Twenty-five RCTs (n=3,290) were included in the review: 16 parallel design and 9 crossover design. Of these, 17 studies were included in the meta-analysis.
Traditional anticonvulsants.
Efficacy of treatment (3 RCTs, n=111): overall, a beneficial effect of treatment was found compared with placebo (OR 5.33, 95% CI: 1.77, 16.02); there was no evidence of statistical heterogeneity. When categorised by outcome (level of pain relief), there was a beneficial effect of treatment for moderate relief of pain (OR 10.63; 95% CI: 2.25, 50.13) but no difference between groups for a 50% reduction in pain (OR 3.04; 95% CI: 0.88, 10.54).
Withdrawals (4 RCTs, n=181): no between-group difference was found for number of withdrawals related to adverse events for traditional anticonvulsants compared with placebo.
Newer generation anticonvulsants. Efficacy of treatment (4 RCTs, n=623): overall, a beneficial effect of treatment was found compared with placebo (OR 3.25, 95% CI: 2.27, 4.66); there was no evidence of statistical heterogeneity. Categorisation by pain relief (50% reduction or moderate pain relief) did not substantially alter this result.
Withdrawals (5 RCTs, n=811): more withdrawals were found for those treated with newer generation anticonvulsants than those treated with placebo (OR 2.98, 95% CI: 1.75, 5.07).
Antidepressants.
Efficacy of treatment (3 RCTs, n=122): a significant effect in favour of tricyclic antidepressants (TCAs) was found compared with placebo (OR 22.24, 95% CI: 5.83, 84.75). Categorisation by outcome (notable improvement in global assessment of pain or moderate pain relief) did not substantially alter this result. A beneficial effect of duloxetine (60 mg and 120 mg) was found compared with placebo (OR 2.55, 95% CI: 1.73, 3.77 and OR 2.10, 95% CI: 1.03, 4.27, respectively). A beneficial effect of mexiletine and opioids was also found compared with placebo (weighted mean difference, WMD -1.87, 95% CI: -2.64, -1.11 and OR 4.06, 95% CI: 1.16, 14.21, respectively). No evidence of statistical heterogeneity was found.
Withdrawals: greater numbers of withdrawals due to adverse events of treatment were found for duloxetine 60 mg and 120 mg (OR 2.36, 95% CI: 1.05, and OR 4.65, 95% CI: 2.18, 9.94, respectively) and opioids (OR 4.06, 95% CI: 1.16, 14.21). No significant between-group differences in the number of withdrawals due to adverse events for TCAs or mexiletine were found.