Magnetic brain stimulation for depression: what is proven and what is still promising

This article is general information, not medical advice.

Transcranial magnetic stimulation (TMS) places a coil against the scalp and sends short magnetic pulses into the brain. The pulses create small electric currents in the outer layer of the brain, and repeated pulses over one area can change how active that area is.

Newer versions aim to shorten the course. The most discussed is accelerated TMS, which packs many short sessions into about five days instead of several weeks. This article separates three things: what is in routine use, what has been shown in trials, and what remains speculative. Anyone considering TMS should talk to a clinician.

Established: standard TMS for depression

Standard repetitive TMS (rTMS) is a clinic treatment given over several weeks. The US Food and Drug Administration (FDA) granted a De Novo request for the NeuroStar TMS System, classed as a transcranial magnetic stimulator, on 7 October 2008 (FDA De Novo record DEN070003). The database entry does not set out the exact indication, so we do not restate it here.

The main recognised risk is seizure. A 2020 literature review puts the overall risk of TMS-related seizures at under 1%, describes seizures as a rare but serious effect, and says they are usually self-limiting (review on TMS and seizures).

Standard rTMS is therefore the established baseline. Accelerated versions are a different protocol and cannot be assumed to share its regulatory status or its evidence.

Shown in trials: faster and accelerated TMS

The results below come from peer-reviewed studies and apply to the people and conditions each study tested.

Faster pulse pattern versus standard treatment. The THREE-D trial compared intermittent theta-burst stimulation (iTBS) with standard 10 Hz rTMS in adults aged 18 to 65 with treatment-resistant depression (Blumberger and colleagues, The Lancet, 2018). It randomised 414 people at three Canadian university hospital centres: 205 to 10 Hz and 209 to iTBS. An iTBS session takes 3 minutes, against 37.5 minutes for a standard 10 Hz session. Treatment ran five days a week for four to six weeks. iTBS was non-inferior to 10 Hz on change in HRSD-17 score, against a pre-set margin of 2.25 points. Scores fell from about 23.5 to 13.4 with 10 Hz and from 23.6 to 13.4 with iTBS. Headache was the most common treatment-related side effect (64% with 10 Hz and 65% with iTBS). Self-rated pain during treatment was slightly higher with iTBS (3.8 out of 10, against 3.4 with 10 Hz). The authors state that devices could treat several times as many patients per day with iTBS. That is a claim about throughput, not a tested clinical result.

Accelerated SAINT protocol, open-label. Stanford Neuromodulation Therapy (SNT, first called SAINT) combines iTBS with targeting based on each person's brain scan. In an open-label study, 22 people with treatment-resistant depression started treatment and one withdrew. Each received 50 sessions, ten a day over five consecutive days. Of the 21 who completed, 19 (90.5%) met remission criteria; counting all 22 who started, the figure was 19 of 22 (86.4%) (Cole and colleagues, American Journal of Psychiatry, 2020). With no control arm, the authors say double-blind sham-controlled trials are needed to confirm this.

Accelerated SAINT protocol, sham-controlled. A double-blind trial had enrolled 32 people at a planned interim analysis, and 29 who still met the entry criteria were treated: 14 with active SNT and 15 with sham. Four weeks after treatment, the mean reduction in depression scores on the MADRS was 52.5% with active treatment and 11.1% with sham. The authors state that further trials are needed to determine durability and to compare SNT with other treatments (Cole and colleagues, American Journal of Psychiatry, 2022).

Replication at one site. A later double-blind, sham-controlled trial at Stanford randomised 48 people (24 per arm), with the study running from June 2021 to June 2024. Remission at one month, defined as a MADRS score of 10 or lower, was 50.0% with active treatment and 20.8% with sham (p = 0.035). The authors report no serious adverse events (Kratter, Austelle, Lissemore and colleagues, World Psychiatry, 2026).

Regulatory clearance. On 1 September 2022 the FDA found the Magnus Neuromodulation System with SAINT Technology, model 1001K, substantially equivalent to an existing device (FDA 510(k) record K220177). The indication in that record is major depressive disorder in adults who have not improved enough on prior antidepressant medication in the current episode. This is a device decision. It does not show that accelerated treatment beats standard care.

Comparison with standard TMS. A 2024 meta-analysis of 14 studies, of which only four were randomised trials, found no statistically significant difference in effect between accelerated and standard TMS. The pooled standardised effect was -0.67 (95% confidence interval -1.62 to 0.27, p = 0.16), with heterogeneity of 93% (Shi and colleagues, BMC Psychiatry, 2024). The authors say accelerated TMS can improve symptoms more quickly but that efficacy does not differ significantly, and they caution that only three studies informed that comparison. The 2026 SNT trial's authors also note that SNT has yet to be tested directly against standard rTMS.

Longer-term maintenance. A different protocol, clustered rTMS, was tested for preventing depressive relapse in 281 randomised patients in China who had already improved on antidepressants. Twelve-month relapse or recurrence was 24.2% with rTMS alone, 15.9% with rTMS plus antidepressants, and 44.4% with antidepressants alone. Outcome assessors were blinded, but there was no sham arm and participants knew their assignment (Wang and colleagues, Translational Psychiatry, 2017). It shows that repeated sessions can be studied for maintenance, but it does not test SAINT.

Limits and open problems

• Response varies between people. Trial results are averages across groups, and the sources do not show who benefits most.
• Durability is unknown for accelerated treatment. The 2026 trial's authors state that longer-term durability "remains unknown", and the 2022 trial's authors call for further trials on durability.
• Trials are small and narrow. The largest sham-controlled SNT trial we found had 48 people at one site. Its authors describe participants as mostly highly educated, White or Asian, and non-Hispanic or Latino.
• Outcomes differ. Some trials report remission and others report change in a score, so results cannot be compared directly.
• Cost and access are not established. The sources we checked did not give figures for cost, waiting times or insurance cover. Individual targeting also needs a brain scan and processing before treatment.
• Side effects. Headache is the most common treatment-related effect reported in THREE-D, and seizure is the main serious risk, estimated at under 1% in the 2020 review. The sources checked for this article did not give figures for other effects such as scalp discomfort.
• UK position. The sources used here do not cover NICE or NHS guidance on accelerated protocols such as SNT, so UK availability is not addressed in this article.

Speculative

• Shorter courses in hospital or emergency settings. A five-day course might suit some patients better than a six-week schedule. None of the trials read here tested this setting.
• Predicting who will respond. The 2026 trial reports a candidate EEG biomarker: higher pre-treatment beta power in the left anterior cingulate cortex was linked to greater clinical benefit in the active arm but not in the sham arm. The authors call for prospective, biomarker-guided trials, so this is a hypothesis to test, not an established tool.
• Other psychiatric conditions. The 2026 trial's authors mention a separate SNT study in bipolar depression at two other academic sites that reported positive findings. We have not read that study, and all participants in the trials described here had major depressive disorder.

What to watch next

• A multi-site, sham-controlled SNT trial with follow-up beyond one month.
• Trials that compare accelerated protocols directly with standard TMS in larger groups.
• Independent replication by groups not involved in the existing SNT trials.
• Any UK guidance from NICE or the NHS on accelerated TMS, and any payer decisions on coverage.