Page 62

62

consider future updates to the CMS-HCC risk adjustment model for future years and appreciate commenter input. Comment: Commenters suggested several proposals for model revisions that they believe would improve the model, such as including additional sources of clinical and non-clinical data or removing data or conditions that they believe lead to more variation in coding between MA and FFS. Response: We appreciate the extensive and thoughtful comments and feedback we received on improving the CMS-HCC risk adjustment model. Comment: Many commenters recommended CMS either stop, delay, or lengthen the phase-in period of the 2024 CMS-HCC model to allow stakeholders more time to assess the impact of full implementation, with multiple commenters noting that pausing would allow the new administration additional time to evaluate the model. Response: The three-year phase-in of the 2024 CMS-HCC risk adjustment model is consistent with how CMS has approached other instances in which model updates have been phased in over time (e.g., the 2014 model was phased in over three years and the 21st Century Cures Act model requirements were phased in over four years, with the final model adopted in the CY 2020 Rate Announcement and phased in over three years). We also note that MA organizations have had over a year of risk scores calculated under the 2024 CMS-HCC model, including CY 2024 and the initial risk scores in CY 2025. MA organizations that conduct analyses on the risk scores are already aware of the 2025 mid-year risk scores. Given this experience, we do not think it is necessary to further delay the full implementation of the updated model for the purpose of stakeholders gaining more experience. CMS appreciates the concerns raised by the commenters on the timeline for finalizing the implementation of the 2024 risk adjustment model. However, it is important to maintain or improve the accuracy of the risk adjustment model by updating it to reflect more recent relative costs, treatment and utilization patterns, and coding practices. We have previously noted that as a model ages and is used to predict expenditures for more recent enrollees in MA plans, that predictive accuracy begins to decline. Delaying the phase-in of a risk adjustment model that is based on more recent underlying data will prolong the use of an older risk adjustment model that, though still accurate according to CMS’ measures (i.e., having a predictive ratio between 0.90 and 1.10), is waning in its ability to predict current costs. We also note that the 2024 CMS-HCC model changes are a combination of routine data updates (e.g., updating the years of data used when calibrating the model) and clinical updates to the HCCs that were required to develop a model using the ICD-10 diagnosis codes implemented in 2015 needed to keep MA payments up-to-date and to improve payment accuracy to MA plans. For CY 2026, 100 percent of the risk scores for MA plans and certain demonstrations will be calculated using the 2024 CMS-HCC model.