Naltrexone is one of two first-line medications for alcohol use disorder, but its effect is modest: a 2023 meta-analysis of 118 trials found it lowered the chance of returning to heavy drinking, with about 11 people needing to take it for one to benefit. [mcpheeters-2023] The largest single trial found a similar picture, with naltrexone and supportive medical management helping and the pill alone not being the whole story. [anton-2006] This guide covers what the main review and the COMBINE trial found, how naltrexone compares with other drugs, the side-effects, what the research could not show and what to ask a prescriber. Our overview of addiction covers the wider picture.
Does naltrexone work for alcohol use disorder?
It helps with heavy drinking, but modestly. The 2023 review pooled 118 clinical trials with 20,976 participants who had alcohol use disorder, all lasting at least 12 weeks. Compared with placebo, oral naltrexone at 50 mg a day was associated with lower rates of return to heavy drinking (risk ratio 0.81, 95% confidence interval 0.72 to 0.90), with a number needed to treat of 11 (95% CI 5 to 41). [mcpheeters-2023]
For avoiding any drinking at all, the number needed to treat was 18 (95% CI 4 to 32; 16 trials, 2,347 people). Injectable naltrexone was not associated with a lower rate of return to any drinking or to heavy drinking, but did reduce the percentage of drinking days (weighted mean difference -4.99 days over the 30-day period). [mcpheeters-2023] A number needed to treat of 11 means that for every 11 people who take it, one avoids heavy drinking who would not have with placebo, which is a real but limited effect.
How does naltrexone compare with acamprosate and other drugs?
It was no clearly better than acamprosate and was ahead of the other options. Acamprosate had a number needed to treat of 11 for any drinking, but it was not associated with lower return to heavy drinking (risk ratio 0.99). Four trials with 1,141 people that compared acamprosate and naltrexone directly found no significant difference between them on either outcome. [mcpheeters-2023]
Disulfiram, an older drug, showed no benefit over placebo for preventing any drinking in 3 trials with 622 people. Topiramate, which is not approved for alcohol use disorder, had moderate-strength evidence for fewer drinking days, but caused more cognitive dysfunction (risk ratio 2.37), paraesthesia and taste changes. Baclofen and gabapentin had low-strength evidence at best. [mcpheeters-2023]
| Medication | Return to any drinking | Return to heavy drinking | Strength of evidence |
|---|---|---|---|
| Oral naltrexone (50 mg a day) | Number needed to treat 18 | Risk ratio 0.81, number needed to treat 11 | Moderate |
| Acamprosate | Number needed to treat 11 | No benefit (risk ratio 0.99) | Moderate |
| Injectable naltrexone | Not lower | Not lower | Low |
| Disulfiram | No benefit in 3 trials | Insufficient | Low |
| Topiramate (off-label) | Not the main finding | Fewer heavy drinking days | Moderate, with more adverse effects |
What did the largest single trial find?
The COMBINE trial found naltrexone helped when combined with supportive care, and acamprosate did not. It randomised 1,383 recently abstinent people with alcohol dependence at 11 US academic sites to 16 weeks of medical management with naltrexone, acamprosate, both or placebos, with or without a combined behavioural intervention, plus a group with the behavioural intervention only. [anton-2006]
During treatment, people on naltrexone plus medical management (80.6), on the behavioural intervention plus medical management and placebos (79.2) or on both (77.1) had more days abstinent than the 75.1 percent of people on placebos and medical management only. Naltrexone also reduced the risk of a heavy drinking day (hazard ratio 0.72, 97.5% CI 0.53 to 0.98). Acamprosate showed no significant effect, alone or in combination. [anton-2006]
Percent days abstinent during treatment in the COMBINE randomised trial, Anton et al. (2006), 1,383 recently abstinent adults with alcohol dependence. Medical management means regular supportive appointments. The behavioural-only group had no pills and no medical management.
The authors concluded that no combination did better than naltrexone or behavioural therapy alone in the presence of medical management, and that placebo pills and meeting with a health professional had a positive effect during treatment. One year after treatment the between-group effects were similar but no longer significant. [anton-2006]
What are the side-effects of naltrexone?
Mostly gastrointestinal and mild. Compared with placebo in the 2023 review, naltrexone was associated with more nausea (risk ratio 1.73, 95% CI 1.51 to 1.98), vomiting (1.53, 1.23 to 1.91) and dizziness (1.99, 1.47 to 2.69). Acamprosate was associated with more diarrhoea (1.58), and in direct comparisons acamprosate caused less nausea and vomiting than naltrexone. [mcpheeters-2023]
The review also notes that oral naltrexone is contraindicated for people with acute hepatitis or liver failure and for people using opioids or likely to need them, because it can precipitate severe withdrawal in people dependent on opioid medications. [mcpheeters-2023] That is a reason to tell a prescriber about every medicine and liver problem before starting.
What can this research not show?
It cannot show how naltrexone works for you, over years, or without support. Five limits matter most:
- Most trials added support. Of the 118 trials, 100 included a co-intervention such as medical management or counselling, so the results reflect medication plus support against placebo plus support. [mcpheeters-2023]
- Participants were mostly moderate to severe cases. Applicability to mild alcohol use disorder is uncertain, and of the trials that reported it, most enrolled mainly White men, typically aged 40 to 49. [mcpheeters-2023]
- Short trials were excluded and long-term harms are unknown. The review included only trials of at least 12 weeks, and long-term adverse-effect data were not available. [mcpheeters-2023]
- Health outcomes are unclear. Evidence on mortality, injuries, crashes and quality of life was insufficient or low strength for most medications. [mcpheeters-2023]
- The big trial is old and benefits faded. COMBINE ran from 2001 to 2004 at US academic sites, and its group differences were no longer significant a year after treatment. [anton-2006]
What should you ask a prescriber?
Ask whether naltrexone suits your drinking pattern and health, how it will be combined with support, and what happens if it does not help. This section is our suggestion and is not something the studies tested, although both found that support mattered. [anton-2006]
Useful questions are whether your liver and any other medicines make it unsuitable, whether the aim is cutting down or stopping, how often you will be seen and how to tell a prescriber about a slip. Our guides to relapse, long-term recovery and how to help someone who drinks too much cover the surrounding support.
No screener on this site covers alcohol use. The self-assessment hub lists the seven we do have, and none is built for this question.
When to seek help
Speak to a doctor if your drinking is causing harm to your health, work, relationships or safety, or if you have tried to cut down and could not. If you drink heavily every day, do not stop suddenly without medical advice, because alcohol withdrawal can be dangerous. If you feel unable to stay safe, contact your local emergency services or a crisis helpline.
How MyFreud can help
A record shows a prescriber what is changing. Daily mood tracking in MyFreud shows how your days run over weeks, and the notepad is somewhere to note drinking days, cravings, side-effects and what helped. It does not replace a doctor or a treatment service.
Download MyFreud and start today: App Store or Google Play.