HCC Risk Adjustment Coding 2026

What V28's Full Phase-In, the MEAT Standard, and Full-Scale RADV Audits Mean for Your Documentation

By Yagnesh Dave  ·  October 9, 2026  ·  11 min read

One unvalidated HCC on a RADV sample doesn't cost a Medicare Advantage plan one patient's risk score. It costs a percentage of the whole contract. In the OIG's audit of MMM Healthcare's contract H4003, reviewers sampled 688 enrollee-HCC pairs and couldn't validate 108 of them in the medical record, a 16% failure rate that got extrapolated across the full MA population the plan covered that year. That's the mechanic every HCC coder and every practice billing risk-bearing Medicare Advantage panels needs to understand in 2026, because 2026 is the first payment year scored entirely on the new V28 model, and it's also the year CMS stopped sampling a handful of plans and started auditing every RADV-eligible contract, every year.

This isn't a coding nuance for a specialty niche. If your practice bills under a Medicare Advantage risk-sharing or value-based contract, your documentation is now, indirectly, the plan's audit evidence. A thin note doesn't just fail to raise a risk score, it creates extrapolated liability someone eventually traces back to the chart you wrote.

Here's what changed with V28, what MEAT documentation actually requires, and where the audit exposure sits in 2026.

HCC Risk Adjustment Coding in 2026: What the V28 Model Changed

CMS pays Medicare Advantage organizations a risk-adjusted monthly amount per enrollee, calculated from the diagnoses that map to Hierarchical Condition Categories, or HCCs, documented and coded for that enrollee in the data collection year. The more and the higher-weighted the HCCs, the higher the payment. CMS finalized the move to the CMS-HCC V28 model in the CY 2024 Medicare Advantage and Part D Rate Announcement, and it phased the model in over three payment years rather than switching all at once.

V28 isn't a cosmetic refresh. It re-weighted roughly 2,300 ICD-10-CM codes and dropped more than 200 codes out of the model entirely, which means some diagnoses that used to raise a risk score in V24 no longer do, and others carry different weight than they did two years ago. A practice coding the same conditions the same way it did in 2023 can see a materially different risk score in 2026, not because the patient changed, but because the model did.

The MEAT Standard: What a Chart Needs to Support an HCC

CMS doesn't pay for a diagnosis code. It pays for a documented, face-to-face clinical encounter that supports the diagnosis in that calendar year, every year, regardless of whether the condition was addressed the year before. The documentation framework coders use to test whether a note supports an HCC is MEAT: Monitor, Evaluate, Assess, Treat.

A diagnosis sitting only on a problem list, copied forward from last year's note without any of the four MEAT elements this year, is exactly the kind of chart that fails a RADV sample. It isn't fraud, usually. It's a documentation habit that worked fine when nobody was checking every contract every year. That habit doesn't survive 2026's audit scale.

Think about the revenue math on a single mid-weighted HCC, worth roughly $1,000 to $3,000 a year in risk-adjusted payment depending on the condition and the plan's benchmark. A primary care panel of 600 Medicare Advantage lives that's under-capturing or under-documenting even 15% of its legitimately supportable HCCs is leaving tens of thousands of dollars in accurate, defensible revenue on the table every single year, separate from and in addition to the audit exposure on the other side of the ledger.

RADV Audits in 2026: Full Scale, Full Extrapolation

The Risk Adjustment Data Validation program is how CMS checks whether the diagnoses a plan submitted are actually supported in the medical record. Under the RADV final rule, CMS-4185-F2, finalized in early 2023, CMS extrapolates audit findings starting with payment year 2018 forward, and it chose not to apply a Fee-for-Service Adjuster that would have offset the extrapolated recovery amount. That combination, extrapolation without an offset, is what turns a sample of a few hundred charts into a contract-wide repayment demand.

For 2026, CMS moved from auditing a small rotating sample of contracts to auditing every RADV-eligible Medicare Advantage contract every year, a scale increase from roughly 60 plans historically reviewed to all eligible contracts, matched by a large expansion of CMS's certified coding review workforce to handle the volume. If your practice's patients are enrolled in a contract CMS is reviewing, and in 2026 that's effectively every contract, your documentation from 2024 and 2025 encounters is part of what gets pulled.

This is also why risk-sharing and value-based Medicare Advantage contracts increasingly push retrospective chart review and documentation requirements down to the practice level. The plan carries the RADV liability on paper, but the plan's actuaries and compliance teams know exactly where the unsupported diagnoses came from, and that knowledge shows up in the next contract negotiation.

Building an HCC Workflow That Survives an Audit

The practices that come through RADV review clean aren't doing anything exotic. They're running a few disciplined habits consistently, year over year, instead of relying on a single annual wellness visit to recapture every chronic condition at once.

None of this requires slowing down a visit. It requires one or two extra sentences in the note that tie the diagnosis to what actually happened in the room, which is the same documentation a well-run practice should want for clinical continuity regardless of what CMS pays for it.

Frequently Asked Questions About HCC Risk Adjustment Coding

What is the MEAT standard in HCC risk adjustment coding?

MEAT stands for Monitor, Evaluate, Assess and Treat. For a chronic condition to support an HCC in 2026, the medical record needs a face-to-face note showing the condition was monitored through signs or symptoms, evaluated through test results or response to treatment, assessed in the clinical decision-making, or treated with a medication, therapy or referral tied to that diagnosis by name. A diagnosis listed only in a problem list or copied forward from a prior note without one of these four elements won't hold up under a RADV audit.

What changed with the CMS-HCC V28 risk adjustment model?

V28 re-weighted roughly 2,300 ICD-10-CM codes and removed more than 200 codes from the risk adjustment model entirely, changing which diagnoses raise a patient's risk score and by how much. CMS phased V28 in over three payment years: 33% of the risk score ran on V28 in 2024, 67% in 2025, and 2026 is the first payment year scored 100% on V28, with the older V24 model no longer blended in for non-PACE organizations.

How far back can a RADV audit go, and can it extrapolate?

Under the RADV final rule (CMS-4185-F2), CMS extrapolates audit findings starting with payment year 2018 forward, meaning a sample of unsupported diagnoses found in a handful of charts can be projected across the entire contract's enrollment for that year. CMS does not apply a Fee-for-Service Adjuster to offset that extrapolated amount, which is why a small sample finding can turn into a large repayment demand.

Who is responsible for HCC coding accuracy, the practice or the Medicare Advantage plan?

Both carry risk. The Medicare Advantage organization submits the diagnosis data to CMS and owns the RADV repayment liability, but that data comes from the provider's documentation and the practice's or vendor's coding of the encounter. Plans increasingly push documentation requirements and retrospective chart review back onto practices through risk-sharing and value-based contracts, so a practice's MEAT gaps become the plan's audit exposure and, eventually, the practice's contract problem.