Thirty-nine RCTs of high quality (Jadad score≥3) were included in the review: 20 RCTs of major depressive disorder (n=1,998); 15 RCTs of post-stroke depression (n=1,680); two RCTs of pre-menstrual depression; and one RCT for each of peri-natal depression and comorbid depression. Only studies of major depressive disorder and post-stroke depression were included in the meta-analysis, due to the small numbers of studies and participants in the other disease subgroups. Four studies were awarded a maximum Jadad score of 5 points, four scored 4 points and the remaining studies scored 3 points.
For patients with major depressive disorder there were no statistically significant differences between acupuncture monotherapy and control groups for clinical response or symptom severity. Subgroup analyses failed to find any statistically significant differences between acupuncture and antidepressants, sham acupuncture and waiting list control. Two pooled trials of acupuncture plus antidepressants versus antidepressants alone showed a statistically significant improvement in HAMD scores in favour of the combination treatment (WMD 2.38, 95% CI 0.62 to 4.12, p=0.007, I2=0%)
For patients with post-stroke depression there was a statistically significant difference favouring acupuncture monotherapy in comparison with control groups for clinical response (RR 1.36, 95% CI 1.24 to 1.50, p<0.00001, I2=36%; 13 RCTs) and symptom severity (WMD 2.54, 95% CI 1.11 to 3.97, p=0.0005, I2=79%; 14 RCTs). Both these pooled analyses were associated with moderate to high levels of heterogeneity.
Subgroup analyses showed statistically significant differences in favour of acupuncture monotherapy in comparison with antidepressants and waiting list controls for both clinical response and symptom severity (further details reported in the review).
Incidence of adverse events was lower with acupuncture and sham acupuncture than with antidepressants (10.2% versus 40.4%, Χ2=389.457, d.f.=1, p<0.001).
There was no evidence of publication bias.