Background
In late May 2025, the Center for Medicare and Medicaid Innovation (CMMI) announced updates to the Accountable Care Organizations (ACO) Realizing Equity, Access, and Community Health (REACH) Model for performance year (PY) 2026.1 According to CMMI, these changes are meant to improve the model’s long-term success by adjusting the financial approach, and are based on early findings from the PY2023 evaluation of the program.2
These changes seem to reflect CMMI’s concerns about increased spending over the model’s duration and alignment with its new strategic direction.3 In this paper we describe the PY2026 ACO REACH changes and quantify their estimated impact. The analysis section for each change estimates the impact of that change assuming no other changes were implemented, unless otherwise noted.
1. Modify risk score growth constraints
Description: Under the current ACO REACH methodology for Standard ACOs, CMMI applies a 3% symmetrical cap on the normalized risk score growth between the reference year (2022) and the performance year. The performance year risk scores are then further adjusted by a coding intensity factor (CIF), if applicable, up to a maximum of 1%.
CMMI identified that some ACOs experienced significant risk score growth between the third benchmark year (i.e., 2019) and the reference year that was not limited in any way by the current symmetrical risk cap. To mitigate the impact of this growth, CMMI will apply an additional 3% cap to the risk score growth between 2019 and 2026. The cap applies after the existing cap and CIF, and unlike the existing cap, it is not symmetrical (i.e., there is no floor to go with it).
For High Needs ACOs, CMMI will raise the cap on the High Needs CIF from 1% to 2%. This CIF is calculated based only on the risk score change for High Needs ACOs. CMMI is also introducing an 8% cap on risk score growth (from 2023 to 2026) for new beneficiaries who voluntarily aligned to a High Needs ACO, without a minimum beneficiary threshold.
Additionally, for all REACH ACOs, CMMI will remove an exemption from risk score caps currently in place for ACOs whose beneficiary count has more than tripled between the performance year and the reference year.
Impact: For Standard ACOs whose capped, CIF-adjusted risk score growth between 2019 and 2026 is less than 3%, there will be no effect. Standard ACOs that saw larger increases in normalized risk scores between 2019 and 2022 will now have less of that growth reflected in their benchmarks. Because risk scores affect benchmark calculations, which in turn affect shared savings/losses, this new cap will have an adverse effect for some ACOs: a smaller shared savings potential and a higher risk of shared losses.
Additionally, since the 2019–2026 risk score growth cap is one-sided, ACOs with risk score decreases will not get any benefit. This creates an asymmetry that could disadvantage some ACOs without benefiting others.
Analysis: Using Milliman's ACO Builder, we estimated the impact of this change on Standard REACH ACOs’ PY2026 projected gross savings. We projected PY2026 savings for all Standard ACOs both before incorporating the new risk score cap and after and then calculated the difference in projected savings.4 Figure 1 shows that we estimate roughly one in six Standard ACOs in the program would see a decrease in projected savings with the new risk score cap, with the rest unaffected. We estimate that approximately 4% of ACOs will experience a large (> 5%) decrease in savings as a result.
Figure 1: Estimated PY2026 gross savings percent impact of additional risk score cap
2. Reduce the regional component of the benchmark for all REACH ACOs
Description: In the PY2026 update, CMMI also indicated that it would reduce the weight of regional expenditures to 40% for PY2026 (for High Needs / New Entrant ACOs this becomes 45%), as seen in Figure 2.
Figure 2: PY2025 and PY2026 regional weights in baseline adjustment
| REACH ACO type | PY2025 | PY2026 |
|---|---|---|
| Standard ACOs | 45% | 40% |
| High Needs or New Entrant ACOs | 50% | 45% |
Impact: With the PY2026 blend of 60% historical and 40% regional, a Standard ACO’s own historical expenditures will have a greater influence on its PY2026 benchmark than was the case in PY2025.
ACOs whose historical expenditures were regionally efficient in the benchmark period—i.e., whose 2017–2019 expenditures per beneficiary per month (PBPM were lower than the risk-adjusted regional rate from the ACO REACH Rate Book) may be negatively impacted by this change because a lower weight on the regional rate will create a smaller regional rate baseline adjustment. Conversely, ACOs that were not regionally efficient in the benchmark period may benefit from this change because the regional rate will have a smaller weight and thus such ACOs will have a larger regional rate baseline adjustment. The ceiling and floor on regional adjustments can affect the extent to which ACO benchmarks are affected by this change.
