Twenty-seven RCTs (n=5,310) were included. Of these, 15 RCTs evaluated treatment effects in patients with MDD (n=3,430), 6 RCTs evaluated patients with OCD (n=718) and 6 RCTs evaluated patients with non-OCD anxiety disorders (n=1,162).
The studies were generally of good quality; the median Detsky score was 0.88 for the 23 trials with adequate data. Methodological problems included a lack of strict intention-to-treat analysis and inadequate reporting of randomisation methods and exclusions.
MDD.
Antidepressants were associated with a significantly increased treatment response rate compared with placebo, 61% versus 50%; the RD was 11% (95% CI: 7.1, 14.9; 13 trials) and the NNT was 10 (95% CI: 7, 15).
Antidepressants were associated with a non-statistically significant increased risk of suicidal ideation/suicide attempt compared with placebo, 3% versus 2%; the RD was 0.9% (95% CI: -0.1, 1.9) and the NNH was 112.
OCD.
Antidepressants were associated with a significantly increased treatment response rate compared with placebo, 52% versus 32%; the RD was 20% (95% CI: 13, 26.6; 6 trials) and the NNT was 6 (95% CI: 4, 8).
Antidepressants were associated with a non-statistically significant increased risk of suicidal ideation/suicide attempt compared with placebo, 1% versus 0.3%; the RD was 0.5% (95% CI: -1.2, 2.2) and the NNH was 200.
Non-OCD anxiety disorders.
Antidepressants were associated with a significantly increased treatment response rate compared with placebo, 69% versus 39%; the RD was 37.1% (95% CI: 22.5, 51.7; 6 trials) and the NNT was 3 (95% CI: 2, 5). All studies showed benefit with antidepressant use but significant heterogeneity was found (p<0.001).
Antidepressants were associated with a non-statistically significant increased risk of suicidal ideation/suicide attempt compared with placebo, 1% versus 0.2%; the RD was 0.7% (95% CI: -0.4, 1.8) and the NNH was 143.
There was an increased RD of suicidal ideation/suicide attempt across all trials and indications for drug versus placebo (RD 0.7%, 95% CI: 0.1, 1.3; 27 trials). There were no completed suicides in any of the studies.
There was no evidence of publication bias for any of the medical conditions.