equal weighting).
The implications for ACOs include:
- The assigned beneficiary population in later performance years may look very different from the assigned population in the benchmark years due to the seven-year gap between BY1 and the last performance year. This large gap will increase the likelihood that newly added physicians will impact performance year expenditures but not benchmark year expenditures, a phenomenon observed under the Next Generation ACO program.
- The longer agreement period will magnify the effects of the proposed 3% cap on risk adjustment and trend changes discussed below.
- ACOs with a stable participant list can expect to have a relatively stable benchmark due to both the longer agreement period and the risk adjustment cap discussed below.
Ultimately, ACOs have five years to work within their benchmark before the benchmark is rebased. ACOs with a favorable benchmark may be in a good position while ACOs with an unfavorable benchmark will have to look to care management improvements, participant list changes, improved coding, and other changes as rebasing will occur further down the road.
Regional FFS adjustment
The regional FFS adjustment will continue to be based on each ACO’s beneficiary distribution by county and enrollment type (Aged Non-Dual, Aged Dual, Disabled, and end-stage renal disease [ESRD]). However, CMS has proposed to limit the regional FFS adjustment to plus or minus 5% of national assignable per capita expenditures by enrollment type in BY3. CMS has also proposed changes to the weights given to the regional benchmark, including changing the first agreement period methodology to give weight to the regional benchmark, as shown in Figure 1 below.
Figure 1: Regional benchmark weight by ACO agreement period and spending level
||ACO spending relative
Note: Under both the current and proposed methodologies, the weights apply in progression to when an ACO is first subject to the regional FFS adjustment. For example, an ACO currently participating in the MSSP and not subject to a regional FFS adjustment will be subject to the Agreement Period 1 weights in its next agreement period.
The implications of these changes for MSSP ACOs include the following:
- ACOs with benchmarks significantly below their regional benchmarks will have the “windfalls” they might receive through the current regional FFS adjustment limited by the regional adjustment cap, while ACOs with benchmarks significantly above their regional benchmarks will find that the regional FFS adjustment may not be as prohibitive of a barrier to participating in the MSSP.
- Because all agreement periods now include the regional benchmark adjustment, ACOs can have a consistent perspective across all agreement periods about which tax identification numbers (TINs) to include in their participant list. Previously, ACOs may have benefited from including less efficient TINs in their ACO participant list for the first agreement period, but not the second or later agreement periods.
- ACOs considering joining the MSSP will need to understand how their costs compare to their regional costs before they start their first agreement period rather than prior to their second agreement period.
The effect of these changes ultimately depends on how an ACO’s historical benchmark compares to the regional benchmark. ACOs that have a high market share in their region will continue to see little impact from the regional FFS adjustment because the ACO’s experience is not excluded from the regional benchmark.
CMS did not propose changes to adjust the financial benchmark for ACOs based on the efficiency of their region as compared to national Medicare FFS efficiency levels. This is in contrast to the Center for Medicare and Medicaid Innovation's (CMMI's) Next Generation ACO model, where the financial benchmark is adjusted upwards for ACOs in "efficient" regions (regions which are lower than national benchmarks) and adjusted downwards for "inefficient" regions (regions which are higher than national benchmarks). As such, ACOs in lower-cost regions may have greater difficulties identifying financial opportunities in the MSSP as compared to ACOs in higher cost regions.
Currently, risk scores are used to adjust the MSSP benchmark in three ways:
1. Risk adjust each historical benchmark year to BY3.
2. Risk adjust the regional benchmark to the ACO’s average risk by enrollment type to calculate the regional FFS adjustment.
3. Risk adjust the ACO’s historical benchmark to each performance year
The first two uses described above continue to apply full risk score adjustment. Currently under use #3, the benchmark is fully adjusted for the new ACO beneficiaries’ risk score, but the risk score adjustment is effectively capped by a demographic adjustment for beneficiaries continuing in the ACO. CMS is proposing to use full risk score adjustment for both newly and continuously assigned beneficiaries, but capping the overall risk score adjustment at plus or minus 3%. It is important to note that the cap applies to the cumulative risk score adjustment between BY3 and a given performance year.
The implications for ACOs include:
- Complete and accurate coding is necessary to maintain an ACO’s benchmark. This has always been the case, but the proposed rules provide additional incentives to ensure proper documentation.
