CMS Wrote a Medicaid Rule That Punishes People for Staying in Recovery

On June 18, a Georgetown University research professor named Leonardo Cuello wrote something unusually blunt for a policy memo: the agency behind a federal rule posted seventeen days earlier was “confusing its Constitutional role — to implement the law — with Congress’s exclusive role in writing new laws.” Not a minor technicality. He meant the rule that now governs whether people with substance use disorders keep their Medicaid in 43 states and Washington, D.C. — including Arizona, where the compliance clock now reads under five months.

Here’s the plain version. Congress, in the 2025 law creating Medicaid work requirements, wrote an exemption into the statute itself: people with a substance use disorder don’t have to prove they’re working to keep coverage. No asterisk, no severity test. The Centers for Medicare & Medicaid Services (CMS) was supposed to translate that into instructions states could follow. Instead, in an interim final rule — a regulation that takes legal effect on its own timeline, ahead of and separate from the usual public comment-and-revise process, issued June 1 and effective July 31 — CMS added a test Congress never wrote: your condition must also “significantly impair” your ability to comply.

CMS added a legal standard Congress never wrote, and used it to write people in stable, long-term recovery out of the one exemption meant to protect them.

That sentence is not editorializing. It is close to a direct paraphrase of what happens next in the rule’s own text: anyone in “active recovery” from a substance use disorder for five years or more no longer automatically qualifies as what CMS calls “medically frail” — the legal term, largely invisible to the people it governs, for someone excused from proving 80 hours a month of work, school, job training, or volunteering (what the rule calls the “community engagement requirement”) to keep their health coverage. Five years clean, on this rule’s logic, is evidence you might not need the exemption anymore. It is one of the stranger inversions in recent health policy: the longer and more successfully you’ve stayed in recovery, the less the federal government is willing to assume you’re still vulnerable.

The standard Congress didn’t write

Start with the mechanics, because they matter for anyone trying to figure out if this touches them. The underlying law — the reconciliation package widely known as H.R. 1 — told states they could exempt five categories of people from the new work requirement without a fight: people who are blind or have a disability, people whose disability limits daily activities, people with a substance use disorder, people with a “disabling” mental health condition, and people with a “serious or complex” medical condition. Read the statute and the substance use disorder exemption is unconditional. It doesn’t say “unless you’re doing well.” It doesn’t mention how long you’ve been in recovery. It just says: if you have a substance use disorder, you’re exempt.

CMS’s rule changes that in practice, even while leaving the statutory language technically intact. The interim final rule ties every one of those five categories to a functional test — whether the condition “significantly impairs” a person’s ability to meet the work requirement — a standard KFF’s Amaya Diana, Jennifer Tolbert, and Robin Rudowitz describe as “a restrictive definition of medical frailty that differs from states’ early expectations and ties medical frailty to an individual’s ability to meet the community engagement requirements.” States can no longer simply exempt someone because they carry a diagnosis. A caseworker — or, starting in 2028, a treating provider filling out documentation — now has to make a separate judgment call about whether that diagnosis is disabling enough, on top of confirming it exists.

Cuello’s argument, laid out across two Georgetown Center for Children and Families analyses this summer, is that CMS doesn’t have the authority to make that call. “There can only be one reasonable or serious interpretation of Congress’s legislation here,” he wrote, arguing the statute’s exemption categories were meant to function as bright lines, not entry points into a second, unwritten eligibility test. “CMS’s job is to implement the statute, and CMS just ignored it.” It’s worth sitting with what that accusation actually is: not that the rule is unwise, but that an executive agency added a threshold to a law that a co-equal branch of government did not include, and is now using that invented threshold to decide who keeps health insurance.

Cuello’s argument, laid out across two Georgetown Center for Children and Families analyses this summer, is that CMS doesn’t have the authority to make that call.

Getting better is the disqualifying event

The five-year cutoff is where the rule stops being abstract and starts being personal for a specific, identifiable group of people: those who did the hardest part of recovery already, and did it well enough to sustain it. Deborah Steinberg, senior health policy attorney at the Legal Action Center, laid out the clinical case against the cutoff in a Georgetown-published analysis. She noted that the research CMS leans on to justify treating long-term recovery as lower-risk is thin and dated — drawn largely from studies conducted between 1989 and 2007, mostly on alcohol use disorder, mostly built around abstinence-only definitions of recovery that don’t reflect how clinicians think about substance use disorders today. Her point, stated as plainly as a policy attorney gets to state anything: for someone in stable recovery of five-plus years, “the risk of SUD recurrence… is approximately the same as the general population.” CMS’s rule doesn’t treat it that way. It treats the five-year mark as a graduation date from protection, not a milestone worth protecting.

If you’re reading this with your own sobriety date in mind, do the arithmetic CMS is effectively asking states to do. Somewhere around your fifth year clean — the year the panic about triggers started to loosen, the year you maybe stopped white-knuckling every family holiday, the year recovery started to feel less like a daily emergency and more like a life — is the year this rule decides you no longer automatically need Medicaid’s protection from an 80-hour-a-month reporting requirement. The medication that keeps you stable, the therapist you see every other week, the primary care doctor who caught your blood pressure problem early because you actually had insurance to go in — none of that stops being necessary at year five. The rule just stops assuming it’s necessary.

