Last updated: June 2026
Quick Answer: Risk adjustment coding is the process of assigning ICD-10-CM diagnosis codes to documented patient conditions so a Medicare Advantage plan’s payment reflects how sick its members actually are. Each code maps to a Hierarchical Condition Category (HCC), and those HCCs build a member’s Risk Adjustment Factor (RAF) score. In 2026 the bar is no longer how many codes you find. It is whether each diagnosis is linked to a real encounter, backed by MEAT evidence, and able to survive a RADV audit. This guide covers HCC coding, MEAT criteria, annual recapture, and defensible documentation.
Introduction
Risk adjustment coding is the systematic assignment of ICD-10-CM diagnosis codes to a patient’s documented conditions so that payment to a Medicare Advantage plan matches the real clinical complexity of its members [1]. It is different from fee-for-service coding, which bills for procedures. Risk adjustment coding captures health status: the chronic and severe conditions a plan is responsible for managing across a full year.
Here is what changed. For a decade, many programs treated this work as a hunt for more diagnoses. That era is closing. In 2026, regulators judge a coded chart by one question: can you prove it? A March 2026 Department of Justice settlement and a series of Office of Inspector General (OIG) audits have made the same point in public, with dollar figures attached. Accurate HCC coding still protects fair reimbursement for sicker patients. But accuracy now means defensible: every diagnosis tied to a face-to-face encounter, supported by clinical evidence, and ready for a RADV audit (https://www.raapidinc.com/blogs/radv-audits-2025/). This guide walks coders, compliance leads, and risk adjustment directors through how to get there.
Key Takeaways
- Risk adjustment coding assigns ICD-10-CM diagnosis codes to documented conditions, mapping them to HCCs that determine a member’s RAF score and a Medicare Advantage plan’s risk-adjusted payment [1].
- MEAT criteria (Monitor, Evaluate, Assess, Treat) are the evidence test auditors use to decide whether an HCC diagnosis is supported. Any one of the four, documented at the encounter, can validate a condition.
- Chronic conditions must be recaptured every calendar year. RAF scores reset each January 1 and do not carry forward.
- The biggest 2026 compliance risk is add-only coding: programs that submit new diagnoses but never remove unsupported ones. A March 2026 DOJ settlement penalized exactly that pattern [6].
- An OIG audit released in March 2026 found unsupported high-risk diagnosis codes in 247 of 271 sampled enrollee-years, a 91% error rate [5].
- Defensible coding is two-way: it adds codes that evidence supports and removes codes it does not.
What Is Risk Adjustment Coding?
Risk adjustment coding is the practice of translating a patient’s documented diagnoses into ICD-10-CM codes that feed the CMS-HCC risk adjustment model and set a member’s risk score [1]. It tells the Centers for Medicare & Medicaid Services (CMS) how complex a population is, so plans that care for sicker members receive appropriate reimbursement.
The mechanics are direct. A clinician documents a condition during a visit. A coder assigns the most specific ICD-10-CM code. That code maps to a Hierarchical Condition Category, and the member’s HCCs combine with demographic factors such as age and gender to produce a Risk Adjustment Factor (RAF) score. The risk adjustment model uses that score to predict healthcare costs and adjust payment. A plan treating members with heart failure, chronic kidney disease, diabetes, and morbid obesity carries a higher risk profile, and the model is built to fund that added cost.
Risk adjustment coding sits at the center of value based care. It is how a health plan proves, in data, that it is managing real disease burden and not simply enrolling healthier members.
How HCCs and RAF Scores Work
A Hierarchical Condition Category (HCC) is a payment grouping that bundles ICD-10-CM diagnosis codes sharing similar clinical profiles and expected healthcare costs. The CMS-HCC model assigns each qualifying diagnosis to a category, and the categories a member accumulates drive the math behind their risk score.
HCC assignments generate a member’s RAF score, which CMS recalculates each year. A score near 1.0 represents an average-cost member. Members with multiple chronic conditions can reach 2.5 to 4.0, reflecting the higher costs of managing complex disease [1]. The risk adjustment factor is additive across body systems: a member with diabetes, congestive heart failure, and chronic kidney disease accumulates HCCs from each, and those risk scores aggregate into one total.
The dollars are large. CMS paid an estimated $615 billion to Medicare Advantage plans in 2026, and risk scores decide how that money is distributed across plans and members [8]. Accurate HCC coding is what makes the model work: the risk scores behind every payment are only as reliable as the codes feeding them. For a deeper look at the current model year, see our guide to the CMS-HCC V28 model.
