The report by Martin et al (Reference Martin, Barbanoj and Schlaepfer2003) seems in conflict with previous meta-analyses of repetitive transcranial magnetic stimulation (rTMS) (Reference Holtzheimer, Russo and AveryHoltzheimer et al, 2001; Reference McNamara, Ray and ArthursMcNamara et al, 2001; Reference Burt, Lisanby and SackeimBurt et al, 2002). We wish to provide a broader context for interpreting these results.

The analysis by Martin et al was designed to minimise type 1 error – to identify the level of confidence that can be placed in purported antidepressant effects of rTMS. It combined only studies with similar methodologies, included only studies that met high standards of randomisation and blinding, and analysed only end-point depression ratings (rather than analysing change scores or controlling for baseline depression severity). With this approach, the review found a statistically significant effect size for high-frequency (> 1 Hz) rTMS applied to the left prefrontal cortex (– 0.35, 95% CI – 0.66 to – 0.04, P=0.03), but did not find evidence that antidepressant effects were clinically significant or that they persisted over time.

The other meta-analyses attempted to minimise type 2 error – to identify whether there is reason to believe that rTMS might have significant antidepressant properties warranting further investigation. They combined studies with different methodologies and calculated effect sizes based on changes in depression severity over time. Such a technique can be important when analysing studies where different treatment arms may start at different baselines. Using these analytic techniques, prior meta-analyses found effect sizes for high-frequency, left prefrontal rTMS ranging from 0.5 to 0.8, suggesting that rTMS does indeed have statistically significant antidepressant effects. However, these analyses all agree that the clinical significance of these effects is not yet established.

The results of the Martin et al review do not suggest at all that rTMS has no antidepressant effects. On the contrary, this methodologically rigorous review identifies statistically (but not clinically) significant, short-term antidepressant effects for 2 weeks of high-frequency, left prefrontal rTMS and recommends further studies to establish efficacy and identify optimal parameters. Even more importantly, numerous studies have shown that rTMS alters brain functioning, with effects ranging from altered gene expression in animals to modified cerebral perfusion in humans; in many cases, these effects are very similar to those seen with established antidepressant treatments.

With these points in mind, we offer the following recommendations to help guide use of rTMS in clinical and research settings.

1. (a) Given the small clinical effects seen with rTMS in studies to date, it does not seem that rTMS is appropriate for widespread clinical use at this time.

2. (b) Large, multi-site trials are warranted to clarify the antidepressant effects of rTMS.

3. (c) Future studies of rTMS should incorporate several improvements in study design, including appropriate (and well-documented) randomisation, adequate blinding of subjects and therapists (probably requiring an improved sham condition), and better assessment of the duration of any antidepressant effects.

4. (d) More research should be directed at clarifying which patient and treatment characteristics might lead to greater antidepressant effects with rTMS.

5. (e) More research should be directed at identifying and testing potential mechanisms by which rTMS produces antidepressant effects.