Key points are not available for this paper at this time.
OBJECTIVE: So far no convincing answer has emerged to the question of whether transcranial magnetic stimulation (TMS) can make a clinically useful contribution to the treatment of depression. Here we examine whether multiple sensitivity analyses can highlight parameters that predict a favorable treatment response. DATA SOURCES: Medline, Embase, and the Cochrane database for controlled trials were searched for relevant randomized controlled trials using the expression (transcranial magnetic stimulation or TMS) and depression. STUDY SELECTION: Thirty-three studies were identified and included in the random-effects meta-analysis, and between 17 and 31 studies were included in the secondary analyses comparing outcome of studies with different parameters. DATA EXTRACTION: Study data were extracted with a standardized data sheet. A meta-analysis based on Cohen d effect size measure was done for all studies and various subsets. Regression analysis of effect sizes with study parameters was done in 24 studies. DATA SYNTHESIS: Active TMS treatment was more effective than sham, but variability was too great to take any single study design as paradigmatic. No significant predictors of study effect size were found. Mean effect sizes were reduced, although still significant, in studies with stimulation intensity below 90% of motor threshold and new medication starting within 7 days before to 7 days after start of TMS. CONCLUSIONS: The absence of significant outcome predictors in the presence of significant variability of outcome measures can be interpreted in 2 ways: either study sizes and numbers and designs are insufficient to afford the power necessary to detect such predictors or TMS has a nonspecific effect on depression that is not influenced by study parameters. Large-scale comparative trials are necessary to decide between these interpretations.
Herrmann et al. (Fri,) studied this question.