Money, explained for the rest of us.

Get our free daily email →

Medicaid will use a diagnosis code, not a doctor’s note, to decide who is too sick to work

By

Image Credit: Saeed Ghezelbash - CC BY 4.0/Wiki Commons

Starting no later than January 1, 2027, states running Medicaid’s new work requirement will decide who is too sick to comply largely by matching a diagnosis code already sitting in a claims file, not by waiting on a letter from a treating doctor. The code comes out of the 2025 Working Families Tax Cut law, which conditioned Medicaid eligibility for certain adults on 80 hours a month of work, training, education, or community service and, in the same provision, excused anyone found to be “medically frail.” Federal regulators spent the summer turning that single word into a formal rule, then used a September briefing to states to name the exact diagnosis codes that settle the question without any paperwork changing hands.

A Rule Already in Force, Not a Proposal

The medically frail exclusion is not sitting in draft form waiting for a future vote. The Centers for Medicare & Medicaid Services wrote it into an interim final rule with a comment period, docket CMS-2454-IFC, which carries the legal weight of a final regulation from the day it takes effect even while the agency keeps collecting public comments on it.

That rule, formally titled the Medicaid community engagement requirement regulation, was published in the Federal Register on June 3, 2026, took effect July 31, 2026, and was followed by a technical correction on June 29, 2026 that left its substance intact. It remains the operative federal standard today. The underlying statute behind it, Section 71119 of Public Law 119-21, was signed on July 4, 2025 and sets the 80-hour monthly community-engagement threshold along with a January 1, 2027 nationwide implementation deadline, though a state may move earlier. Nebraska already has, according to CMS’s own announcement of the rule’s rollout.

States are not being asked to build this system on their own. Federal guidance tied to the same statute makes $200 million in grants available for the 2026 fiscal year to help state Medicaid agencies build the eligibility systems this rule requires, on top of enhanced federal matching funds states can claim for the technology work itself. That money is a signal of how much administrative machinery a single word, “frail,” now has to run through before a Medicaid case is settled either way.

None of this is abstract for the people it touches. A caseworker cannot simply take someone’s word that a health condition makes working 80 hours a month unrealistic; the state has to be able to point to a code, a claim, or documentation that satisfies the federal definition, and it has to be ready to defend that determination if it is ever audited.


Free download: Every document a Medicaid renewal may require, a deadline and proof tracker, and what to do after a termination notice. Get the free Medicaid renewal checklist.

Why One Diagnosis Code Can Close the File

CMS’s September 8, 2026 guidance to states lays out an example framework for sorting medically frail cases using data states already hold, built around three tiers. In the first tier, certain ICD-10 diagnosis codes recorded on a medical claim or managed-care encounter in the prior 12 months are enough, by themselves, to confirm someone is exempt from the work requirement, with no separate note or certification from a physician required.

CMS’s implementation guidance names several codes that qualify on their own: C25.0, malignant neoplasm of the head of the pancreas; G12.21, amyotrophic lateral sclerosis; N18.6, end-stage renal disease; and B22.0, HIV disease resulting in encephalopathy. The list is described as non-exhaustive, and states are expected to build their own broader code sets, but every state list has to stay “auditable” and “justifiable” against the regulatory definition of medical frailty and cannot expand the five categories the underlying statute allows.

The Five Categories a Diagnosis Code Has to Fit

A code cannot simply be sympathetic; it has to map onto one of five statutory grounds for exclusion. Those are: being blind or disabled under the Social Security Act’s own disability definition; having a substance use disorder that is not in stable, five-year-plus recovery; having a disabling mental disorder; having a physical, intellectual, or developmental disability that significantly limits one or more daily activities; or having what CMS defines as a serious or complex medical condition, a standard borrowed from an Institute of Medicine report that covers conditions that are life-threatening, severely disabling, or require major, ongoing caregiver effort.

Within that fifth category especially, CMS’s own worked examples show the outcome hinges on the specifics of the code and the surrounding data, not the diagnosis name alone. A recent hospitalization for decompensated heart failure lands someone in the automatically-qualifying tier; a diagnosis of anxiety managed entirely through a low-dose prescription and routine primary care visits does not, absent something more severe in the record.

How Often the Code Itself Gets Rechecked

Not every record is this tidy. CMS’s framework describes a second tier, where a qualifying condition shows up in the data but has to be paired with other coded information, such as pharmacy claims or durable medical equipment codes, before the exclusion is confirmed, and a thinner third tier for the small share of cases where the state’s data does not yet show enough to decide. Even there, CMS’s stated design goal is a data-first system: settle as many cases as possible on coded claims and encounter data alone, and treat anything beyond that as the exception rather than the everyday process.

Verification does not happen just once. States must attempt to confirm medical frailty using claims and encounter data from the preceding 12 months, or the preceding five years for a substance-use-disorder determination, and must reverify every enrollee’s status at least once every 12 months. Before January 1, 2028, a state may accept a signed statement made under penalty of perjury each time it verifies someone’s status if the data trail is thin; after that date, a state may only accept such a statement once per continuous period of enrollment, and every verification after that has to run back through the reliable data the state already has on file.


The Paperwork Trail Now Falls on the Household

The frailty rule above shows how much of the new Medicaid work requirement rides on data the state already has, but the same overhaul is tightening renewal schedules and document requests for everyone else on Medicaid and SNAP, frail exclusion or not. Few households keep a running record of which forms a caseworker already has on file versus which ones still need to be resent, and a missed request is often the difference between staying enrolled and losing coverage before an appeal can even begin.

The SNAP & Medicaid Renewal Organizer combines a 13-page organizer and 51 state packs into one reference a household can keep next to its own Medicaid file.

Look up the renewal document checklist in The SNAP & Medicaid Renewal Organizer.

AI-assisted research and drafting; reviewed by an editor.


Spotted an error? Tell us at [email protected]. We fix mistakes fast and in the open — see how we work on our standards page.

Get the money news that affects your wallet — free, every weekday morning.