CMS Rewrote Who Counts as Medically Frail. Getting Better Now Counts Against You.

Theresa spent six years sober in Alcoholics Anonymous and, by her own account, was miserable for most of it. It wasn’t until she found a harm-reduction taper program that she got herself off a daily half-gallon of vodka and onto a maintenance dose of naltrexone — a medication she still takes every day, more than three years later. “I’m not powerless,” she told Filter. “I’m in control.” That ongoing, unglamorous management of a substance use disorder — the daily pill, the check-ins, the years of not being finished — is exactly what federal law has always recognized as a reason someone might need Medicaid to stay flexible with them.

A rule the Centers for Medicare & Medicaid Services issued on June 1 says that isn’t necessarily true anymore. Buried in the interim final rule implementing the new 80-hour-a-month Medicaid work requirement is a sentence that reads, in effect, like an actuarial judgment on recovery itself: people with a substance use disorder generally qualify as “medically frail” and exempt from the work mandate — unless they’ve been in active recovery for five years or more, in which case a caseworker can decide they don’t.

The rule treats five years of staying well not as proof recovery works, but as proof you no longer need protecting.

That’s backwards, and it’s about to become federal law for 43 states and the District of Columbia, effective July 31 — the same date public comments on the rule are due. States have until January 1, 2027 to build the system that decides who still counts.

The rule doesn’t ask if you’re using. It asks how long you’ve stopped.

The mechanics matter, because they’re stranger than “cut” or “kept.” Congress’s own language, tucked into last year’s budget reconciliation law, lists five categories of people automatically exempt from proving 80 hours a month of work, school, or community service to keep Medicaid: people who are blind or disabled, people with a physical or developmental disability limiting daily activities, people with a “serious or complex” medical condition, people with a disabling mental disorder — and people with a substance use disorder. Congress did not attach a duration limit to any of those five categories. CMS added one, for exactly one of them.

Under the interim final rule, a caseworker reviewing someone’s SUD exemption is now instructed to distinguish between people in “active treatment,” people in “early or sustained recovery,” and people in what the agency calls “stable recovery” — defined as five years or more. The first two groups keep the presumptive exemption. The third does not. They can still apply for an exemption, but only by clearing a second, CMS-invented bar that isn’t in the statute at all: proving their condition “significantly impairs” their ability to comply with the work requirement, as the agency’s own fact sheet describes it. Blindness doesn’t come with a sunset clause. Diabetes doesn’t come with a sunset clause. Substance use disorder, alone among the five statutory categories, now does.

Anthony Wright, executive director of the patient advocacy group Families USA, put the practical effect plainly: “This guidance significantly raises the barrier for demonstrating medical frailty, meaning many patients in the middle of treatment will have the new hassle of proving their condition, over and over, with any mistake or gap being penalized by the loss of their health care and coverage.” Sen. Ron Wyden of Oregon called the broader rule “a grim step in America’s march towards a health care system that further restricts access to health care.” CMS Administrator Dr. Mehmet Oz has defended the agency’s posture toward self-attestation in blunter terms: “We are serious about the consequences of dishonesty.”

Nobody at CMS has explained, in the rule or anywhere else, why five years was the number, or why recovery is the one condition on the list that the agency decided could expire.

Twenty-five states, including Arizona, are already suing to stop it

On June 29, twenty-five states and the District of Columbia — Arizona among them — filed suit in the U.S. District Court for the District of Massachusetts, Commonwealth of Massachusetts et al. v. Oz et al. (No. 1:26-cv-12962, before Judge Richard G. Stearns), seeking to have the court throw out the specific regulatory provision — 42 C.F.R. § 435.554(c)(5)(i) — that lets CMS demand proof of “significant impairment” on top of Congress’s five statutory categories. The states argue the addition is “contrary to law and arbitrary and capricious” under the Administrative Procedure Act, because Congress never wrote a functional-impairment test into the exemption in the first place — CMS added it during rulemaking. They’ve also filed an emergency motion for a preliminary injunction, trying to get relief before the January 1 implementation deadline arrives.

On June 29, twenty-five states and the District of Columbia — Arizona among them — filed suit in the U.S.

It is that same “significantly impairs” language — the piece the states are suing over — that CMS is using to carve five-year recovery out of the SUD exemption. Kill the provision in court, and the recovery carve-out likely goes with it. Seventeen expansion states, all with Republican governors, chose not to join the suit. Arizona, which expanded Medicaid under the ACA and where AHCCCS covers a meaningful share of the state’s substance use disorder treatment population, is one of the twenty-six plaintiffs betting that a federal judge in Boston will decide this before it becomes state agencies’ problem on New Year’s Day.