Analysis: Figure 3 summarizes the estimated proportion of Standard ACOs and the estimated PY2026 gross savings impact of this update. More ACOs will be impacted by this update than by the additional risk score cap, but this change will generally have a smaller effect on projected savings for affected ACOs.
Figure 3: Estimated PY2026 gross savings percent impact of regional weight update
3. Narrow the first risk corridors to 10% for REACH ACOs in the global risk option
Description: REACH ACOs are subject to risk corridors, which limit their shared savings or losses. Under the Global track in PY 2025, ACOs retain 100% of savings or losses up to 25% of their benchmark (i.e., the first corridor). Generated savings or losses greater than 25% of the benchmark are then shared with CMMI, with CMMI’s share increasing with each subsequent corridor.
For ACOs participating in the Global track in PY2026, CMMI has announced plans to reduce the first corridor to 10% of the benchmark.
Impact: ACOs will be less exposed to large losses (and savings), sharing in full savings/losses only up to 10% of the benchmark, rather than 25%. This lowers both risk and reward. For example, an ACO that generated 20% of savings will now only get to keep 15% (10% + (20% - 10%) * 50%5) of it instead of the full 20%. Conversely, an ACO that generated 20% losses would only be responsible for 15% of those losses, with the rest being shared with the Centers for Medicare and Medicaid Services (CMS).
Analysis: The majority of the Standard ACOs in the Global track in PY20226 (Direct Contracting) and PY20237 (REACH) generated savings or losses within the 10% symmetrical corridors, while roughly 14% each year generated savings between 10% and 25% of benchmark. After incorporating the new risk score cap and regional rate blend percentage in ACO Builder, we estimate that in PY2026 approximately 7% of Standard ACOs in the Global track will generate enough savings to be impacted by this change (i.e., > 10% gross savings). As can be seen in Figure 4, this implies that a slightly higher proportion of ACOs are estimated to have savings reduced rather than losses reduced as a result of this update.
Figure 4: Estimated percentage of ACOs impacted by risk corridor update
4. Increase quality withhold from 2% to 5% while proportionally increasing the high performers pool bonus
Description: CMMI applies a quality withhold to ACO benchmarks that can be earned back during the performance year. Beginning in PY2023, CMMI reduced the withhold to 2% and made it pay-for-performance. Beginning in PY2026, CMMI announced that it will increase the withhold to 5%.
Currently, in addition to earning back the quality withhold, REACH ACOs that meet the continuous improvement / sustained excellence performance (CI/SEP) criteria and have an average quality score percentile rank of 70% or more across all claims-based measures are eligible for additional funds from the high performers pool (HPP). The HPP is funded by pooling the quality withholds of REACH ACOs that did not meet the CI/SEP criteria during the performance year.
Impact: ACOs are at greater risk for their performance on the ACO REACH quality metrics.
ACOs that achieve high quality scores, especially those meeting CI/SEP criteria and ranking above the 70th percentile across claims-based measures, will benefit from a larger HPP. Because this pool is funded by unearned withholds, a higher total withhold means a larger potential bonus for those ACOs that qualify to share in the HPP.
On the other hand, ACOs with lower quality performance could face larger financial losses because more of their benchmark will be at risk. For example, an ACO with a 75% quality score would see its benchmark reduced by an extra 0.75%: i.e., unmet quality score * additional quality withhold, or (1 - 75%) * (5% - 2%).
Analysis: Based on ACO REACH PY2023 results, 24% of all ACOs achieved a 100% total quality score and earned back the entire 2% quality withhold. Approximately 55% of all ACOs satisfied the CI/SEP requirement, and of those, 33% received the HPP bonus. This bonus was $9.38 PBPM paid across 24 ACOs.7
Figure 4 illustrates the proportion of ACOs in PY2023 under each quality score quartile and the estimated benchmark decrease if the quality withhold was 5% instead of 2%. This illustrates how higher quality scores can shield ACOs from more material impacts of this change.
Figure 5: Estimated PY2023 gross savings percent impact of quality withhold update
5. Update risk adjustment model
Description: In 2024, CMMI announced plans to continue transitioning from the V24 risk score model to V28. It utilized a phase-in approach with the V28 weight being 33% in 2024, 67% in 2025, and 100% in 2026.
In this latest announcement, CMMI essentially reiterates its original plans (i.e., no changes were included).
Note that in the official announcement, CMMI had made a typo stating the PY2025 blends as 67% V24 and 33% V28. The appropriate blends are 33% V24 and 67% V28.
Impact: No new impact expected beyond what had been previously communicated by CMMI.
6. Adjust PY2024 expenditures for significant, anomalous, and highly suspect billing
Description: In the past, CMMI has flagged significant, anomalous and highly suspect (SAHS) billing patterns. In 2024, CMMI had finalized a ruling8 to “hold ACOs harmless,” which means that clinicians, hospitals, other healthcare providers, and ACOs are not unfairly held responsible for this spending.
In PY2024, CMMI will remove two specific Healthcare Common Procedure Coding System (HCPCS) codes for urinary catheters and ostomy bags (A4353 and A5057) from settlement calculations that showed SAHS billings. These codes will not only be excluded from PY 2024 but also from historical benchmark years and from future performance years in which 2024 serves as a benchmark year (e.g., PY2026 for the calculation of regional baseline rate adjustment for voluntarily aligned beneficiaries), and these exclusions will impact the retrospective trend adjustments (RTA) for PY2024 and PY2025. The PY2026 rate book is expected to incorporate the removal of these billing codes.
This is a shift from their approach in PY 2023, where these codes were excluded from performance year expenditures but not from historical benchmarks. It is not known at this time whether CMMI will publish a revised PY2024 rate book.
Impact: The effect of potential SAHS billing on PY2026 results will not be known unless and until such time as an SAHS determination for PY2026 is made.
What’s next
While there are some situations where ACOs could see increased savings, generally these changes will increase pressure on REACH ACOs to decrease expenditures and increase quality of care. For some ACOs, REACH may no longer be a viable option due to these changes. The future of REACH beyond PY2026 is also uncertain. For these reasons, we recommend that REACH ACOs consider their options; for example, determining whether the Medicare Shared Savings Program (MSSP) could be a better fit for one or more of the ACO’s participant providers. ACOs should carefully consider all available options for program participation.
Please reach out to your Milliman consultant for impact estimates and potential mitigation strategies for your ACO.
Guidelines issued by the American Academy of Actuaries require actuaries to include their professional qualifications in all actuarial communications. The authors of this paper are member of the American Academy of Actuaries and meet the qualification standards for performing the analyses in this report.
1 Centers for Medicare and Medicaid Services. (May 29, 2025). ACO REACH Model performance year 2026 model update: Quick reference. Retrieved June 11, 2025, from https://www.cms.gov/aco-reach-model-performance-year-2026-model-update-quick-reference.
2 Center for Medicare and Medicaid Innovation. (May 21, 2025). Preview of findings from the evaluation of ACO REACH Model for performance year 2023. Retrieved June 11, 2025, from https://www.cms.gov/priorities/innovation/data-and-reports/2025/aco-reach-preview-py2023-evaluation.
3 Sutton, A. (May 13, 2025). CMS Innovation Center strategy to make America healthy again. Centers for Medicare and Medicaid Services. Retrieved June 11, 2025, from https://www.cms.gov/priorities/innovation/about/cms-innovation-center-strategy-make-america-healthy-again.
4 Key assumptions used in projection: CIF of 1.01, quality score of 75%, and projecting from calendar year 2024 experience.
5 Fifty percent of shared savings/losses incurred within the second risk corridor (i.e., 10%–35% of benchmark now) are shared with CMS.
6 Center for Medicare and Medicaid Innovation. (October 20, 2023). Direct Contracting PY2022 participant-level financial and quality results. Retrieved June 11, 2025, from https://www.cms.gov/global-and-professional-direct-contracting-gpdc-model-performance-year-2022-financial-and-quality.
7 Center for Medicare and Medicaid Innovation. (November 8, 2024). ACO REACH PY2023 participant-level financial and quality results. Retrieved June 11, 2025, from https://www.cms.gov/priorities/innovation/aco-reach-py2023-financial-and-quality-performance-results-fact-sheet.
8 Centers for Medicare and Medicaid Services. (September 27, 2024). Medicare program: Mitigating the impact of significant, anomalous, and highly suspect billing activity on Medicare Shared Savings Program financial calculations in calendar year 2023 (42 CFR Part 425). Retrieved June 11, 2025, from https://public-inspection.federalregister.gov/2024-22054.pdf. .