Repetitive transcranial magnetic stimulation for the treatment of depression.

The use of repetitive transcranial magnetic stimulation (rTMS) as a therapeutic tool in neuropsychiatry is increasing, and careful assessment of the supporting scientific evidence is needed. Most studies to date investigating the efficacy of rTMS in medication-resistant depression have had small sample sizes and have used different methods and controls. Therefore, meta-analyses are valuable. Couturier,1 in a meta-analysis of randomized controlled trials investigating the efficacy of rTMS for the treatment of a major depressive episode, concluded that rapid-rate rTMS is no different from sham treatment in major depression. However, her conclusions are based on only 6 studies. She conducted a thorough analysis of 19 randomized controlled trials, but the eventual exclusion of 13 of them (to obtain a homogeneous sample) decreased the external validity of this meta-analysis. More important, as Couturier acknowledges, an analysis of 6 studies with a total of only 68 patients is insufficient to allow conclusions to be drawn about the efficacy of rTMS in depression. We believe that quantitative analysis of these data, with such a low number of patients and studies, should have been avoided, to prevent bias and misleading results. For instance, the author compared post-treatment scores on the Hamilton Rating Scale for Depression (HAM-D) for the sham and active groups rather than the difference between pre-treatment and post-treatment scores for these groups. Given the small sample sizes in the trials analyzed, the baseline values might be different between the active and placebo groups, and thus analysis of only the post-treatment scores could ignore significant weighted differences. For example, in one of the studies that was included in the meta-analysis (by George et al2) the post-treatment HAM-D scores were 22.2 for the active group and 19.0 for the sham group, whereas the pre-treatment (baseline) scores were 30.0 for the active group and 23.8 for the sham group. Comparing the post-treatment scores only, Couturier reported that this study favoured sham rTMS, when in fact patients in the active rTMS group had a larger improvement. The same flaw applies to the interpretation of the study by Padberg et al,3 in which the baseline HAM-D scores were 30.2 for the active rTMS group and 22.2 for the sham group. In this case, after treatment, the score for the sham group worsened but that for the active rTMS group improved. Although the difference was not significant, the weighted mean difference should have been in favour of active rather than sham rTMS. Finally, for the study by Eschweiler et al,4 Couturier reported the weighted mean difference as 0; however, the active rTMS group improved by 4.2 points and the sham rTMS group worsened by 3.0 points in terms of HAM-D results. We also find the exclusion criteria questionable. The author excluded 2 studies because they used the 25-item HAM-D5 or the 17-item HAM-D.6 However, Couturier could have calculated the effect size (instead of the mean difference) to allow these studies to be included, which would have increased the information and therefore the power of her meta-analysis. Indeed, the technique of meta-analysis was developed as a way to combine data from studies with different methods, a strength that seems to have been underutilized by the author. For these reasons we fear that this meta-analysis might mislead readers. There is no question that “randomized controlled trials with sufficient power to detect a clinically meaningful difference are required.” However, the questionable conclusions of Couturier's meta-analysis might discourage, rather than encourage, further prospective, well-designed clinical trials investigating the effects of rTMS on depression.