How CMS’s crackdown on unlinked chart reviews impacts risk adjustment

‍

CMS issued its 2027 Advance Notice on January 26, 2026, with a proposal that will significantly change how Medicare Advantage (MA) plans conduct retrospective risk reviews. Unlinked chart review records (CRRs) will no longer count toward risk adjustment scores. Now, every diagnosis submitted for risk adjustment must be tied to an encounter.

For years, unlinked claims have been flagged as a significant driver of MA overpayments. This latest move from CMS marks a significant crackdown on the practice and a shift in expectations for MA plans and how they conduct retrospective risk reviews.

MA plans will face a challenge: ensuring that the risk of their member population is accurately captured while having a strong data infrastructure to reliably connect all diagnoses to encounters. Doing so will require advanced technology, including artificial intelligence (AI), to support highly transparent, traceable documentation and point-of-care solutions to support HCC capture during the encounter.

In an era of heightened payment scrutiny, including new RADV audit requirements, it is more critical than ever that every MA plan take the steps needed to minimize the potential financial impact of CMS’s changes and prepare for a new level of regulatory accountability.

‍

Overview of proposal to exclude unlinked chart review records (CRRs)

Unlinked CRRs introduce new, risk-adjustable diagnoses into CMS’s data without tying them to a particular billable service or encounter. When conducting retrospective chart reviews, many MA plans may find documentation in a member’s medical record that supports a diagnosis for diabetes, for example, without the diagnosis having appeared on any previously submitted claim for that member.

Until recently, that plan could have submitted a CRR with the diabetes diagnosis without referencing the Internal Control Number (ICN) of any existing encounter data record or chart review and without specifying the actual visit. The resulting CRR is therefore unlinked because it introduces a new risk‑adjustable diagnosis into CMS’s data without tying it to a particular billable service or encounter. CMS’s systems see “extra” diagnoses but no underlying encounter to support them.

CMS’s 2027 Advance Notice explicitly excludes unlinked CRRs from being used in risk adjustment scoring. The practice, the agency estimates, is costing billions of dollars.

‍

The cost of unlinked CRRs: CMS, OIG, and MedPAC estimates

Several years ago, the HHS Office of Inspector General (OIG) released a report claiming that unlinked CRRs cost the government about $2.7 billion in overpayments in 2017. In response to that report, CMS claimed that the OIG had overestimated the impact and that the actual number was substantially lower.

Still, MA overpayments continued to be under the spotlight with varying reports and national attention focusing on reducing overpayments to MA plans. Last year, for example, MedPAC released a report estimating $84 billion in MA overpayments due to coding intensity as well as favorable selection. Around the same time, Congress introduced S. 1105, the “No UPCODE Act”, which would require various changes to MA risk adjustment, including excluding the submission of unlinked claims as well as diagnoses identified via health assessments.

CMS’s own estimates have also evolved significantly. In its 2027 Advance Notice, CMS projected that excluding unlinked CRRs, combined with other payment changes, would reduce MA payments by about 1.53%, or roughly $7.12 billion, in 2027. Plans relying heavily on unlinked submissions may experience a significant financial setback from these changes.

‍

What MA plans can expect from CMS’s claims linking requirements

New claims linking requirements will apply to 2026 dates of service, impacting next year's risk scores and payments. MA plans that have historically relied on a significant number of unlinked claims must move quickly to implement new data integration standards and point-of-care practices to ensure all diagnoses are captured and documented.

The stakes of inaction are high. Significant drops in RAF scores could introduce new revenue volatility. When taken together with CMS’s lower-than-expected projected rates (.09% versus the expected 4-6% increase), many MA plans could face a compounding financial impact. Acting early to build strong data integration and documentation practices will help MA plans protect their risk scores as well as position themselves better for RADV audits in a time of heightened scrutiny.

‍

What “good” looks like under CMS's proposed rules

Meeting CMS’s new standards requires a fundamental rethinking of how retrospective risk programs are structured. Here’s what best-in-class will look like starting today:

1. Every HCC has a claim behind it: Under CMS’s proposal, chart‑review findings as well as AI‑detected diagnoses must consistently tie back to specific encounters, dates of service, and claim identifiers. This requires plans to treat claims and encounter data as the foundation of their risk program, and chart reviews must be done to enrich the foundation, not be implemented separately from it.
2. Provider‑centric, point‑of‑care documentation is the first line of defense: HCCs must be captured accurately during the visit in the EHR so that most conditions originate from clean, encounter‑level documentation. Embedded prompts, pre‑visit summaries, and in‑workflow nudges can help clinicians close HCC and manage care gaps while they practice medicine, not after the fact.
3. Chart review programs are anchored to encounters: Instead of relying on unlinked retrospective reviews, plans will need workflows that use chart reviews and AI to validate and complete what happened in real encounters. This means ensuring supplemental evidence is systematically connected back to the right claim.
4. Transparency and traceability replace ad‑hoc spreadsheets: Today, many teams manually reconcile spreadsheets to figure out which chart‑based diagnoses belong to which claims. Going forward, matching logic must be standardized, documented, and easily visible so that every supported diagnosis has a defensible evidence chain.
5. AI is used to prioritize: AI‑enabled HCC detection should prioritize which charts to review and which conditions to validate. Then, teams can feed those results into a claims‑linked pipeline rather than generating unanchored codes.
6. Success is measured by defensibility: The new benchmark for retrospective risk programs is the ability to walk an auditor, step by step, from a risk score factor to a claim and supporting documentation.

‍

MA plans need a retrospective risk adjustment solution that prioritizes data integrity and audit defensibility. Reveleer Claims Linking is built for this environment, turning coded charts into submission-ready, encounter-tied data.

Reveleer Claims Linking:

• Automatically links coder-documented and AI-detected diagnoses codes to claim ID’s and dates of service.
• Uses standard parameters to classify each diagnosis as an Exact Match, Possible Match, or No Match to a claim.
• Supports an end-to-end pipeline from chart retrieval and AI-enabled HCC capture to submission-ready HCCs for retrospective risk and RADV workflows.
• Pre-builds the “show me the claim behind this HCC” story to prevent last-minute audit scrambles.

‍

Reveleer Claims Linking helps plans prepare for CMS’s upcoming requirements and bolster audit readiness going forward in an era of heightened scrutiny. Schedule a demo to explore how to reduce unsupported revenue risk, strengthen defensibility, and improve year-round audit readiness.

‍

‍

‍