Risk Adjustment Documentation · Hub
MEAT Criteria for HCC Coding
MEAT — Monitor, Evaluate, Assess, Treat — is the documentation standard CMS auditors look for on every chronic condition reported for Medicare Advantage risk adjustment. A diagnosis on a claim without MEAT-level evidence in the chart is a deletion waiting to happen during a RADV review.
This hub walks through what each letter requires, how CMS encoded the standard in the 2023 RADV final rule, how MEAT shows up in the OIG's recent audit work, and the provider-query templates coders actually use when a chart is thin.
Quick answer
MEAT criteria is how risk adjustment coders remember what counts as sufficient chart evidence. For every HCC on a claim, the encounter note must show at least one of: the provider is monitoring the condition (labs, vitals, symptom review), is evaluating its response to therapy, is assessing its severity or stability, or is treating it actively (medications, procedures, counseling). One MEAT-supported visit per calendar year is the CMS minimum; two to three per year is the compliance target most payers use to absorb audit risk.
What each MEAT letter means in practice
M — Monitor
Monitoring is any documented observation that tracks the condition over time: A1c trending, blood pressure log, INR range, weight change on heart-failure visits, peak-flow numbers on a COPD follow-up. Monitoring does not require a new intervention — it requires the provider to look. A diabetic patient whose chart note records "A1c 7.2 today, up from 6.8 in January" has Monitor evidence for E11.9 — Type 2 diabetes without complications, even if no plan changes were made.
E — Evaluate
Evaluation is the provider's judgment about how the condition is responding to current management. Phrases like "controlled on current regimen," "improving since referral," or "worsening despite dose increase" are Evaluate evidence. Evaluation is what distinguishes a thoughtful HCC submission from a problem-list carry-forward: a condition that "exists" but is never evaluated is exactly the pattern RADV auditors remove from the payment year's RAF.
A — Assess
Assessment is the clinician's severity or stability call — "heart failure with reduced ejection fraction, NYHA class II, stable," or "Type 2 diabetes, uncontrolled." Assess evidence is what justifies coding more-specific ICD-10 codes over the generic unspecified ones, and it is the element coders most often cite when querying providers for specificity. An Assess statement is also what anchors the correct HCC category when two codes could compete — see the HCC hierarchy and trumping rules for when severity-level assessment changes the HCC that pays.
T — Treat
Treatment is the most straightforward letter: any active intervention counts. Prescription refills, dose adjustments, counseling, ordered labs, referrals, or procedures. Even a refill of a standing medication ("continue metformin 1000 mg BID") is Treat evidence. The gotcha is status codes — an amputation status (Z89.xx) has no ongoing treatment, so Treat rarely applies; it is usually documented via Monitor or Assess instead.
Why CMS requires MEAT-level evidence
Medicare Advantage plans are paid a risk-adjusted per-member, per-month capitation tied to the HCCs submitted for each enrollee. Because the payment is prospective — higher RAF equals higher revenue — CMS needs a retrospective check that every HCC claimed was really present and really managed. That check is Risk Adjustment Data Validation (RADV), codified in the ICD-10-CM Official Guidelines for Coding and Reporting and operationalized through the CMS Medicare Advantage risk adjustment program.
The HHS Office of Inspector General's V24/V28 trend report documents that a small number of HCC categories account for the majority of RAF inflation risk: acute stroke, acute myocardial infarction, major depression, and morbid obesity. These categories share a pattern — they carry significant RAF weight and are coded from diagnoses that can be easy to infer from a problem list without ever being actively managed in the current year. MEAT evidence is what separates a legitimate chronic-care claim from an audit deletion in these high-risk categories.
CMS's 2023 RADV final rule also ended the fee-for-service adjuster, which means unsupported HCC deletions now flow through to payment recovery without a buffer. A plan that routinely submits HCCs without MEAT-level evidence is no longer just risking the HCC — it is risking the extrapolated recovery across the entire contract sample. That is the regulatory pressure that made MEAT documentation a board-level compliance concern for most MA plans.
