CMS finalized the CY2027 Medicare Advantage and Part D Rate Announcement on April 6, 2026, and the result is meaningfully different from what the industry saw in January. The average expected payment change rose from 0.09% in the Advance Notice to 2.48%, more than $13 billion. Including CMS's estimate of risk score trend, the expected increase is 4.98%. The difference came almost entirely from one decision: CMS did not finalize its proposed recalibration of the Part C risk model.

The other major proposal survived. Beginning in 2027, diagnoses from unlinked chart review records will be excluded from risk score calculations, with one exception. For plans and risk-bearing providers, that combination calls for very different responses.

What Changed Between January and April

CMS's own comparison of the Advance Notice and the final announcement shows where the movement occurred:

  • Effective growth rate: 4.97% proposed, 5.33% final.
  • Rebasing and re-pricing: not available in January, -0.17% final.
  • Risk model revision and normalization: -3.32% proposed, -1.12% final.
  • Sources of diagnoses (unlinked chart reviews and audio-only): -1.53% in both.
  • Star Ratings: -0.03% in both.
  • MA coding pattern adjustment: 0% in both.

CMS expects underlying coding trend to raise average risk scores by about 2.50%, which is not included in the 2.48% figure.

The Risk Model: V28 Retained Without Recalibration

In January, CMS proposed keeping the 2024 CMS-HCC model structure (V28) but recalibrating it to 2023 diagnoses and 2024 expenditures. In the final announcement, CMS kept the existing 2024 model calibrated on 2018 diagnoses and 2019 expenditures. That removed most of the proposed negative adjustment and avoids a shift in relative condition weights that would have required plans to re-forecast revenue by condition.

For finance teams, the practical consequence is continuity. The coefficients that drove 2026 revenue still apply. Organizations that built models around the proposed recalibration should revert to current coefficients for bids and provider contract projections. CMS also finalized the exclusion of diagnoses from audio-only encounters from risk scoring, which CMS estimates has no effect on average.

Unlinked Chart Reviews: Finalized, With an Exception

CMS finalized the exclusion of diagnoses from unlinked chart review records from risk score calculation for both Part C and Part D. The final policy includes an exception for beneficiaries who switch from one MA organization to another, recognizing that a plan may need to document conditions for new members whose prior encounters were submitted by a different organization. CMS estimates that without this exception the impact would have been -1.78%; with it, the impact is -1.53%. Coverage of the final policy indicates the exception does not extend to beneficiaries moving from Original Medicare into MA.

Because the effect is concentrated among organizations that relied heavily on unlinked records, the average understates the impact for many plans. A plan that drew a large share of its RAF from retrospective unlinked submissions could see a reduction well beyond the average, even with the overall rate increase.

Part D Model Changes

Part D moves in the opposite direction from Part C. CMS finalized an updated Part D risk adjustment model calibrated on 2023 diagnoses and 2024 costs, reflecting Inflation Reduction Act benefit changes taking effect in 2027, including the phased-in manufacturer discount for specified small manufacturers. The model applies the same audio-only and unlinked chart review exclusions, and it is calibrated separately for MA prescription drug plans and standalone PDPs, with separate normalization factors for each. Organizations with both lines of business should re-forecast Part D revenue separately and should not assume their MA-PD and PDP risk scores will move together.

What Health Plans Should Do Now

  1. Finalize bids on the right assumptions. Bids for 2027 are due the first Monday in June. Confirm that revenue models use the retained Part C model, the new Part D model and the unlinked exclusion with the switcher exception.
  2. Quantify residual unlinked exposure. Identify, by contract, the HCCs that would have come only from unlinked chart reviews for non-switching members, and treat that revenue as gone for 2027.
  3. Build a switcher workflow. Flag members who joined from another MA organization so documentation for them is handled under the exception, and confirm that compliance can verify eligibility for it.
  4. Accelerate prospective and concurrent capture. Encounters in 2026 drive 2027 payment. The remaining months of this year are the window to document chronic conditions during visits rather than after.
  5. Repurpose retrospective review. Linked corrections, deletions and validation still matter. Retrospective work should increasingly focus on accuracy and encounter linkage.

What Risk-Bearing Providers Should Do Now

  • Review contract mechanics. Determine how plan revenue changes flow through to percent-of-premium and shared-risk arrangements, and whether contracts reference plan-run retrospective programs.
  • Own documentation at the visit. Provider groups that document chronic conditions completely during encounters are now more valuable partners, and less exposed to a plan's retrospective strategy.
  • Tighten encounter submission. Linked chart reviews depend on complete, timely encounter data. Rejected or missing encounters now carry a direct revenue cost.
  • Respond to queries promptly. Signed addenda tied to the original encounter will be a larger part of compliant capture.

A Quick Readiness Check

  • Revenue forecasts updated for the retained V28 coefficients and new Part D model.
  • Unlinked chart review exposure quantified, with the switcher exception modeled.
  • Pre-visit and concurrent review active for high-risk members.
  • Encounter rejection rates and linkage rates monitored monthly.
  • Provider query and addendum workflow documented and auditable.

How a Platform Approach Helps

The final notice rewards organizations that can trace every HCC to an encounter. Curitics Health supports this with patient-level CMS-HCC V28 RAF scoring, AI chart classification that separates documented from inferred HCCs for coder review, prospective, concurrent and retrospective coding review with provider queries and signed addenda, and automated 837P claims delivery that helps keep encounter data complete.

The Bottom Line

The 2027 rate is better than January suggested, but it is not a return to business as usual. CMS chose model stability while closing a diagnosis source it views as weakly supported. The revenue that remains will flow to organizations that document conditions at the point of care and can prove it.