For the First Time in Decades, Stimulants Kill More Americans Than Opioids. The Treatment System Has No Answer.

When JJ Keller read the texts on his phone — messages he didn’t remember sending — he checked himself into a hospital. His last dose of methamphetamine had sent him into a psychotic state. He was homeless. He had attempted suicide. The phone was, according to the account he later shared publicly with Stand Together, the thing that anchored him back to reality. He went on to build a life in recovery, teaching other people what he called “life on life’s terms.”

What that story doesn’t tell you — what almost no recovery narrative told at scale ever does — is what a physician could prescribe for his methamphetamine use disorder the morning he walked into treatment. Not buprenorphine, the medication that has saved hundreds of thousands of opioid-dependent people. Not methadone. Not naltrexone. Not anything. The FDA has never approved a medication for methamphetamine use disorder. After decades of research, billions of dollars in federal investment in addiction science, and a sea change in how this country thinks about opioid treatment, there is still no pill, no injection, no pharmaceutical answer for the drug killing more Americans than fentanyl now does.

That gap — between where the crisis has moved and where the treatment infrastructure is still pointing — is the story of American addiction policy in 2026.

The numbers shifted, and nobody told the treatment system

For most of the last fifteen years, the opioid epidemic defined the national conversation. Rightly so: synthetic opioids drove overdose death counts to 107,000 in 2023, a number so large it rewired federal policy, unlocked billions in Medicaid funding, and made buprenorphine access a bipartisan cause. The treatment system organized itself around opioids. Medication-Assisted Treatment became shorthand for buprenorphine and methadone. The harm-reduction infrastructure — naloxone distribution, fentanyl test strips, the whole apparatus — was designed first around opioid overdose.

Then the numbers moved.

Provisional CDC data cited in an NPR investigation published in April 2026 found that for the first time in decades, stimulant overdoses — primarily cocaine and methamphetamine — now kill more Americans than opioids do when counted separately. The shift is partly a success story: opioid-specific deaths have fallen sharply as naloxone saturated more communities and buprenorphine became more accessible. But the decline in opioid-only deaths hasn’t corresponded with any equivalent fall in stimulant deaths, which reached nearly 65,000 in the 12 months ending mid-2024, according to the CDC’s MMWR — including 29,449 cocaine-involved and 34,855 methamphetamine-involved deaths. Most co-involve fentanyl, which is now so pervasive in the supply that it’s nearly impossible to use any street drug without encountering it.

The practical implication is that a person with methamphetamine use disorder is now statistically more likely to die of their addiction than a person with opioid use disorder — and is guaranteed to encounter a treatment system that has far fewer tools for them.

The medication gap is not a gap. It’s an absence.

This is worth saying precisely: there is no FDA-approved medication for stimulant use disorder. This is not a case of a drug that exists but isn’t covered, or one that’s effective but restricted. The drug doesn’t exist. Decades of clinical research into cocaine and methamphetamine pharmacology have failed to produce an approved treatment.

What researchers do have is evidence — a growing body of it — that the combination of extended-release naltrexone and oral bupropion reduces methamphetamine use and is tolerated reasonably well. The MOTIVE trial published in NEJM in 2021 showed it. Follow-up research has confirmed the signal. But this combination is not approved for meth. Physicians who prescribe it do so off-label, fighting insurance denials and prior-authorization cliffs every step of the way. The FDA’s standard approval pathway requires substantial evidence of efficacy and safety. The evidence exists — marginally. The political will and pharmaceutical investment to push it through approval have not materialized, because there is no blockbuster drug to patent.

The FDA’s standard approval pathway requires substantial evidence of efficacy and safety.

For cocaine, the picture is even bleaker. Cocaine use disorder has no off-label pharmacological backstop comparable to naltrexone/bupropion. Researchers have studied dozens of candidates over forty years — modafinil, disulfiram, topiramate, tiagabine, N-acetylcysteine, propranolol. None crossed the finish line in Phase 3 trials. Cocaine remains, in the clinical sense, medication-resistant.

Contingency management: the treatment that works, the funding that never came

Into this vacuum steps the one behavioral intervention with a genuine evidence base for stimulant use disorder: contingency management. The concept is not complicated. Patients receive small rewards — vouchers, gift cards, small cash prizes — for verified abstinence, typically confirmed through drug testing. The evidence for CM reducing methamphetamine use is among the strongest in addiction medicine. It works for cocaine too. The effect sizes in randomized trials beat most pharmaceutical treatments for other conditions.

And yet, for decades, federal policy effectively banned it.