- Coding improvement can lead to a higher benchmark, but only up to the 3% limit. While the risk score adjustment is limited to 3%, maintaining or improving coding and documentation may be the difference between shared savings and shared losses for many ACOs.
- The model may not fully account for significant population changes due to the 3% risk score adjustment limit. This could benefit or hurt an ACO, depending on if the population changes are expected to increase or decrease costs.
With agreement periods now lasting five years, there is the potential for up to seven years of risk score adjustment from BY1 to the last performance year (PY5). However, only the cumulative adjustment from BY3 to each performance year is capped at plus or minus 3%. Therefore, significant population changes may occur over the course of the five years of the agreement period and not be fully captured in the risk score adjustment due to the 3% cap.
MSSP Financial Benchmark Basics
The MSSP financial benchmark is the cost target used to measure each ACO’s financial performance. The sharing of savings or losses is based on how the ACO’s costs (under the Medicare FFS payment schedules) compare to the financial benchmark.
The financial benchmark is based on the ACO’s costs for the three benchmark years prior to the start of each agreement period. Benchmark years 1, 2, and 3 are referred to as BY1, BY2, and BY3, respectively –with BY3 being the most recent year. The benchmark is based upon adjusting each benchmark year to BY3 and blending each benchmark year into a composite per capita target. The benchmark can also be adjusted based on the BY3 expenditure levels in the ACOs region–this is called the regional FFS adjustment.
Each agreement period is made up of performance years. The first performance year is called performance year 1 or PY1. Savings or losses are shared after each performance year between the ACO and CMS. The sharing of savings or losses depends on the MSSP track that the ACO participates in. Figure 2 below provides a simplified view of how the benchmark and performance year 1 settlement is calculated.
This illustration is a simplification for many reasons, including the fact that the benchmark is constructed separately for each enrollment type (Aged Non-Dual, Aged Dual, Disabled, and ESRD), and there are adjustments for population changes in addition to trend.
Like risk scores, trends are used to adjust all historical benchmark years to BY3 and then BY3 to each performance year. While trends continue to be based on the assignable population and retrospectively determined at the end of each performance year, CMS has proposed to replace the national trend adjustments in Agreement Period 1 and the regional trend adjustments in Agreement Periods 2 and later with a blend of regional and national trends for all agreement periods. When blending the regional and national trends, the national trend will be weighted by the ACO’s average market share in its region, and the regional trend will receive the remaining weight. For example, the trend for an ACO with 70% market share will be weighted 70% national trend and 30% regional trend.1
These changes will affect both how the ACO benchmark is constructed, i.e., how costs are trended to BY3 as well as how the benchmark is adjusted to each performance year. The effect on individual ACOs will depend on their market share and the relationship between the trends in their region and the national trends. In general, ACOs with high market share (e.g., greater than 50%) can expect to have a benchmark trend primarily based on national trends while ACOs with low market share can expect to have a benchmark trend primarily based on regional trends.
Key implications include:
- Although the risk for high market share ACOs negatively impacting their own benchmark through strong performance is mitigated by the proposal, reverting to a national trend adjustment reintroduces some of the original challenges of the MSSP’s national trend adjustment (e.g., misalignment in national and regional fee schedule trends).
- The MSSP trend target continues to be retrospective in nature. Therefore, ACOs will not fully understand their savings or loss position until well after the end of the performance year.
- ACOs in regions with high participation in Medicare risk-sharing programs (e.g., MSSP and Next Generation ACO) may see lower regional trends and therefore a lower financial benchmark.
Under the MSSP proposed rule, an ACO’s financial benchmark will continue to be largely based upon the ACO’s prior experience. However, the proposed changes in the MSSP’s financial benchmark methodology will have significant implications for most ACOs. Given the increase in the agreement period length from three to five years, it is critical that ACOs assess how the proposed rule will affect their financial benchmark and related strategies.
Figure 2: Simplified illustration of Benchmark and Performance Year 1 settlement
Note: The MSR is the “Minimum Saving Rate” an ACO must achieve before it shares in first dollar savings. The MSSP also has a Minimum Loss Rate or MLR. No settlement occurs below the minimum loss rate, but ACOs share in first dollar losses once the loss rate exceeds the MLR.
1Note that the national and regional trend blending weights are calculated separately for each of the four beneficiary enrollment categories. Market share is calculated for each of the ACO’s service area counties and then is weighted by the proportion of the ACO’s enrollment in each county.