How MEAT shows up in a RADV audit
A RADV audit starts with CMS or its contractor pulling a sample of enrollees and requesting the full medical record for a specific payment year. For each HCC submitted on each enrollee's claims, the auditor reads the record looking for MEAT-level evidence in at least one face-to-face encounter during that calendar year. The encounter must be from an acceptable provider type (most commonly physicians, NPs, and PAs — home health visits and most lab-only visits do not count), and the diagnosis must be supported by sufficient documentation to be coded per the CMS ICD-10-CM Official Guidelines.
If the auditor cannot find MEAT evidence, the HCC is removed. The resulting RAF recalculation reduces the capitation paid to the plan, and under the 2023 final rule that reduction can be extrapolated across the contract — turning a handful of missing MEAT statements into an eight-figure recovery. The 2026 RADV audit expansion and payment year 2020 RADV that covers 471 contracts have raised the stakes: MEAT documentation is no longer the payer compliance team's problem alone — it is a shared responsibility between coders, providers, and the EHR's problem list.
A separate CMS proposal — the proposed ban on unlinked chart reviews for MA plans — would further tighten the link between submitted HCCs and documented encounters, making MEAT evidence the only defensible audit trail for retrospectively added diagnoses. Coders preparing for the expansion should also review the RADV audit prep guide and the broader CMS-HCC V28 model overview to understand which categories are most exposed under the expanded audits.
MEAT vs TAMPER — which framework to teach providers
Different health systems and coding vendors teach different documentation frameworks. The two most common are MEAT (four letters: Monitor, Evaluate, Assess, Treat) and TAMPER (six letters: Treatment, Assessment, Monitor/Medicate, Plan, Evaluate, Referral). TAMPER breaks Treatment into two pieces — the acute treatment and the follow-up referral — and explicitly names "Plan" as a separate element.
In practice both frameworks substantiate the same diagnosis in the same RADV audit. Auditors do not check for a specific framework label; they check that the documentation contains one or more of the underlying elements. The AAPC risk adjustment MEAT brief remains the most widely circulated provider-facing reference, which is why MEAT tends to win in provider education decks — fewer letters, faster to remember, same audit outcome.
The one place TAMPER's extra granularity matters is when a specialist consult is the only MEAT-adjacent evidence in the chart. TAMPER's explicit "Referral" element makes it easier to defend a chronic-condition HCC that was otherwise only acknowledged by a PCP referring out. If your organization's audit-prep workflow depends heavily on specialist referrals, TAMPER's explicit Referral element can help provider-query writers pull the right snippet.
MEAT expectations by condition type
Not every HCC lives up to MEAT the same way. The framework bends depending on whether the condition is chronic, acute, status, or resolved.
Chronic progressive conditions
Diabetes (E11.x), heart failure (I50.x), CKD (N18.x), COPD (J44.x) — these produce MEAT evidence naturally because the provider is actively managing them. The risk is that the management is so routine the note collapses to "continue current regimen," which can fail the Evaluate element on audit. See the condition-specific guides for diabetes, heart failure, CKD, and COPD for condition-specific MEAT patterns.
Acute conditions
Acute MI, acute stroke, acute pulmonary embolism — these produce dense MEAT evidence during the acute encounter but often disappear from subsequent notes once the patient stabilizes. If the plan is submitting the acute ICD-10 code (rather than the "history of" Z-code), MEAT evidence has to live in the acute encounter itself; afterward the patient's diagnosis should shift to the chronic or history variant. The same pattern applies to I50.9 heart failure unspecified once a more specific heart-failure code has been assigned, and to J44.1 COPD with acute exacerbation once the exacerbation resolves.
Status codes (Z-codes)
Amputation status, ostomy status, transplant status, and other Z-codes capture a permanent anatomical change. Because the "condition" is not actively progressing, Treat rarely applies — MEAT evidence usually lives in Monitor (e.g., "stump site intact, no breakdown") or Assess ("above knee amputation, prosthesis fit stable"). Status codes are also the codes most often lost in audit because providers stop mentioning them once the condition is chronic.
