TL;DR — V28 HCC Documentation Strategy for Medical Directors:

• The CY2027 Advance Notice confirms the full transition to the V28 CMS-HCC model, which removes many "vague" diagnosis pathways and demands explicit clinical specificity.

• Under V28, dictating "diabetes, CKD" without stage, causal linkage, and MEAT (Monitor, Evaluate, Assess, Treat) forfeits HCC value—an estimated 0.28 RAF drop (~$2,500/year in capitation) per affected member.

• Scribing.io embeds a real-time V24→V28 RAF delta simulator that fires inline when a chronic condition is about to lose HCC value, prompting the clinician to verbalize causal linkage and stage.

• Human-attested MEAT evidence is persisted in FHIR Condition.evidence, restoring risk score AND creating an audit-ready trail (e.g., E11.22 + N18.32).

• The V28 Cliff and the CY2027 Advance Notice

• MEAT, Causal Linkage, and the Anchor Truth

• The 72-Year-Old Diabetic CKD Follow-Up

• Technical Reference for ICD-10 Standards

• Operational Rollout for Medical Directors

• Compliance and RADV Audit Posture

Medicare V28 HCC Risk Adjustment: An AI Documentation Strategy for Medical Directors

The V28 Cliff: Why the CY2027 Advance Notice Redefines Documentation Risk

CLINICAL UPDATE 2026: Revised for new CMS CPT G2211 standards, SB 1120 compliance, and FHIR interoperability.

The CY2027 Advance Notice, released January 26, 2026, finalizes the payment trajectory Medical Directors have braced for. The CMS-HCC V28 model is now the operative framework (Attachment VI, Table VI-4). The document is exhaustive on actuarial mechanics but silent on clinical workflow.

That silence is the entire problem. V28 restructured HCC categories, expanded the count of codes mapping to no-payment categories, and raised specificity requirements. Understand how Scribing.io closes that gap before rate-setting math ever reaches your panel.

For a Medical Director managing a Medicare Advantage panel, the risk is no longer under-coding. It is documenting a real, treated condition in a way V28 refuses to risk-adjust. This playbook addresses that operational blind spot directly.

Explore how documentation standards vary across our Clinical Specialties Directory and how capture connects to your systems in our EHR Integration Library.

What the Advance Notice Left Unsaid: MEAT and Causal Linkage

The CMS Advance Notice is the authoritative source of the numbers. What it structurally cannot provide is the documentation bridge between a treated patient and a captured risk score. This is the information-gain pillar of the playbook.

The V28 transition requires explicit MEAT documentation—Monitor, Evaluate, Assess, Treat—for every chronic code. A diagnosis floating in a problem list without contemporaneous MEAT evidence is, under audit, an unsupported diagnosis.

Under V28's tighter hierarchies, unsupported diagnoses are the first to fall out of a RAF recalculation and the first a RADV audit challenges. The gap between clinical reality and captured value is where capitation quietly bleeds.

The Delta Simulator: Detecting Forfeited HCC Value

No rate notice or competing scribe addresses this: the moment of forfeiture is invisible to the clinician at the point of care. A physician who says "diabetes, CKD" believes the condition is documented. Clinically, it is.

But under V28, that phrasing may drop the encounter's HCC value silently. Scribing.io embeds a real-time V24→V28 RAF delta simulator inside the note that compares legacy capture against V28 capture for the same dictated language.

When the delta simulator detects forfeited value, it fires an inline flag and auto-prompts the clinician to verbalize causal linkage and stage. The prompt targets the exact missing element—not a generic reminder.

• Continuous dual-model comparison runs against both V24 and V28 hierarchies as language is captured.

• Inline flags fire only when a chronic condition is about to lose HCC value—no alert fatigue.

• Human-attested MEAT evidence persists in FHIR Condition.evidence as a clinician-verbalized fact, never an AI inference.

This is the difference between an audit-vulnerable auto-code and an audit-ready evidentiary chain. The clinician evaluates; Scribing.io captures and structures.

Scribing.io Clinical Logic: The 72-Year-Old Diabetic CKD Follow-Up

The following scenario is the canonical V28 failure mode—and the demonstration of the clinical logic engine within Ambient Clinical Intelligence.

The Encounter reads as follows: a 72-year-old with type 2 diabetes and chronic kidney disease returns for follow-up. The clinician dictates "diabetes, CKD" and adjusts medications, but never states stage, cause, or monitoring plan.

The Silent Failure without intervention: under V28, this encounter fails to risk-adjust. The unspecified codes drop out of the HCC hierarchy, and the RAF and capitation impact follow directly.

The Scribing.io Intervention fires inline. As the clinician dictates, the V24→V28 delta flag prompts a targeted clinical question:

"Is CKD due to diabetes? What's the latest eGFR/stage and treatment?"

The clinician verbalizes the specificity: "Type 2 diabetes with CKD stage 3b; eGFR 38; continue ACEI; labs q3 months."

The Result restores full value. The chart captures complete MEAT with E11.22 (ICD-10-CM) and N18.32 (ICD-10-CM). Risk is restored and the audit trail is intact.

Ten additional seconds of dictation recovered an estimated $2,500 in annual capitation—the core equation modeled in our AI Medical Scribe ROI Calculator.

Technical Reference: ICD-10 Documentation Standards

The two codes at the center of the diabetic-CKD scenario are the archetypes of V28 specificity. Both require explicit clinical language to trigger—neither can be inferred safely from vague dictation.

Note the coding-pair dependency here. E11.22 alone confirms causality but N18.32 supplies the stage that V28 risk-adjusts. Both must appear, and both must be MEAT-supported within the encounter.

Operational Rollout: Deploying the Strategy Across a Panel

A Medical Director does not roll out documentation change one clinician at a time. Sequence the deployment against your highest-risk chronic cohorts first, where forfeiture density is greatest.

1. Identify the V24→V28 delta cohort: members whose prior-year RAF depended on codes now demoted under V28 hierarchies.

2. Enable inline delta prompting for those cohorts before annual wellness visits and chronic follow-ups.

3. Review FHIR Condition.evidence output weekly to confirm MEAT completeness, not merely code presence.

Governance sits with clinical leadership, not coding vendors. Scribing.io captures the clinician's verbalized attestation; the physician remains the sole author of the diagnostic judgment. Compare deployment tiers in Scribing.io Pricing & Plans.

Compliance and RADV Audit Posture Under V28

RADV audits in 2026 challenge the evidentiary chain, not the code string. A code without contemporaneous MEAT is an extrapolation liability across your entire sampled population.

Clinical-Grade Scribing preserves the attestation as a discrete, timestamped FHIR element tied to the encounter. Review applicable state and federal requirements, including SB 1120 disclosure standards, in our AI scribe legal reference.

• Every risk-adjusting diagnosis carries monitor, evaluate, assess, and treat evidence within the same note.

• Causal linkage is verbalized by the clinician, never auto-asserted by the model.

• The audit trail reconstructs the clinical reasoning, satisfying RADV documentation-support standards.

The strategic conclusion is direct: V28 penalizes vague language, not sick patients. Medical AI Scribing that prompts specificity at the point of care converts clinical truth into captured, defensible risk score.