Medicare Risk Coding Tightens in 2027: Unlinked Chart Reviews Will Not Count

Medicare Advantage risk adjustment is moving toward a clearer encounter test. The 2024 CMS-HCC model is fully in use for 2026 and will continue in 2027, but CMS has finalized a consequential change for 2027: diagnoses from unlinked chart-review records will generally be excluded from risk-score calculations. An exception applies when a beneficiary changes Medicare Advantage organizations.

For clinicians, coders and plan teams, the implication is practical: finding a diagnosis somewhere in a chart will no longer be enough when that diagnosis is not connected to a specific beneficiary encounter. The change strengthens the case for documentation that records what the clinician evaluated, the condition’s current status and the resulting plan—not merely a code selected after retrospective review.

What changes in 2027—and what does not

An unlinked chart-review record contains diagnosis information that is not associated with a specific beneficiary encounter. Under CMS’s finalized 2027 payment policy, those records will no longer contribute diagnoses to Medicare Advantage risk scores, apart from the plan-switching exception. CMS is also continuing the 2024 CMS-HCC model rather than introducing a replacement model for 2027.

The policy does not prohibit chart review, prevent correction of genuine documentation errors or eliminate diagnoses obtained from records linked to qualifying encounters. It changes whether a diagnosis disconnected from a service can affect payment. Organizations therefore need to distinguish a useful retrospective review from a submission that lacks the required encounter relationship.

For 2026 payments, non-PACE Medicare Advantage organizations are already using the 2024 CMS-HCC model for 100% of the risk-score calculation after a three-year phase-in. That model status and the coming record-linkage rule are separate issues: one determines how eligible diagnoses are grouped and weighted, while the other limits which diagnosis information can enter the calculation.

Accurate coding supports care, but it does not guarantee an outcome

Risk adjustment primarily determines payments to Medicare Advantage organizations based on the expected cost of caring for enrolled populations with different health needs. It should not be treated as a clinical outcome measure. A more complete risk score does not, by itself, show that a patient received medication management, follow-up testing, specialist care or help overcoming an access barrier.

Documentation can nevertheless support better care when it gives the clinical team usable information. A record that identifies the condition being addressed, its present status, relevant findings and the management decision can help the next clinician understand what remains unresolved. The same record may support an appropriate ICD-10-CM code, but its clinical value comes from the information and subsequent action rather than from the code alone.

This distinction prevents two opposite failures. Under-documentation can hide relevant complexity from both care teams and the payment system. Over-documentation—such as carrying forward an obsolete diagnosis without reassessment—can produce a misleading clinical picture and an unsupported payment submission.

Build the diagnosis from the encounter

A defensible workflow begins with the patient encounter, not with a desired HCC. The clinician establishes and documents the diagnosis; the coder translates that documentation under the applicable ICD-10-CM conventions and model mappings. If the record does not support the required specificity, the appropriate response is clarification through a compliant query process, not inference from medications, laboratory results or an old problem-list entry.

The official FY 2026 ICD-10-CM guidelines describe complete and accurate reporting as a joint effort between the healthcare provider and coder. They also direct reviewers to consider the entire record to identify the reason for the encounter and the conditions treated. That makes documentation quality a shared operational responsibility while leaving diagnosis establishment with the qualified provider.

For each condition relevant to the visit, a useful record should make several points discoverable without requiring a reviewer to reconstruct the clinician’s reasoning:

• the condition or symptom actually evaluated during the encounter;
• its current status, severity or relevant complication when clinically known;
• the evidence and assessment supporting the diagnosis;
• the treatment, monitoring, referral or other management decision;
• the identity and credentials of the responsible provider and the date of service.

Not every chronic condition must generate a new treatment. Continuing medication, reviewing control, ordering surveillance or deciding that no change is needed may all be clinically meaningful, but the record must truthfully show what occurred. A diagnosis should never be added solely because it increases a risk score.

Use technology to expose gaps, not manufacture certainty

Electronic health records and analytics can identify possible gaps: an old condition still on the problem list, a medication associated with several diagnoses, or documentation that lacks expected specificity. These signals are prompts for review. They are not independent clinical confirmation and should not automatically populate a signed assessment.

Well-designed tools keep the evidence trail visible. A reviewer should be able to see which dated encounter supports the diagnosis, which provider documented it, which code-set version applies and whether the information reached the plan’s submission system correctly. Version control matters because ICD-10-CM codes and CMS model mappings can change on different schedules.

Organizations should also measure corrections in both directions. Finding omitted supported conditions can improve completeness, while removing unsupported or outdated diagnoses protects record integrity. A system rewarded only for adding codes creates a distorted incentive and misses half of the accuracy problem.

Audit readiness is part of the care-record workflow

CMS does not validate a diagnosis merely because it appeared in submitted data. During Risk Adjustment Data Validation, the agency checks whether diagnoses used for payment are supported by enrollees’ medical records; its current RADV program page states that unsupported diagnoses may lead CMS to collect overpayments. Audits occur after the final submission deadline, so organizations must preserve the supporting record and its encounter connection rather than attempt to recreate support later.

A practical internal review should sample both accepted and rejected diagnoses, trace each one to its encounter, and separate three questions: Was the condition clinically documented? Was the ICD-10-CM code supported? Was the diagnosis eligible under the applicable risk-adjustment rules? Combining those questions into a single “HCC captured” check can conceal why an error occurred.

The patient-care test remains equally important. When documentation identifies uncontrolled disease, a new complication or a missed follow-up need, the information should enter an accountable clinical workflow. Accurate coding can make patient complexity visible and support appropriate plan payment, but better outcomes depend on whether the care team acts on what the encounter established.

Also read:

• Few Industries will not be Transformed by AR and VR on 2025
• Tufts Report: 9.3 Million US Jobs at Risk from AI in the Next 5 Years — And the Hits Are Coming for High-Tech Hubs, Not Rust Belt Towns
• Stack Overflow Figured Out How Not to Die Because of AI. Will It Work?