Stimulant Trials Have Been Grading Every Medication Pass/Fail. A New Meta-Analysis Says That's Why None of Them Passed.

There is no FDA-approved medication for cocaine or methamphetamine use disorder. Not one. Buprenorphine and methadone exist for opioids; naltrexone and acamprosate exist for alcohol. For the roughly 1.8 million Americans with cocaine use disorder and 2.5 million with methamphetamine use disorder, the medicine cabinet is empty — not because nobody’s tried, but because trial after trial of promising candidates has come back reading as a failure. A meta-analysis published in JAMA Psychiatry on June 3 makes a case that a lot of those “failures” weren’t failures of the drugs. They were failures of the ruler used to measure them.

Stimulant trials have spent two decades asking a yes/no question — did you stop completely — when the honest question was always going to be: did you use less.

The meta-analysis, drawing on 12 randomized controlled trials and roughly 2,000 adults enrolled in cocaine- or methamphetamine-use-disorder pharmacotherapy studies, tested nine medications — topiramate, bupropion, modafinil, ondansetron, tiagabine, cabergoline, reserpine, selegiline, and baclofen — against two different scorecards. The first was the field’s long-standing default: abstinence, meaning zero use, verified by urine screens, for a sustained stretch near the end of the trial. The second was a looser but arguably more honest bar: a meaningful reduction in use from where the person started, whether or not they reached zero.

By the abstinence scorecard, the story was bleak and familiar: 13.3% of participants across all trials reached abstinence, and no single medication beat placebo by a statistically significant margin. That is the number that has justified, trial after trial, headlines like “another stimulant medication fails.” By the reduced-use scorecard, the picture changed: 31.2% of participants meaningfully cut back, and one medication — cabergoline, a dopamine agonist most familiar as a treatment for prolactin disorders — separated from placebo with a real, measurable effect. Participants on cabergoline hit a predicted reduced-use rate of 44.2%, against 27.6% on placebo. That’s not a rounding difference. It’s a small-to-moderate treatment effect that the abstinence-only lens had been erasing for years.

Why “did you stop” is the wrong question for a dopamine-driven disorder

Here’s the mechanism in plain terms, because the reason this matters isn’t just statistics — it’s how stimulant dependence actually works in the brain. Cocaine and methamphetamine flood the brain’s dopamine system far more forcefully than opioids flood the endorphin system, and the crash that follows — the anhedonia, the flatness, the sense that nothing feels good without the drug — is part of what keeps people returning to use even when they’re actively trying to cut down. Expecting a single medication to take a person from that state to zero use in a matter of weeks was always a higher bar than the neuroscience supports. Reduction — using less often, using smaller amounts, going longer between episodes — is a real, clinically meaningful step toward the dopamine system recalibrating, even if it isn’t the finish line. Treat every trial that produces “less” instead of “none” as a failure, and you throw away exactly the signal you’re looking for.

This is also, not incidentally, closer to how clinicians already talk to patients who use stimulants — in reductions, in fewer bad weeks, in a longer stretch between relapses — than the all-or-nothing framing that has governed FDA trial design. A 2024 companion analysis from Johns Hopkins reached a parallel conclusion using a separate set of 13 multisite trials: reduced use predicted the same downstream benefits — fewer overdose events, better retention in care, improved functioning — that abstinence was assumed to be a proxy for. If reduced use gets you most of the real-world benefit that abstinence gets you, and if measuring for abstinence-only is hiding medications that work, the trial design itself has been the obstacle.

None of this means cabergoline is ready for your formulary. The authors are explicit that the finding needs confirmation in a dedicated, adequately powered trial designed around reduced use as a primary endpoint rather than uncovered secondarily in re-analysis — and dopamine agonists carry their own monitoring burden, including a rare but serious risk of cardiac valve changes at the doses used for other conditions, that any stimulant-use-disorder trial would need to track closely. What the meta-analysis does support is narrower and still significant: the absence of an FDA-approved stimulant medication is not proof that pharmacotherapy for stimulant use disorder doesn’t work. It may be proof that the industry has been running trials against a scorecard that made it nearly impossible for a real, partial effect to count as a win.

What the meta-analysis does support is narrower and still significant: the absence of an FDA-approved stimulant medication is not proof that pharmacotherapy for stimulant use disorder doesn’t work.

If you are one of the people this trial design failed — if you cut back and still felt like you’d failed because you didn’t reach zero — the data says otherwise. Less is not nothing. For a field that has spent twenty years telling people who reduce their use that it doesn’t count, that’s the finding worth sitting with.