I think that there are two major issues with the literature on TMS:
1. Many of the studies only look at treatment-resistant depression, so it would be expected for it to be moderately effective.
2. All of the studies treat depression as a homogeneous illness, and they use essentially the same methodologies. This works when you're testing a drug, since most drugs nonspecifically target a downstream effect (decreased serotonin in the synaptic cleft) of the neuroplastic malfunction that's led to the phenotype of depression. But when you're talking about an intervention that's working directly on action potentials, you should expect that there will be different wiring malfunctions that are causing the same phenotypes with similar downstream effects. TMS studies generally haven't considered this factor because we don't have a good way to do that yet, and so they all target the prefrontal cortex. When it becomes more widely accepted, I think we'll have more research on finding different brain regions to target based on different clinical presentations. An easier thing (but still difficult to study) might be to do trials of TMS in 2-3 different locations for the same patient... I think that such a study would be expensive and difficult, but it would show that different people with similar phenotypes will respond to different forms of TMS because their underlying brain dysfunction is different. Our current approach is akin to treating all leukemia with cytotoxic chemotherapy rather than treating CML with imatinib.