The problem was a federal rule — not an addiction policy, actually, but an anti-kickback statute in an unrelated law — that capped the value of incentives Medicaid providers could offer patients at $75 per year. You cannot run a meaningful contingency management program on $75 a year. The result was a treatment that everyone knew worked but that federal funding rules made functionally impossible to offer at scale.

In January 2025, SAMHSA finally raised the cap, increasing the allowable incentive to $750 per patient per year. That was a genuine win. But it didn’t resolve the underlying coverage question: $750 is still low enough to constrain program design, and more importantly, most states haven’t approved CM for Medicaid reimbursement at all. As of mid-2026, only five states — California, Delaware, Hawaii, Montana, and Washington — have used Section 1115 Medicaid waivers to fund contingency management for stimulant use disorder. The remaining 45 states, including Arizona, have not.

Arizona’s case is instructive. The state’s annual substance use treatment report filed with Governor Katie Hobbs in April 2026 documents 29,707 active methamphetamine and stimulant cases served through AHCCCS-funded treatment in state fiscal year 2025 — making meth the second-most-common primary substance in Arizona’s treatment population, behind only alcohol. Nearly 30,000 people. Not one of them is receiving contingency management through Medicaid, because AHCCCS has not filed the waiver that would allow it.

The five states that did apply for the waiver moved first. Everyone else is still deciding.

The thing people keep not saying about GLP-1s

There is, possibly, a new pharmacological answer coming — but it isn’t here yet, and the way the field is talking about it risks overclaiming.

A study published June 3, 2026 by researchers analyzing data from more than 600,000 U.S. veterans found that GLP-1 receptor agonists — the class of drugs that includes semaglutide, sold as Ozempic and Wegovy — were associated with a 14-20% reduction in substance use disorder risk across alcohol, cannabis, cocaine, opioids, and nicotine. For people already struggling with addiction, GLP-1 users had fewer overdoses, fewer hospitalizations, and fewer drug-related deaths.

That’s a remarkable signal. And it fits with a mounting body of animal and mechanistic research showing that GLP-1 receptors in the brain’s reward circuitry appear to dampen dopamine responses to addictive stimuli — a potential common pathway underlying multiple SUDs.

But a few things are also true. This is an observational study of veterans, not a randomized trial for stimulant use disorder. The SHIFT study — a randomized controlled trial specifically testing semaglutide for methamphetamine use disorder — only started enrolling in May 2026 and won’t produce results before early 2027. A published RCT protocol for cocaine use disorder (Yamamine and Schmitz, University of Texas, 2026) exists, but the trial itself is running. The Lancet RCT that published in May 2026 — showing semaglutide significantly reduced heavy drinking in people with alcohol use disorder — is compelling evidence that these drugs work for at least one SUD. But it was a 26-week trial in 108 participants for alcohol, not methamphetamine.

A published RCT protocol for cocaine use disorder (Yamamine and Schmitz, University of Texas, 2026) exists, but the trial itself is running.

The honest position is: GLP-1s are the most promising lead in stimulant pharmacotherapy in a generation. The evidence is not yet sufficient to prescribe them for that purpose, and clinicians who are doing so are getting ahead of the trial results. The people who will decide whether this pans out are the trial investigators, not the venture capitalists.

The question that matters right now

The problem with “wait for the trials” is that the trials take time that people in active methamphetamine use don’t have.

There are approximately 30,000 people in Arizona’s public behavioral health system right now being treated for stimulant use disorder. They have access to residential programs, outpatient counseling, group therapy. What most of them don’t have access to is contingency management through their Medicaid coverage — because their state hasn’t filed the paperwork — or any medication specifically approved for what they’re going through.

The five states that do offer Medicaid-covered CM are doing something that works. They took the waiver application, filed it, and got it approved. Arizona has not. That is a decision with a body count.

What shifts this is usually not science — the science on CM has been settled for years — but political and administrative will. Someone at AHCCCS has to decide the waiver is worth pursuing. Someone at the legislature has to stop treating contingency management as a controversial “cash for drugs” scheme — the absurd mischaracterization that held up the federal cap for a decade — and recognize it as the evidence-based practice it is.

JJ Keller found his way to recovery on the strength of something that no treatment protocol gave him: a moment of self-recognition, a phone screen, the sheer will to check himself in. He has, by any measure, done better than the system around him had any right to expect. But recovery built on personal resilience alone is not a health policy. It’s a bet against the odds, and the house usually wins.

The drugs that are killing Americans have changed. The treatment system needs to catch up.

---

The Rize Newsroom covers the stimulant treatment gap as part of our ongoing Treatment & Recovery series and our Stimulants substance pillar. For people seeking treatment today, find facilities near you.