Page 88

88

enrolled in Part D. Further, because Part D payment parameters are set across the enrolled population, in contrast with how Part C payment parameters are set across the FFS population, we set the 1.0 risk score across the entire enrolled population, including both the denominator of the model and the normalization factor(s). The Part D risk adjustment model and Part D normalization factors are developed and applied consistent with section 1860D-15(c)(1)(A) of the Act. The Part D model and normalization factor(s) together account for variation in costs between PDPs and MA-PD plans based on the differences in actuarial risk of different enrollees being served without changing the aggregate amounts payable. Calculating separate normalization factors for PDPs and MA-PD plans does not imply a different methodology as long as those factors are designed to maintain an average risk score of 1.0 across the Part D program, which has been a long-standing goal of normalization. Section 1860D-15(c)(1)(A) of the Act clearly permits CMS to modify the risk adjustment methodology to account for differences in the actuarial risk of enrollees being served by the PDP and MA-PD plan market – a requirement that does not exist for Part C. In this vein, applying a single normalization factor, as we do for the Part C risk scores, would lead to risk adjustors that do not appropriately account for the variation in costs between MA-PD and PDP enrollees. The risk score and normalization factor combined account for the expected risk of a Part D enrollee relative to the expected average across Part D in the payment year. A single normalization factor results in a risk adjustor that overestimates actuarial risk for MA-PD plans and underestimates actuarial risk for PDPs. As stated previously in this section of the CY 2026 Rate Announcement, and in the CY 2026 Advance Notice, CMS’ goal in applying separate normalization factors is to improve the predictive accuracy of PDP and MA-PD plan risk scores. The intent of normalization is to keep risk scores at consistent levels year over year so that the level of payment is driven by the bid, not the risk scores, and normalization adjusts for any and all drivers of risk score trends. Applying separate normalization factors calculated with the multiple linear regression methodology will reflect the change in population and health status from the denominator year to the payment year similar to how a single normalization factor would but will better account for differences in costs between PDPs and MA-PD plans. Accurately predicting the relative risk for PDPs and MA-PD plans is a long-standing goal of the Part D risk adjustment methodology. Similar changes to the Part D risk adjustment methodology were made in 2016 when CMS incorporated both FFS and MA-PD data into the Part D model to allow MA-PD coding and utilization patterns to be accurately reflected in the Part D relative costs and improve the predictive accuracy of the RxHCC model.19 While the differences between the PDP and MA-PD plan sectors are longstanding, the degree to which PDP and MA-PD risk scores have diverged is more recent. CMS analysis finds that PDP and MA-PD risk scores have been growing farther apart since 2015 but that the difference has been increasing more quickly in recent years. The percent difference between PDP and MA-PD

19 https://www.cms.gov/medicare/health-plans/medicareadvtgspecratestats/downloads/advance2016.pdf.