MEAT Criteria: The Evidence Standard Behind Every HCC
MEAT criteria (Monitor, Evaluate, Assess, Treat) are the four kinds of clinical evidence used to confirm that an HCC diagnosis was actively managed at an encounter, not just listed in a patient’s history [2]. Any one of the four, clearly documented, can support the code. It is an industry convention for reading a chart the way an auditor does, not a CMS-named rule.
Here is what each component looks like in a note:
- Monitor: tracking a condition over time through labs, vitals, or recurring checks.
- Evaluate: reviewing test results, imaging, or clinical findings tied to the condition.
- Assess: recording clinical judgment about current status and the management plan.
- Treat: an active intervention, such as a medication change or referral.
The link between MEAT documentation and RADV defensibility is tight. Without evidence that meets at least one MEAT component, an HCC can be invalidated during an audit and the payment recouped. A note that reads “history of CHF” with no current management does not support an active congestive heart failure HCC. A note that records an ejection fraction (evaluate), adjusts a diuretic (treat), and schedules a follow-up echo (monitor) does. For a closer look at applying these standards, see our breakdown of MEAT criteria in HCC coding.
Why Defensible Coding Replaced Revenue Capture in 2026
The 2026 enforcement record makes the case better than any vendor pitch could. Risk adjustment has moved from a capture problem to a compliance and accountability problem, and the cost of getting it wrong is now public.
In March 2026, the OIG published audit A-07-22-01207, a review of one Medicare Advantage organization. Of 271 sampled enrollee-years, 247 contained unsupported high-risk diagnosis codes, a 91% error rate [5]. Two condition groups, acute stroke and acute myocardial infarction, had 100% error rates. The most common pattern was a history-of condition coded as active: a past stroke submitted as an acute stroke, a resolved cancer coded as a current malignancy.
The Department of Justice went further. On March 11, 2026, Aetna agreed to pay $117.7 million to resolve False Claims Act allegations [6]. Of that, $44.2 million resolved claims that a retrospective chart review program added diagnosis codes to boost risk scores but failed to delete unsupported codes the same reviews had identified. The whistleblower was a former risk adjustment coding auditor. The OIG’s February 2026 Medicare Advantage compliance guidance reinforced the message: failing to remove unsupported codes is itself a compliance failure [7].
The lesson is structural. A coding program that only adds diagnoses, and never removes the ones evidence does not support, now reads to regulators as intent to inflate payment. Defensible coding is two-way. It adds what the chart supports and deletes what it does not. That is the foundation of defensible RAF scores, and it is why compliance officers, not revenue teams, increasingly own the coding standard.
Autonomous Retrospective Risk Adjustment Solution
One platform. Every HCC validated. Revenue secured.
How Does Annual Recapture Work?
Annual recapture is the requirement to document every chronic condition fresh each calendar year so it keeps counting toward a member’s RAF score. CMS does not carry diagnoses forward automatically. Risk scores reset every January 1, which means a condition documented last year contributes nothing this year unless it is captured again [3].
A member with type 2 diabetes (HCC 19) needs that condition documented, with specificity and MEAT evidence, in the current year for it to hold. When chronic conditions go unrecaptured, they drop out of the risk calculation, and the plan’s payment no longer reflects the member’s real complexity. Recapture is the part of HCC coding most programs underinvest in, and it is also where many lean too hard on retrospective cleanup. A cleaner path is to surface prior-year HCCs before the visit so clinicians can confirm or rule them out at the point of care.
Documentation Standards for Recapture
Recapture rests on a few non-negotiable documentation standards:
- A face-to-face encounter with a credentialed clinician authorized to diagnose the condition.
- ICD-10-CM codes assigned at the highest available specificity (E11.22 for type 2 diabetes with diabetic chronic kidney disease, not a generic diabetes code).
- At least one MEAT component documented for each condition in the current calendar year.
Generic documentation is where recapture fails. “Diabetes” supports far less than “type 2 diabetes mellitus with diabetic chronic kidney disease.” Specificity is the difference between a code that maps to an HCC and one that does not, and precise HCC coding lives or dies on that detail.
How ICD-10-CM Updates and V28 Affect Coding
ICD-10-CM code changes take effect each October 1, and they ripple straight into HCC mapping. New codes, retired codes, and revised crosswalks can change whether a diagnosis lands in an HCC at all [2]. The CMS-HCC model also evolves: the V28 model reshaped category mapping and trimmed some conditions that previously carried risk weight.
