What evidence gaps are holding back transcranial magnetic stimulation?

Does TMS really work for depression, and how well?

The strongest evidence for TMS is in treatment-resistant depression (TRD), but the results are mixed. A 2024 randomized trial of 278 patients found that repetitive TMS (rTMS) was superior to switching antidepressants—reducing depression scores by 17.4 points on the Montgomery-Asberg Depression Rating Scale (MADRS) compared to 13.2 for a medication switch [1]. However, a meta-analysis of 12 sham-controlled studies showed that while rTMS had a moderate-to-large effect size (0.81), the actual response rate was only 13.7% for active TMS versus 7.9% for sham—meaning about 1 in 7 patients improved significantly [2]. This gap between statistical significance and clinical meaningfulness is a key evidence gap: TMS works, but not for most patients, and we don't know why some respond and others don't.

Supporting this, a 2026 expert consensus panel strongly recommended TMS as a next-step treatment for TRD, alongside ketamine and antipsychotics [9]. But the same panel noted that evidence gaps persist, especially for newer interventions and real-world complexity. The takeaway: TMS is a valid option, but its modest effect size and variable response mean it's not a cure-all.

Does TMS work for epilepsy, consciousness disorders, or insomnia?

For epilepsy, the evidence is weak and inconsistent. A Cochrane review of 8 randomized trials (241 participants) found that only 2 of 8 studies showed a significant reduction in seizure frequency (72% and 78.9% reductions), while the other 6 found no benefit [5]. The review rated the evidence as low to very low certainty, meaning we can't confidently say TMS helps epilepsy. This is a major evidence gap: despite a plausible mechanism (low-frequency TMS inhibits cortical excitability), the variability in techniques and outcomes prevents clear conclusions.

For disorders of consciousness (DOC), results depend on the subtype. A 2025 trial in children with traumatic brain injury found that 5 Hz rTMS over the left prefrontal cortex improved consciousness scores (Coma Recovery Scale-Revised) and reduced a brain injury marker (neuron-specific enolase) [3]. But a 2026 review noted that TMS works best in minimally conscious states (MCS), with inconsistent effects in vegetative states, and no evidence in coma [6]. The key gap: we lack biomarkers to predict who will respond, and the evidence is uneven across DOC subtypes.

For insomnia, a 2023 umbrella review found that rTMS improved self-reported sleep quality and insomnia severity, but effects on objective measures (like polysomnography) were inconclusive [4]. A 2022 protocol for a trial in children with autism and insomnia aims to fill this gap, but results aren't out yet [7]. So, while TMS shows promise for sleep, the evidence is too sparse to guide clinical use.

What specific evidence gaps are holding TMS back?

The biggest gap is the lack of standardized protocols and biomarkers. Across the studies here, TMS parameters (frequency, intensity, target location, number of pulses) varied widely, making it impossible to compare results or pool data [5]. For example, the depression meta-analysis found that no study showed a mean Hamilton Depression Rating Scale (HDRS) decrease of more than 50%, and the reasons for response differences across studies were unclear [2]. Without knowing the optimal dose or target, TMS remains a trial-and-error treatment.

Another gap is the lack of long-term data. Most studies, like the 2024 depression trial, lasted only 8 weeks [1]. We don't know if TMS effects persist, or if patients need maintenance sessions. The 2026 expert panel highlighted this as a key limitation [9]. Additionally, for conditions like epilepsy and insomnia, the evidence is low-quality or sparse [4][5]. A scoping review on musculoskeletal conditions found that TMS has been used to measure brain excitability, but no studies have tested it as a treatment for pain or motor deficits [8]—a gap that could be explored.

Finally, there's a gap in understanding who responds. The 2026 DOC review proposed that TMS works best in patients with preserved brain networks (e.g., in MCS) but not in those with severe damage (e.g., coma) [6]. Similarly, the depression meta-analysis noted that response rates were low overall, suggesting that patient selection is critical [2]. Without predictive biomarkers, TMS is applied broadly, diluting its effectiveness.

About These Sources

This answer is built on 9 peer-reviewed studies — published from 2021 to 2026, 6 from 2024 or later, 6 in Q1 journals, collectively cited 324 times — selected as the most relevant from 9 studies that passed quality screening, drawn from 86 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions