The Centers for Medicare & Medicaid Services (CMS) is announcing a coordinated set of changes to the ACO REACH Model starting in Performance Year 2026 (PY 2026) that are expected to improve the model test by adjusting the financial methodology to improve model sustainability based on the findings in the Preview of the PY 2023 Evaluation Report.
These changes are described in the tables below.
| PY2026 Model Update | Description |
|---|---|
| Modify risk score growth constraints | Standard ACOs: New Policy: Some ACOs had large risk score growth in the early years of the model (2019–2022). For PY 2026, after the application of the current Cap and CIF policy, CMS will apply an additional cap of 3% on risk score growth from 2019 to 2026 to constrain risk score growth from the early years of the model. High Needs ACOs: New Policy: In PY 2026, CMS will increase the ceiling on the CIF from 1% to 2% for High Needs ACOs. For newly voluntarily aligned beneficiaries in High Needs ACOs, CMS will apply an asymmetric risk score growth cap of 8% from 2023 to 2026. No minimum beneficiary threshold will apply to the newly voluntarily aligned risk score cap. All ACOs: New Policy: In PY 2026, ACOs with a beneficiary population in the Performance Year that is at least 3 times the reference year population will no longer be exempt from risk score caps. |
| Reduce the regional component of the benchmark for all REACH ACOs | Current Policy: Since PY 2021, Standard ACOs have received benchmarks based on a blend of historical and regional expenditures, with the proportion of regional expenditures increasing each year. In PY 2024 and PY 2025, Standard ACO benchmarks were based upon a 55%/45% blend of historical and regional expenditures, respectively, with the impact of regional expenditures constrained by a ceiling and floor. Prior to PY 2025, New Entrant and High Needs ACOs received benchmarks that were calculated exclusively from regional expenditures. In PY 2025, New Entrant and High Needs ACOs transitioned to a benchmark composed of a 50%/50% blended historical and regional expenditures. New Policy: In PY 2026, CMS will decrease the regional benchmark weighting for all REACH ACOs.
|
| Narrow the 1st Risk Corridors to 10% for REACH ACOs in the Global risk option | Risk corridors serve as a key risk mitigation mechanism in the ACO REACH Model. The aggregate amount of savings or losses that ACOs in Global or Professional tracks are eligible to receive as Shared Savings or are required to repay as Shared Losses are constrained by a series of risk corridors. Current Policy: The 1st risk corridor in the Global risk option (100%) is set so CMS does not share in savings until a REACH ACO has generated savings greater than 25% of their benchmark. New Policy: In PY 2026, CMS will narrow the 1st risk corridor to be 10% (instead of 25%) for REACH ACOs in the Global risk option (100% risk), so that savings and losses above 10% are shared with CMS. |
| Increase Quality Withold from 2% to 5% while proportionally increasing the High Performers Pool (HPP) Bonus | For all REACH ACOs, a portion of the benchmark is held “at risk,” dependent on the ACO’s quality performance. Current Policy: This Quality Withhold is set at 2% of the value of the trended, regionally blended, risk adjusted benchmark. REACH ACOs can “earn back” some or all the Quality Withhold, depending on their performance on a pre-determined set of quality measures. The model also tests the use of a HPP Bonus to further incentivize high performance and continuous improvement. ACOs qualify for a bonus from the HPP if they meet the continuous improvement/sustained exceptional performance (CI/SEP) criteria and demonstrate a high level of performance or meet improvement criteria on the pre-set quality measures. The HPP is “funded” from quality withholds that were not earned back by the ACOs who did not meet the CI/SEP criteria. New Policy: In PY 2026, CMS will increase the Quality Withhold from 2% to 5%, which will accordingly increase the HPP Bonus. |
| Update Risk Adjustment Model | The revised Part C prospective risk adjustment model that is applied under the Medicare Advantage Program is applied to Standard and New Entrant ACOs. Current Policy: In PY 2025, risk scores were blended using 33% of the risk scores under the 2020 risk adjustment model (V24) and 67% of the risk scores under the revised 2024 risk adjustment model (V28). New Policy: In PY 2026, CMS will continue with implementation of the V28 model and increase the weight on the new V28 prospective HCC model to 100%. The concurrent risk adjustment model used for High Needs ACOs is not being modified. |
| Adjust PY 2024 Expenditures for Significant, Anomalous, and Highly Suspect (SAHS) Billing | Current Policy: SAHS billing is defined as a given HCPCS or CPT code that exhibits a level of billing that represents a significant claims increase either in volume or dollars (e.g., dollar volume significantly above prior year or claims volume beyond expectations) with national or regional impact (e.g., not only impacting one or few ACOs) and represents a deviation from historical utilization trends that is unexpected and is not clearly attributable to reasonably explained changes in policy or the supply or demand for covered items or services. The billing level is significant and represents billing activity that would cause significantly inaccurate and inequitable payments and repayment obligations in the ACO REACH model if not addressed. New Policy: ACO REACH will remove two HCPCS codes (A4353 and A5057) associated with intermittent urinary catheter supplies and ostomy bags that were identified for SAHS billing activity from 2024 expenditures and re-calculate the retrospective trend adjustment and stop loss. These two HCPCS codes will be excluded from Performance Year expenditures and Retrospective Trend Adjustments for PY 2024 and from historical benchmark years for PY 2024, unlike with PY 2023 SAHS policy, where historical benchmark adjustments were not made for SAHS, and will be excluded from future performance years where 2024 serves as a benchmark year. |