MEAT Criteria in Risk Adjustment: How to Code a Condition That Survives RADV

The short answer: MEAT criteria in risk adjustment require that each coded chronic condition be supported in a face-to-face encounter note by evidence that the provider monitored, evaluated, assessed or treated it. Coders should tie every HCC to a specific encounter and MEAT element, delete unsupported diagnoses, query providers on documentation gaps and keep an audit trail for RADV. Plans and risk-bearing groups often run this work through an HCC risk adjustment coding team.

What MEAT criteria mean

MEAT stands for monitor, evaluate, assess or address, and treat. A chronic condition is supportable for risk adjustment when the encounter note shows the provider did at least one of these for that condition at that visit: tracked symptoms or results, reviewed tests, documented the status or decided on a plan, or prescribed or adjusted treatment.

A condition that appears only on a problem list, in a past medical history or in a claim, with no evidence of attention at the visit, does not meet the standard.

Start with an acceptable encounter

Before checking MEAT, confirm the encounter itself counts: a face-to-face or qualifying telehealth visit with an acceptable provider type, in the measurement year, with a signed and dated note. Diagnoses from records that do not qualify cannot be submitted, however well documented.

Encounter validation is the first step on every chart, and the most common source of errors for coders new to risk adjustment.

Record the evidence for every code

For each coded HCC, record the encounter date, the provider, the MEAT element and the supporting text from the note. That makes the coding decision reviewable in seconds by a second-level reviewer, a compliance auditor or a RADV reviewer.

Coders who cannot point to the supporting text should not code the condition. See medical coding for the wider coding standards involved.

Supported and unsupported, side by side

A note that lists diabetes with a current A1c, a medication adjustment and a plan supports the condition. A note that lists diabetes in the history with nothing else at that visit does not. A note documenting heart failure with current symptoms, an echo review and a diuretic change supports it; heart failure copied forward in a template does not.

Templates and copied-forward text are a frequent source of unsupported codes, so coders should read for evidence of attention at that visit.

Coding in both directions

Accurate risk adjustment removes as well as adds. When a previously submitted diagnosis is not supported by the record, it should be flagged for deletion and logged with the reason. Programs that only add codes raise risk scores in the short term and audit findings later.

Report deletions alongside additions, so compliance and leadership see that the program codes to the evidence in both directions.

Querying providers on gaps

Sometimes the record suggests a condition, such as a medication or a lab result, without documentation at an acceptable encounter. The right response is a compliant, non-leading query to the provider or a flag for a visit, not a code. Track queries, responses and any addenda.

Query patterns by provider are useful documentation-improvement feedback. See health plan and payer services for provider operations programs.

Second-level review and accuracy

Re-review a sample of each coder’s charts with a senior coder, score agreement and feed results back. Report accuracy by coder and by HCC category so training targets the conditions where errors cluster.

Document the QA methodology itself, so compliance teams and auditors can inspect how accuracy is measured.

Preparing for RADV

For RADV, assemble the best supporting record for each sampled HCC: a qualifying encounter with clear MEAT evidence. Flag HCCs with weak support before submission rather than after.

Charts retrieved late or incomplete weaken RADV packets. See medical records retrieval for how records reach the coding team on time.

Concurrent review and timing

Retrospective review reads the full year’s records after the fact, when it is too late to fix documentation. Concurrent review looks at encounters within days of the visit, while a provider can still add a compliant addendum and the patient’s condition is fresh in mind.

Most programs use both: concurrent review to improve documentation as the year goes on, and retrospective review to make sure every supported condition is captured and every unsupported one removed before submission.

Keeping up with model changes

Risk adjustment models change. The mapping of diagnosis codes to HCCs, the categories themselves and their weights can all shift between years, and different programs such as Medicare Advantage and ACA use different models. Coders need to know which model applies to which members and year.

Assign someone to track model and guidance updates, retrain coders when they change and confirm that coding tools reflect the current mapping.

Watching for template and copy-forward text

Electronic health record templates and copy-forward functions can fill notes with conditions that were not addressed at the visit. A long problem list that is identical across months of notes is a warning sign. Coders should look for evidence specific to that encounter, such as current results, changes in medication or a stated plan.

Patterns of copied-forward text by provider are worth sharing with clinical leaders as documentation feedback, with examples.

Turning findings into documentation feedback

Coding findings are most useful when they reach providers in a form they can act on. Summaries by provider showing common gaps, such as conditions listed without status or plan, or chronic conditions not addressed during the year, with examples from their own notes, lead to better documentation than general training alone.

Keep feedback educational and compliant: the aim is accurate documentation of the care provided, not more codes.

Metrics that show the program is working

Useful metrics include coding accuracy on second-level review, the deletion rate on previously submitted codes, chronic condition recapture year over year, query response rates and RADV sample results. Charts per hour belongs on the report, but never as the main measure.

Report these by provider group and coder, so leaders can see where documentation and coding are improving and where they are not.

Building MEAT into every chart

Train coders to validate the encounter, find the MEAT evidence, record it, delete what is unsupported and query what is missing, then measure accuracy with second-level review. Model updates should trigger retraining.

Our HCC risk adjustment coding outsourcing page explains how vetted Philippine teams of certified coders run retrospective and concurrent review for US plans, and our guide to ABA authorization tracking covers another healthcare revenue control.