High-Dose Baclofen for Alcohol Use Disorder: Off-Label Use When First-Line Options Fail
A patient on stable long-term opioid therapy for chronic pain has already tried and lost access to both standard AUD medications. Baclofen’s French trial record is genuinely split — and his own regimen raises a risk neither trial was designed to test.
The day his pain physician explained that naltrexone would send his body into acute withdrawal against the morphine he takes every day was the first time anyone had told T.O., in plain terms, why his drinking and his pain couldn't be handled as two separate problems. He still remembers it clearly, months later, the way people remember being told a diagnosis rather than a plan. He is fifty-eight, and a warehouse accident crushed two lumbar vertebrae ten years ago; he has been on a stable dose of extended-release morphine since, managed the whole time by the same pain clinic, with no history of dose escalation or aberrant use. He still tries to get out to the lake most Saturday mornings, a habit from before the injury that now costs him more than the fishing gives back, and his wife rides along less for the fishing than to keep him company and, he suspects, to keep an eye on him, since his drinking has become something they talk about openly now rather than something skirted around.
His alcohol use disorder developed gradually over the pain years — roughly six to eight drinks most evenings for the past five — and he has tried twice before: acamprosate for four months last year, taken exactly as prescribed, with no reduction in his drinking either he or his wife could point to; and naltrexone, which never actually started, for the reason he now explains almost verbatim the way his pain doctor first laid it out to him. He is otherwise healthy — no liver disease, no kidney disease, no psychiatric history beyond the drinking — and he read about the French experience with high-dose baclofen online before this visit, arriving with a specific question rather than a vague one: why isn't this the obvious next option, given that the other two didn't work and the third was never available to him at all?
The drug that worked in France, on a patient already taking an opioid
He's not asking for baclofen instead of trying — he's asking for it because he genuinely tried and the standard path closed on him twice, once by non-response and once by a real contraindication. Bacloville, the largest and most methodologically careful trial we have, used individualized titration up to 300 mg a day across sixty-two French centers and found significantly more low-risk or abstinent drinkers at one year than placebo. That's a real signal in exactly the population he resembles — someone for whom the standard drugs didn't work.
I don't think the evidence is as one-sided as Bacloville alone suggests. ALPADIR — also a randomized, placebo-controlled French trial, using a fixed 180 mg target rather than individualized titration — found no significant benefit at six months. And separate from trial efficacy entirely, Chaignot and colleagues' 2018 national cohort of over 165,000 French patients found that doses above 80 mg a day carried more hospitalizations and deaths than acamprosate or naltrexone at comparable follow-up. Baclofen has no FDA approval for this indication in the US for a reason.
Bacloville's positive result used individualized dosing that in practice often exceeds that same 80 mg threshold flagged in the safety data — the trial that worked and the dose range that raised concern aren't cleanly separable from each other.
Both of those trials matter, but neither one enrolled patients on a stable, long-term opioid regimen the way he is, and that's the piece I want to make sure doesn't get lost in a debate about average trial outcomes. Baclofen is independently sedating, and combining it with his existing extended-release morphine raises the real, if individually variable, risk of additive respiratory depression — not a theoretical drug-class interaction, a documented one.
If we go forward, it should be at conservative starting doses, titrated far more slowly than either trial protocol, with his pain physician looped in explicitly rather than treating this as a pure addiction-medicine decision that happens to run alongside his opioid regimen.
Agreed: a trial of baclofen, started conservatively and titrated more slowly than either French protocol, with his pain clinic informed and involved from the first dose rather than after the fact.
Not agreed: whether baclofen should be offered this readily to a patient outside a specialized program at all, given its lack of FDA approval and the genuinely mixed trial record. The clinical pharmacologist would have preferred referral to a research or specialty program with closer monitoring infrastructure before prescribing it in a general clinic setting; the addiction medicine specialist felt that bar would functionally deny him any further option, given how few programs exist. T.O. left having heard both concerns plainly stated, not softened for his benefit.