Two practical effects matter for coders. First, crosswalks between diagnosis codes and HCCs shift between model years, so a code that mapped to an HCC last year may not this year. Second, CMS has finalized the exclusion of chart-review diagnoses that are not linked to a clinical encounter, which raises the bar on evidence [4]. HCC coding teams need retraining whenever the model updates, and they need current crosswalks loaded before the new code year begins. Precise HCC coding starts with crosswalks that match the current model year. Our 2026 ICD-10-CM update guide (https://www.raapidinc.com/blogs/2026-icd-10-cm-updates-medicare-advantage/) tracks the specific changes.
How to Build a Risk Adjustment Coding Process
A reliable program turns these standards into a repeatable workflow. The steps below apply to retrospective chart review and concurrent coding alike.
- Identify managed chronic conditions. Review provider notes, specialist reports, labs, and pharmacy data to find conditions a clinician is actively managing.
- Validate MEAT evidence. Confirm at least one MEAT component exists in the record for each candidate HCC diagnosis before coding it.
- Code to the highest specificity. Assign the ICD-10-CM code that captures the condition fully, so it maps cleanly to the correct HCC.
- Check the recapture requirement. Verify each condition appears in the current calendar year with a qualifying face-to-face encounter.
- Build the evidence trail. Document the path from clinical note to submitted code, and remove any code the evidence does not support.
That last step is what separates a defensible program from a risky one. A two-way process that records both the adds and the deletes is what an auditor wants to see, and it is what RAAPID’s retrospective risk adjustment (https://www.raapidinc.com/blogs/retrospective-risk-adjustment/) approach is built around.
Documentation Types and RADV Audit Defense
Not every document carries the same weight when an auditor validates an HCC. Risk adjustment data validation compares submitted codes against the medical record, and weak source documents are where plans lose.
Documentation Type | MEAT Evidence Strength | RADV Defense Level |
Provider progress notes | Strongest when MEAT components are explicit | Highest; the primary source auditors trust |
Specialist reports | Strong with clinical detail and a treatment plan | High when signed and dated |
Health risk assessments | Limited without explicit MEAT | Weak; needs clinical validation and attestation |
During risk adjustment data validation, auditors pull the medical record for each sampled HCC and check it against MEAT evidence. This is where HCC coding either holds up or collapses. Plans that maintain a single source of truth, with EHR and claims data reconciled, defend their codes far more easily than plans chasing records across fragmented systems. The OIG’s 91% error rate in audit A-07-22-01207 shows what happens when that evidence is not there [5].
Common Coding Challenges and How to Fix Them
Three problems account for most lost HCC coding accuracy. Each has a practical fix.
Missing chronic condition documentation. High-prevalence conditions like diabetes, hypertension, chronic kidney disease, and major depression often go undocumented in a given year even when a clinician is managing them. Fix it with targeted chart review on high-impact conditions, clinician education on what specificity coders actually need, and pre-visit workflows that surface prior-year HCCs for confirmation.
Generic coding. When a clinician documents without specificity, the code may not map to an HCC at all. “Diabetes” instead of “type 2 diabetes mellitus with diabetic chronic kidney disease” is a coded chart that still loses risk-score accuracy. Fix it at the point of care, where clinical decision support can prompt for the detail before the note is signed.
Add-only review. The riskiest pattern, and the one regulators now target, is adding codes without ever removing unsupported ones. Fix it with a two-way process that deletes codes the evidence does not support. RAAPID’s Neuro-Symbolic AI links every suggested HCC to MEAT-based evidence in the note and flags codes that should come out, so coders work from an explainable, audit-ready trail. In practice this has cut chart review time to 8 to 12 minutes per chart and lifted coding-team productivity by 60 to 80 percent. Out-of-box AI accuracy runs about 92 percent, reaching 98 percent or higher after human review.
Certified Risk Adjustment Coders and Ongoing Training
Skilled people make the model work. A certified risk adjustment coder (CRC) is trained to read clinical documentation, apply MEAT criteria, and assign ICD-10-CM codes at the specificity HCC capture demands. Credentialing matters because the rules move: model years change, crosswalks update, and enforcement standards tighten.
Effective programs invest in continuous HCC coding education for their certified coders, covering new HCC model versions, the nuances of coding complex conditions such as congestive heart failure and major depression, and the evidence patterns auditors flag. Precise HCC coding is a skill that has to be maintained, not assumed. The goal is not speed for its own sake. It is consistency: coders who apply the same defensible standard to every chart, every time.
