Twenty-one RCTs evaluated acamprosate (n=5,280). Twenty RCTs evaluated naltrexone (n=2,182).
The risk of having a first drink after abstinence was reduced significantly with acamprosate compared to placebo (RR 0.84, 95% CI: 0.78, 0.91; NNT 7.7, 95% CI: 5.6, 13.0) and with naltrexone compared to placebo (RR 0.93, 95% CI: 0.88, 0.99; NNT 17.4, 95% CI: 9.7, 111.0). The sample weighted proportions of unpublished data were 10.8 per cent for acamprosate and 33.8 per cent for naltrexone. Significant heterogeneity was found for both analyses: p<0.00001, I2 83.6% for acamprosate; and p=0.08, I2 33.8% for naltrexone.
The risk of relapse to heavy drinking was significantly reduced with acamprosate compared to placebo (RR 0.82, 95% CI: 0.73, 0.92; NNT 8.6, 95% CI: 5.7, 18.7) and for naltrexone compared to placebo (RR 0.80, 95% CI: 0.71, 0.91; NNT 8.1, 95% CI: 5.5, 16.5). The sample weighted proportion of unpublished data was 79.2 per cent for acamprosate and 0 for naltrexone. Significant heterogeneity was found for both analyses (p<0.0001, I2 75.5% for acamprosate and p< 0.0001, I2 64.6% for naltrexone).
For non-abstinent drinkers there was no significant difference in the risk of heavy drinking between acamprosate and placebo (RR 0.98, 95% CI: 0.94, 1.02), but the risk of heavy drinking was significantly reduced with naltrexone compared to placebo (RR 0.88, 95% CI: 0.80, 0.96; NNT to prevent one additional relapse to heavy drinking was nine).
The funnel plot was asymmetrical suggesting the potential for publication bias, but Begg's test showed no significant evidence (p=0.09 and 0.31 for the two main analyses).