The KFF analysis of the rule points to the state that already ran this experiment: New Hampshire’s earlier, since-abandoned work-requirement program required enrollees to get a doctor to certify they were “unable to work” before an exemption would be granted — a standard so far removed from the actual statutory language that “enrollees with physical and behavioral health problems struggled when applying for exemptions,” according to KFF’s review. CMS is now asking 43 states to build a national version of the same system, for the same population, on a seven-month clock.

We have watched a federal government decide that a population which meets a written legal standard should have to re-prove it anyway, and watched people fall through that gap, before. In 1988, Congress banned the use of federal funds for the syringe exchanges that were, at the time, the only intervention slowing the spread of HIV among people who inject drugs — not because the science was unclear, but because lawmakers worried support looked like endorsement. The ban held for 21 years before Congress finally lifted it in 2009 — then reinstated it three years later, and didn’t settle into a lasting partial repeal until 2016. The mechanism is different this time — an administrative exemption test instead of a funding ban — but the instinct is the same: treat evidence that a harm-reduction or recovery-support intervention is working as a reason to make it harder to reach, not a reason to protect it.

The five-year line assumes recovery is a light switch. The research says it’s closer to weather.

If you’re five years out and proud of it, you should be — and you also already know that pride isn’t the same thing as being finished. CMS’s rule treats “stable recovery” as a status you graduate into and then hold, permanently, the way you hold a diploma. The clinical research on relapse doesn’t describe it that way. Harvard’s John F. Kelly, whose recovery-monitoring work is among the most cited in the field, has tracked people well past the five-year mark and found that relapse after long stretches of stability is rarely sudden. As Dr. Mark Gold summarized Kelly’s findings in Psychology Today, people who relapsed after an average of 3.6 years of sobriety showed warning signs for months beforehand — more than 80% had already shown “decreased attention to recovery activities” before anything visible happened, alongside building depression, anxiety, isolation, and disrupted sleep. Long-term recovery isn’t a wall you get to stand behind once you’ve built it high enough. It’s a thing you keep tending, the same way you’d keep taking a blood pressure medication that’s working. Nobody tells a patient whose blood pressure has been controlled for five years that they no longer need the prescription, or the doctor.

That’s the deeper problem with a bright line at five years: it isn’t a clinical finding, it’s a bureaucratic convenience, applied to the one chronic condition on CMS’s list still tangled up in a century of moral judgment about who deserves the benefit of the doubt. The same federal government that put out a formal request for comment in June asking whether addiction should be classified as a chronic disease is, simultaneously, writing a rule that treats a chronic-disease diagnosis as something that expires on a timer no other chronic disease gets. You can’t credibly ask the country to debate whether addiction is a lifelong condition in one docket while telling Medicaid caseworkers in another that five years means it’s over.

What actually changes on July 31 — and what doesn’t

Arizona’s own numbers make clear why the state signed onto this suit instead of sitting it out — this is a state where overdose deaths have been moving in the opposite direction of the rest of the country even as the national count falls, which is not the moment to make Medicaid harder to hold onto for people who got well. For the mechanics of the broader Medicaid and behavioral-health funding landscape this rule sits inside, see our ongoing coverage.

If you are currently in treatment, or in the first five years after developing a substance use disorder, this rule does not touch your exemption. That protection is still there, still automatic, in every one of the 43 affected states plus D.C. What changes is narrower and still serious: if you’ve been in stable recovery for half a decade or longer, you now have to clear a second, non-statutory hurdle — proving “significant impairment” — to keep the same exemption Congress already gave you. If you’re a case manager or facility admin with clients or patients approaching that five-year mark on Medicaid, this is the documentation conversation to start now, not in December: get a treating provider’s letter on file describing ongoing functional impact before a state’s verification vendor comes looking for one on a deadline. States have to build the verification systems to make that judgment by January 1, 2027, with essentially no federal funding attached to doing it, which is its own quiet cruelty: cash-strapped state Medicaid agencies now have to spend money they didn’t budget for to decide who’s recovered enough to lose protection.

That protection is still there, still automatic, in every one of the 43 affected states plus D.C.

None of this is settled. The comment period on the rule runs through July 31 — the same day it’s slated to take effect — and a federal judge in Massachusetts could vacate the underlying provision before a single state finishes building its verification system. Twenty-six plaintiffs are betting he will. In the meantime, naloxone access, buprenorphine telehealth prescribing, and Medicaid’s basic behavioral health benefit are untouched by any of this — this fight is over one exemption test, not over whether Medicaid covers treatment at all. That’s a door that’s still open tonight, even for the people this rule is aimed at.

Five years from now, whoever is counting will have to decide whether “stable recovery” was ever the kind of thing that should have had an expiration date on it in the first place. Theresa, three years into her own maintenance dose, isn’t waiting for that answer. She’s just still taking the pill.