There’s a second, quieter problem buried in the same provision: who gets to decide whether a condition “significantly impairs” someone’s ability to work. Steinberg’s analysis flags that many providers are being asked to make a determination outside their clinical training — “it is not within their scope of practice or licensure/certification to determine whether someone’s condition significantly impairs their ability to meet the work reporting requirements.” A psychiatrist can diagnose a substance use disorder. Whether that diagnosis clears a federal work-requirement threshold is a different, essentially bureaucratic question that CMS is now asking clinicians to answer on a form, under audit risk, for patients they may see for twenty minutes a quarter.

A trade group’s victory lap, and the text underneath it

Not everyone in the behavioral health world read the rule as an attack. The day after CMS posted it, Debbie Witchey, president and CEO of the Association for Behavioral Health and Wellness — a trade group whose managed-care members cover roughly 200 million people — said CMS had “listened to the behavioral health community.” ABHW’s read is that the rule adopted real wins the group had lobbied for: exemption criteria anchored to recognized clinical standards (DSM-5, ICD-10, and ISMICC criteria, rather than a caseworker’s improvised judgment), limited self-attestation so people don’t need paperwork from day one, and the option to use standardized screening tools instead of open-ended clinical review.

Those wins are real, and worth naming honestly rather than folding into a simpler story where the rule is uniformly hostile. But they coexist with, rather than cancel out, the “significantly impair” standard and the five-year recovery cutoff — and KFF’s Tolbert pushed back on the celebratory framing directly, warning that the rule’s more restrictive approach “will be challenging for states to implement and could lead to more people falling through the cracks and losing coverage.” Winning better documentation rules while losing the scope of who qualifies in the first place is not obviously a net win for the people the exemption exists to protect.

Those wins are real, and worth naming honestly rather than folding into a simpler story where the rule is uniformly hostile.

The numbers around this are not small. The Congressional Budget Office estimates that roughly 18.5 million people nationally will be subject to the new work requirement, and that about 5.2 million will lose Medicaid coverage specifically because of it — a figure CBO reaches while also concluding, in Tolbert’s paraphrase, that “the new requirements will not meaningfully increase employment by Medicaid enrollees.” That’s the trade at the center of this policy in one sentence: a work requirement that CBO doesn’t expect to put meaningfully more people to work, layered with an exemption standard narrow enough to strip coverage from people who are already stable, working, or both.

We’ve been here before

This isn’t the first time Medicaid eligibility rules have quietly decided who gets addiction treatment and who doesn’t. In 1965, when Medicaid was created, Congress carved out what’s still called the IMD exclusion — a ban on federal Medicaid dollars paying for care inside an “institution for mental disease,” any facility over 16 beds primarily treating mental illness, a category that swallowed most residential addiction treatment. The goal was reasonable: discourage warehousing people in old asylums. The effect, for decades, was not. States could get federal help paying for a three-day detox or an ER visit, but not the 30- or 90-day residential program that might have kept someone out of the ER in the first place. A rule meant to modernize care ended up routing people toward crisis and away from stabilization — a pattern states are still unwinding through waivers Congress didn’t authorize until 2018.

The CMS work-requirement rule isn’t the IMD exclusion. But it rhymes. Both are cases where an eligibility rule, defended on paper as reasonable, ends up sorting people with substance use disorders away from the coverage that keeps them stable and toward the systems that catch people once they’re already in crisis.

What Arizona is actually staring at

Arizona is one of the 43 states this rule applies to, and AHCCCS — Arizona’s Medicaid program — expanded coverage under the Affordable Care Act specifically to reach the low-income adults this new work requirement now targets. Arizona’s broader behavioral health strategy, including how the state is deploying its opioid settlement dollars, assumes Medicaid coverage holds steady for people in treatment and in recovery; that’s the funding floor the rest of the system is built on. A rule that narrows who counts as exempt doesn’t just threaten individual coverage — it threatens the assumption every case manager, every facility billing office, and every discharge planner in the state is currently building around. And Arizona isn’t getting extra runway: the same January 1, 2027, deadline applies here as everywhere else, which means the state’s implementation decisions — how it defines “significantly impair” in practice, how much self-attestation it allows before demanding documentation — are being made right now, this summer, largely outside public view.

Behind all of this sits a number that should reframe the entire debate: according to SAMHSA’s 2025 National Survey on Drug Use and Health, released July 27, only 16% of the estimated 47.2 million people who needed substance use treatment in the past year actually received any — 7.6 million people, out of a need nearly six times that size. The safety net this rule is narrowing was already catching a small fraction of the people who needed it. Making the medically frail exemption harder to qualify for doesn’t fix that gap. It makes it official policy.

If you’re five years into recovery and reading this tonight, here’s the concrete thing worth doing before anything else: find out, specifically, whether your state — Arizona included — is treating self-attestation as sufficient through 2027, or already asking for provider documentation. That answer determines whether this rule reaches you in five months or later, and it’s answerable now, before your next renewal notice arrives instead of a warning.

That answer determines whether this rule reaches you in five months or later, and it’s answerable now, before your next renewal notice arrives instead of a warning.

Cuello ended his June memo with a line that reads less like a legal conclusion and more like a warning about what happens next: CMS, he wrote, is confusing implementing the law with writing it. Somewhere in the space between those two things is a person who spent five years building a life stable enough that a federal agency now considers it evidence they don’t need protecting. The statute Congress passed didn’t ask them to prove that stability was fragile. The rule CMS wrote does.