Thirteen studies (410 participants)were included: 4 RCTs (142 participants), one non-randomised controlled trial (48 participants) and 8 cohorts (220 participants).
The rates of moderate to marked improvement in the participants on heterocyclics (nortriptyline, trazadone, imipramine plus mianserin, and a 'mixture of heterocyclics') ranged from 36 to 75%, compared with 16 to 44% in the comparison groups (no treatment in 3 studies, placebo and desipramine plus mianserin). A statistically-significant reduction in depressive symptoms was reported in 2 of the 3 RCTs, the non-randomised controlled trial and one of the 2 cohort studies. In addition, there was a significant improvement in activities of daily living in one RCT. The rates of discontinuation due to side-effects ranged from 0 to 45% in the participants on heterocyclics, and from 0 to 100% in the comparison groups.
The 2 studies of SSRIs (one RCT comparing citalopram with placebo and one non-randomised trial comparing fluoxetine with no treatment) reported a significant reduction in depressive symptoms. The RCT reported a moderate to marked improvement in 62% of the patients in the intervention group, compared with 29% in the placebo group. Drop-outs due to side-effects were 12% (intervention) compared with 3% (control).
The 5 cohort studies (one prospective, 4 retrospective) of psychostimulants reported rates of improvement in depression of 40 to 80% (there were no placebo or 'no treatment' groups). The one retrospective cohort study of ECT reported a 95% rate of improvement. Drop-outs due to side-effects ranged from 4 to 18%.
The evidence suggested that there were contraindications to treatment for 83% of the patients in a group to receive heterocyclics (imipramine or desipramine plus mianserin), compared with 11% of those in a group to receive SSRIs. No information was presented on contraindications to treatment in patients who were to receive either a psychostimulant or ECT.
The overall inter-rater agreement concerning relevance (based on a sample of 39 titles and abstracts) was 100%. The agreement between abstractors for information relating to study setting, type of treatment, and the four criteria of methodological quality was 77%, 88%, and 62 to 100%, respectively.