On June 1, 2026, CMS published an interim final rule implementing the Medicaid community engagement requirements from the One Big Beautiful Bill Act. States have until January 1, 2027, to comply. What most providers have not yet absorbed is that this rule creates a direct ICD-10 coding obligation. Whether a patient keeps Medicaid coverage may now depend on what a coder documents in the record.
What CMS Finalized
The interim final rule, CMS-2454-IFC, requires certain Medicaid enrollees to complete 80 hours per month of qualifying activities — employment, education, vocational training, or community service — as a condition of eligibility. The rule took effect immediately upon publication, though states have until January 1, 2027, to complete implementation. A public comment period runs through July 31, 2026.
The rule applies broadly to non-elderly adults in Medicaid expansion populations, but it carves out exemptions for individuals who are medically frail, pregnant, caregiving, or actively enrolled in substance use disorder (SUD) treatment. Active SUD treatment qualifies as a full exemption — a significant concession CMS made after sustained advocacy from behavioral health providers. Those in addiction recovery treatment do not need to meet the work hour threshold.
These exemptions are not automatic. They require documentation. And that documentation lives in the clinical record, in the form of ICD-10 diagnosis codes.
Why This Is a Coding Problem
Most of the attention on this rule has focused on patient outreach, state system builds, and eligibility verification workflows. Very little has been written about the coding obligation it creates. Yet the exemption pathway runs directly through the diagnosis record.
For a patient to be identified as medically frail, states need a mechanism to recognize that frailty. The mechanism CMS contemplates is ICD-10-based: states will maintain auditable code lists, run automated eligibility screenings against diagnosis data, and flag individuals who qualify for exemption. If the ICD-10 code isn’t in the encounter record, the automated system won’t see it — and the patient will be treated as work-capable even when they aren’t.
This is not a billing problem. It’s a patient-protection problem. A coder who fails to capture a qualifying chronic condition isn’t just leaving money on the table — they may be contributing to a coverage termination for someone who legally qualifies for an exemption.
The Nebraska Model
Nebraska is the furthest along in operationalizing this. The state already implemented Medicaid community engagement requirements ahead of the federal rule and developed a nearly 300-page index of ICD-10 diagnosis and procedure codes used to identify medically frail individuals. That document represents the most detailed publicly available picture of what “medical frailty” looks like in ICD-10 terms.
Other states will follow Nebraska’s lead, either by adopting a similar code list or by building their own variation. Providers working across multiple states should expect inconsistency in how medical frailty is coded and assessed. The BHB article from June 2 quoted Maeghan Gilmore of the Association for Behavioral Health and Wellness noting that state-by-state variance in definitions of SUD and “disabling mental disorder” is precisely what the industry was hoping to avoid. CMS allowed states significant implementation latitude, which means the ICD-10 code list that protects a patient in Nebraska may not be recognized in Texas.
Exemption Categories Requiring ICD-10 Documentation
Coding teams preparing for January 2027 should focus on the following exemption categories, each of which requires ICD-10 substantiation in the clinical record:
- Medical frailty and chronic conditions: High-risk diagnoses familiar from HCC coding — congestive heart failure (I50.x), COPD (J44.x), end-stage renal disease (N18.6), diabetes with complications (E11.x) — are expected to qualify. These are often captured for risk adjustment but not consistently documented per encounter.
- Serious mental illness: Diagnoses like schizophrenia spectrum disorders (F20.x), bipolar disorder (F31.x), and major depressive disorder with recurrent episodes (F33.x) will likely qualify. Documentation specificity matters — unspecified codes may not meet a state’s threshold.
- Substance use disorder in active treatment: SUD diagnoses (F10-F19) for patients actively enrolled in a treatment program are explicitly exempt under CMS-2454-IFC. Coders must capture both the SUD diagnosis and document the treatment context.
- Physical disability and functional limitation: Codes reflecting blindness (H54.x), paralysis (G81-G83), or other significant functional impairment may qualify as medical frailty depending on state implementation.
- Pregnancy: Pregnancy-related exemptions require current documentation of obstetric status (Z34.x, O codes) through the applicable coverage period.
What Documentation Standards Must Support
CMS expects states to maintain auditable ICD-10 code lists and to run ex parte — meaning automatic, without requiring patient action — eligibility reviews against diagnosis data before sending any termination notice. That audit trail depends on the quality and completeness of the clinical documentation that coders produce.
The documentation burden extends beyond the encounter. States will develop appeals procedures, and patients who lose coverage incorrectly will have the right to contest. In those appeals, the medical record is the primary evidence. A provider whose coding is inconsistent or incomplete faces a compliance and patient relations risk, not just a reimbursement risk.
CMS also left open significant ambiguity around how 42 CFR Part 2, the federal privacy rule governing SUD treatment records, interacts with exemption verification workflows. Providers treating SUD patients should watch for further guidance on what diagnosis data can be shared with Medicaid eligibility systems without additional patient consent.
Where Agentic Coding Tools Can Help
The challenge this rule creates is partly a workflow problem. Most EHR systems are not configured to flag medically frail patients for Medicaid work requirement exemption purposes. Standard coding workflows focus on DRG optimization, E/M level selection, and payer-specific billing rules — not on whether a documented condition qualifies a patient for a regulatory exemption in a state program that hasn’t yet published its code list.
Agentic AI coding tools that continuously analyze encounter documentation — identifying and suggesting ICD-10 codes in real time, with logic that can be updated as states publish their exemption code lists — can close this gap before January 2027. They can also flag inconsistencies across encounters: a patient with CHF documented in one visit but not captured in subsequent encounters is a compliance and coverage risk under this new framework. Systematic, per-encounter diagnosis capture is what this rule effectively requires.
The comment period closes July 31, 2026. If your organization has concerns about how the rule will interact with your patient population or coding infrastructure, this is the window to raise them. CMS may revise the interim final rule based on feedback before January 2027 implementation.
Coding teams that treat this as purely an eligibility issue will be caught off-guard when the first Medicaid work requirement audits arrive. The ICD-10 record is the foundation of the exemption system, and the coder is the one who builds it. Medikode’s automated medical coding platform helps practices capture the right diagnosis codes at the point of care — including the chronic and behavioral health conditions that Medicaid work requirement exemptions depend on. Start the conversation now, before January 2027 becomes a crisis.