Where Risk Adjustment Coding Is Headed
The direction is set. Value based care keeps expanding, Medicare Advantage enrollment keeps growing [9], and risk scores keep driving how plans are paid. What is changing is the standard of proof. Modern risk adjustment software now does the heavy lifting of surfacing candidate conditions and checking evidence, but the human coder keeps final authority over what gets submitted.
The plans that do well will treat risk adjustment coding as a clinical and compliance discipline, not a revenue lever. They will document real patient complexity, link every diagnosis to a real encounter, and keep an evidence trail that holds up under scrutiny. That is the whole point of accurate HCC coding in 2026: the right codes, for the right clinical reasons, that an auditor, a clinician, and a CFO can all trust.
Frequently Asked Questions
Risk adjustment coding assigns ICD-10-CM diagnosis codes to a patient’s documented conditions so a Medicare Advantage plan’s payment matches how complex its members are. The codes map to HCCs, which build the member’s RAF score and adjust reimbursement for expected healthcare costs [1].
Regular medical coding bills for specific services and procedures at a single visit. HCC coding captures a member’s chronic and severe conditions across the year to set a risk score. HCC coding relies on documentation of health status, not procedure volume, and feeds the CMS-HCC risk adjustment model [1].
MEAT stands for Monitor, Evaluate, Assess, and Treat. Any one of these four kinds of evidence, documented at an encounter, shows a condition was actively managed and supports the HCC diagnosis during a RADV audit [2]. Without MEAT evidence, an HCC can be invalidated and the payment recouped.
RAF scores reset each January 1 and do not carry forward. A chronic condition like diabetes must be documented again in the current calendar year, with specificity and MEAT evidence, or it drops out of the member’s risk score and the plan’s payment no longer reflects that complexity [3].
Risk adjustment data validation (RADV) is the CMS audit that checks submitted diagnosis codes against the medical record. Auditors validate each sampled HCC against MEAT evidence. Codes without supporting documentation are removed, and unsupported codes can trigger recoupment, which is why defensible documentation matters [3].
Add-only coding submits new diagnoses to raise risk scores without removing unsupported ones. A March 2026 DOJ settlement penalized a plan $44.2 million for exactly this pattern [6], and 2026 OIG guidance treats failure to remove unsupported codes as a compliance failure [7]. Defensible coding is two-way: add and remove.
A certified risk adjustment coder (CRC) credential signals training in reading clinical documentation, applying MEAT criteria, and assigning ICD-10-CM codes at HCC-level specificity. It is not legally required for every role, but most plans and vendors expect it because model years and rules change often.
Ready to Evaluate a Collaborative Path?
Over a Partnership Evaluation Call.
Conclusion and Next Steps
Accurate risk adjustment coding in 2026 rests on three habits: MEAT-supported documentation, disciplined annual recapture, and an evidence trail that survives a RADV audit. The shift from capture to care is not a slogan. It is written into DOJ settlements and OIG audits, with real dollars attached. Plans that code defensibly protect both their members’ care and their own standing with regulators.
Start here:
- Audit current HCC capture on high-prevalence chronic conditions to find recapture gaps.
- Educate clinicians on the specificity coders need, not just the diagnoses.
- Build a two-way review that removes unsupported codes, not only one that adds.
- Create a single source of truth for member risk that reconciles EHR and claims data.
See how RAAPID’s Clinical AI Platform makes every HCC explainable and audit-ready. Book a demo.
Internal RAAPID benchmark. Figures reflect RAAPID client engagements and proof-of-concept results, not an independent published study.*
About the author
Wynda Clayton, MS, RHIT, CRC
Director of Risk Adjustment Coding & Compliance, RAAPID
Wynda Clayton, MS, RHIT, CRC, Director of Risk Adjustment Coding & Compliance at RAAPID and a former CMS RADV auditor. Wynda leads RAAPID's clinical and compliance review and writes on defensible coding, MEAT documentation, and audit readiness.
Sources
[1] CMS, “Risk Adjustment,” Centers for Medicare & Medicaid Services
[2] ICD-10-CM Official Guidelines for Coding and Reporting, CDC National Center for Health Statistics
[3] CMS, “Medicare Advantage Risk Adjustment Data Validation Program”
[4] CMS, “2027 Medicare Advantage and Part D Rate Announcement”
[7] OIG, “Medicare Advantage Industry-Specific Compliance Program Guidance” (February 2026)
[8] MedPAC, “Report to the Congress: Medicare Payment Policy” (March 2026)
[9] KFF, “Medicare Advantage in 2025: Enrollment Update and Key Trends”
