Eighty-five RCTs (18,937 participants) were included in the review. Follow-up ranged from 12 weeks to 15 months.
Disulfiram (11 RCTs): Nine trials were used to draw conclusions as two used a method of administration without proven efficacy. One of the nine trials had lower potential for bias.
It was unclear whether disulfiram was an effective intervention. There were mixed findings for all comparisons with placebo and active interventions. Some outcome measures showed a statistically significant benefit and others did not.
Topiramate (four RCTs): Three RCTs were judged to have a lower potential for bias.
There was a statistically significant reduction in alcohol consumption for topiramate compared with placebo (three trials; lower potential for bias). There did not appear to be a difference with naltrexone on most outcomes (two trials) but one trial found differences on some outcomes.
Naltrexone (31 RCTs): Seventeen trials were judged to have a lower potential for bias.
Naltrexone was superior to placebo in most trials (25 trials), but no difference or mixed findings were found in some (five trials). Naltrexone combined with sertraline (an antidepressant) was effective in one trial that included participants with alcohol dependence and depression but not in another trial where the participants were not depressed.
The two trials included depot injections and both found statistically significant benefit compared to placebo injections.
Antidepressants (10 RCTs): Seven trials examined the use of selective serotonin reuptake inhibitors (SSRI) and all were judged to have low potential for bias.
Four trials compared sertraline (an SSRI) with placebo in people with depression or PTSD. There was no difference with sertraline in three trials and mixed findings in another.
Sertraline combined with naltrexone was not associated with benefit in one trial but there was a statistically significant benefit in another trial that included people with comorbid depression.
For desipramine (a tricyclic antidepressant) there was some evidence of benefit compared with placebo (one trial). Quetiapine (an antipsychotic) was not associated with benefit when used in combination with lithium and divalproex.
Acamprosate (24 RCTs): Fifteen trials were rated as having a low potential for bias.
There were mixed findings in 11 trials that compared acamprosate with placebo. Six of these trials found some benefits for acamprosate and five trials found either no significant differences or mixed findings.
There were mixed findings for the combination of acamprosate and naltrexone (two trials). One trial found no difference with naltrexone compared with other medications. The other trials found acamprosate to be less effective than naltrexone (one trial) and disulfiram (two trials).
Adjunctive psychosocial interventions (11 RCTs): Most studies of pharmacological interventions summarised above included adjunctive psychosocial interventions. Eleven trials examined this separately and seven were judged to be of low risk of bias. There were mixed findings concerning the benefits of adjunctive psychosocial treatment.