On January 26, 2026, CMS released the CY2027 Medicare Advantage and Part D Advance Notice, and the headline number caught much of the industry off guard. After a 5.06% increase for 2026, CMS projects an average MA payment change of just 0.09% for 2027, roughly $700 million. The effective growth rate is healthy at 4.97%, but two risk adjustment proposals absorb nearly all of it: a recalibrated risk model with updated normalization, and the exclusion of diagnoses from unlinked chart review records.
The Advance Notice is a proposal, not a final policy. Comments are due February 25, 2026, and CMS expects to publish the final Rate Announcement no later than April 6. But the direction is unmistakable, and plans and risk-bearing providers should be using the next two months to understand their exposure.
The Numbers in Context
CMS's own breakdown of the proposed change from 2026 to 2027 shows where the money goes:
- Effective growth rate: +4.97%
- Risk model revision and normalization: -3.32% (normalization alone would be -1.50% if the model were not updated)
- Exclusion of unlinked chart review diagnoses: -1.53%
- Star Ratings: -0.03%
- MA coding pattern adjustment: 0%
- Rebasing and re-pricing: to be determined in the Rate Announcement
CMS separately estimates that underlying coding trend will raise average risk scores by about 2.45%, which would bring the expected average revenue change to about 2.54%. That figure is important: it assumes plans continue to document and code at the same pace. Organizations that lose access to a revenue source, such as unlinked chart reviews, will not see that trend.
The Unlinked Chart Review Proposal
Chart review records allow MA organizations to add or delete diagnoses based on a review of medical records after the encounter has been submitted. A linked chart review is tied to a specific encounter data record. An unlinked chart review is not; the diagnosis it reports does not connect to any particular service submitted to CMS.
CMS proposes that, beginning in 2027, diagnoses from unlinked chart review records would no longer be used to calculate risk scores. Plans could still submit them, but they would not count. CMS's reasoning is that a service should be reported on an encounter data record, and a chart review should not be the only record of a healthcare item or service. The agency noted that some plans appear to be submitting unlinked chart reviews in place of encounter records even when an encounter record could be submitted.
The concern is not new. In 2019, the HHS Office of Inspector General reported that MA organizations used chart reviews almost exclusively to add diagnoses rather than remove them, and that diagnoses reported only on chart reviews drove billions of dollars in risk-adjusted payments. OIG specifically recommended that CMS reassess whether unlinked chart reviews should be used for risk adjustment. According to an analysis from Georgetown University's Center on Health Insurance Reforms, nearly 58% of MA contracts submitted unlinked chart reviews in 2022.
CMS also noted that the payment effect will be larger for MA organizations that rely more heavily on unlinked chart reviews. A 1.53% average reduction will land very unevenly. For plans with aggressive retrospective programs, the effect could be several times the average.
The Proposed Risk Model Recalibration
The second major proposal keeps the structure of the 2024 CMS-HCC model (V28), including its eight segments, demographic variables, HCC definitions and condition count variables, but recalibrates it using 2023 diagnoses and 2024 expenditures instead of 2018 diagnoses and 2019 expenditures. The normalization denominator year would move to 2024.
Recalibration shifts relative coefficients. Some conditions will be worth more, others less, and the net effect for any plan depends on its population mix. Combined with normalization, CMS estimates the change at -3.32% on average. Plans with populations concentrated in conditions whose relative cost has fallen since 2019 will feel more of it.
CMS also proposes to exclude diagnoses from audio-only services from risk score calculations, aligning the model with existing MA diagnosis submission policy.
Part D Changes
For Part D, CMS proposes recalibrating the risk model on 2023 diagnoses and 2024 drug costs, applying the same audio-only and unlinked chart review exclusions, and, for the first time, calibrating separate model segments for MA prescription drug plans and standalone prescription drug plans. Separate normalization factors would apply to each. Organizations with both MA-PD and PDP books should model each separately.
What It Signals
Taken together, the proposals point to a clear CMS position: risk scores should reflect diagnoses that are documented in connection with real clinical encounters. The policy direction favors documentation at the point of care over retrospective harvesting. Even if specific numbers change in April, organizations should expect the scrutiny of diagnosis sources to continue.
What to Do Before April
- Quantify unlinked chart review dependence. Calculate the share of current-year RAF attributable to diagnoses that appear only on unlinked chart review records, by contract and by provider group.
- Re-run risk scores under the proposed calibration. Use the published coefficients to estimate the population-level effect, and identify which conditions drive the change.
- Review retrospective vendor and program contracts. Contingency fee arrangements tied to unlinked chart review findings may lose their value entirely.
- Assess prospective and concurrent capture. Determine how many retrospectively found diagnoses could have been documented during an encounter, and why they were not.
- Coordinate with provider partners. Risk-sharing contracts that pass through RAF changes will affect provider group revenue too.
- Submit comments by February 25 if the proposal creates specific operational problems, such as linking difficulties for members who switch plans.
Building for a Linked-Diagnosis World
The organizations best positioned for 2027 will be the ones that can see, at the patient level, where each HCC came from and whether it is supported by an encounter. A platform approach that combines patient-level CMS-HCC V28 RAF scoring with AI chart classification separating documented from inferred conditions, and that supports prospective, concurrent and retrospective review with provider queries and signed addenda, gives risk adjustment teams that visibility. Curitics Health brings those capabilities together so teams can test their exposure now rather than after the final notice.
The Advance Notice may change before April 6. The underlying message is unlikely to.