Resolved or "history of" conditions
History-of codes (Z86.xx, Z87.xx) do not carry an HCC weight — they are the opposite of a MEAT target. A cancer patient who is NED (no evidence of disease) should be coded to Z85.xx (personal history of malignancy), which maps to no HCC, not to the active cancer code which would require MEAT evidence of ongoing treatment. The compliance risk here is reversed: submitting an active code without MEAT evidence when a history code is more accurate.
The five most common MEAT failures — and how to fix them
- Problem-list carry-forward without narrative. The fix: a single assessment-and-plan sentence per chronic HCC. "Diabetes, stable on metformin, A1c 6.9, continue regimen" is MEAT in fifteen words.
- Unspecified codes when specificity is available. An E11.9 when the chart supports E11.22 (diabetes with diabetic CKD) is a RAF leak and an Assess failure. The fix: query the provider using the provider-query templates and re-encode with the specified code through the HCC Buddy encoder.
- Status codes dropped from chronic-care visits. Above-knee amputation (Z89.619) should be documented annually even if the patient never mentions it. The fix: an annual status-code sweep during the patient's Medicare Advantage Annual Wellness Visit.
- "History of" vs active-condition confusion. Coding an active cancer ICD-10 on a post-remission patient is an audit findable in the opposite direction — the chart will not support active treatment MEAT. The fix: review the top ten commonly miscoded HCCs every quarter and calibrate against current documentation.
- V24 vs V28 category drift. Some conditions that carried HCC weight under V24 no longer do under V28, and vice versa. MEAT evidence that was "sufficient" for a V24-weighted diagnosis may sit on a code that no longer maps in V28, wasting the provider's documentation time. See the V24 vs V28 comparison and check each code's status in the ICD-10 to HCC mapping tool.
MEAT provider-query templates
Provider queries must be compliant under AHIMA and ACDIS query guidelines: non-leading, based on the record, and open-ended enough to allow the provider to clarify without being steered to a specific code. The templates below are neutral prompts that elicit the missing MEAT element without suggesting a diagnosis.
Missing Evaluate
The chart documents [condition]. Based on today's encounter, can you indicate whether this condition is stable, improving, or worsening on the current treatment plan?
Missing Assess — unspecified code
The documentation notes [unspecified condition]. If clinically appropriate, can you document the specific type or severity (e.g., controlled vs uncontrolled, with or without complications) to support the most accurate ICD-10 assignment?
Missing Monitor
[Condition] appears on the problem list but is not addressed in today's note. Can you document any monitoring performed (labs, vitals, symptom review) or indicate whether the condition is no longer present?
Full templates are in the provider-query templates guide. The AHIMA/ACDIS rules of neutrality apply to all of them.
Free MEAT-included cheat sheet
The HCC Buddy cheat sheet includes a printable MEAT quick-reference alongside the V28 decision tree and the top 20 HCC categories. No email for the preview — the full PDF is gated by email.
Get the cheat sheetFrequently Asked Questions
Related coding tools
ICD-10-CM Encoder
Look up any code and see V24/V28 HCC, RAF, and MEAT reminders inline.
ICD-10 to HCC mapping
Type a code, see the HCC. The full mapping hub for 72K codes.
CMS-HCC V28
The 2026 model deep dive — what changed, what pays, what to update.
RAF Score Calculator
Build a full RAF score from a diagnosis list, V24 and V28.
Payer Guidelines Library
Search CMS, Humana, Aetna, and your uploaded payer PDFs.
HCC Buddy Chrome Extension
MEAT reminders and HCC mapping inside your EHR.
Built by a Certified Risk Coder. Sourced from CMS, OIG, and AAPC.
Zero PHI · 2026 CMS V28 current · Reviewed against ICD-10-CM FY2026 Official Guidelines