Substance and claimed effects

Transcranial magnetic stimulation (TMS) is a non-invasive brain stimulation technique that uses a focused, time-varying magnetic field generated by a coil placed against the scalp to induce small electrical currents in underlying cortical tissue. The clinical form is repetitive TMS (rTMS), delivered in structured outpatient courses of typically 20–36 daily sessions over 4–6 weeks. Three indications carry US FDA clearance: major depressive disorder when at least one prior antidepressant has failed (NeuroStar 2008, with broader device class clearance since) FDA 2008, deep TMS for obsessive-compulsive disorder (Brainsway Deep TMS, 2018) FDA 2018, and short-term smoking cessation (Brainsway, 2020) FDA 2020. Claimed effects in scope for this entry are: remission/response in treatment-resistant depression (the dominant use case), symptom reduction in OCD, smoking abstinence, the side-effect profile, response durability, and access considerations (cost, time, geography, insurance). The entry covers all of these holistically; lateral indications under active study (PTSD, anxiety disorders, addiction beyond nicotine, chronic pain, neurorehabilitation) are out of scope.

Evidence by addressing question

Mechanism

Science. Magnetic pulses (~1–2.5 Tesla, ~100–300 µs duration) pass unimpeded through skull and induce currents in cortical neurons within ~1.5–3 cm of the coil (figure-8 coils ~1.5 cm; H-coils for deep TMS ~3–4 cm) Rossi et al. 2021. Stimulation frequency dictates direction: high-frequency (10 Hz, "5-second-on, 10-second-off" trains for standard depression protocols) is excitatory and increases cortical excitability; low-frequency (1 Hz, continuous) is inhibitory Lefaucheur et al. 2020. Theta-burst stimulation (TBS) compresses this into a different pattern — bursts of 3 pulses at 50 Hz repeated at 5 Hz; intermittent TBS (iTBS, 600 pulses over ~3 minutes) produces long-term-potentiation-like effects, continuous TBS (cTBS) produces long-term-depression-like effects.

Mechanism (biological). The depression protocol targets the left dorsolateral prefrontal cortex (DLPFC), which is hypoactive in major depression on functional imaging. Excitatory stimulation of left DLPFC modulates downstream limbic circuitry — particularly the subgenual anterior cingulate cortex (sgACC), which is hyperactive in depression — via established fronto-limbic networks. Functional connectivity between the stimulation site and sgACC predicts response; the Cole/Williams Stanford group used individualized fMRI-guided targeting based on this principle to design the SAINT/SNT protocol Cole et al. 2020. OCD protocols target the medial prefrontal cortex and anterior cingulate using H7-coil deep TMS, the network implicated in compulsions Carmi et al. 2019. The smoking cessation protocol stimulates lateral prefrontal cortex and insula, with symptom provocation (cue-exposure) immediately before each session to engage the addiction circuit Zangen et al. 2021.

Downstream neurochemistry: repeated sessions induce changes in BDNF, cortical GABAergic tone, and dopaminergic signaling in striatal targets; the relative contribution of each to clinical response is unsettled. The honest summary is that TMS produces a verifiable physical effect on cortex (motor-evoked potentials, EEG signatures) but the path from "cortical excitability shifted" to "depression remits" is partly inferred.

Evidence (depression)

Science. The pivotal multicenter sham-controlled RCT (O'Reardon et al. 2007, n=301, antidepressant-resistant adults, 4–6 weeks of daily 10 Hz left DLPFC stimulation) showed response rates of ~24% active vs ~12% sham on the MADRS at week 4, with significant separation on Hamilton-D scales — the basis for the 2008 FDA clearance O'Reardon et al. 2007, FDA 2008. The NIMH-funded OPT-TMS trial (George et al. 2010, n=190, treatment-resistant unipolar depression, sham-controlled) confirmed effect with ~14% remission at 3 weeks vs ~5% sham; open-label extension produced ~30% remission George et al. 2010, Lisanby et al. 2009. Meta-analyses pool effect: Berlim et al. 2014 (n=29 trials, 1,371 patients, treatment-resistant) showed response rate ~29% active vs ~10% sham, NNT ≈ 6; remission ~19% vs ~5% Berlim et al. 2014. Brunoni et al. 2017 network meta-analysis ranked bilateral, high-frequency left, and low-frequency right rTMS all superior to sham with effect sizes typical of antidepressants Brunoni et al. 2017. Naturalistic real-world data (Carpenter et al. 2012, n=307 across 42 US clinical sites) showed response ~58% and remission ~37% — higher than RCT averages, attributable to longer treatment courses (~30 sessions) and selection of patients with less severe resistance Carpenter et al. 2012. Theta-burst non-inferiority: the THREE-D trial (Blumberger et al. 2018, n=414, Canadian multicenter) showed iTBS — at ~3 min/session instead of ~37 min — produced equivalent response (~49% vs ~47%) and remission (~32% vs ~27%) compared to standard 10 Hz Blumberger et al. 2018. The Stanford SAINT/SNT trial (Cole et al. 2022, n=29, double-blind sham-controlled) used 10 daily iTBS sessions over 5 days, individualized fMRI-targeted, and produced ~79% remission at week 4 — but with very small n; replication is ongoing Cole et al. 2022. The accelerated protocol won FDA clearance in 2022.

Effect-size comparisons: Mutz et al. 2019 network meta-analysis comparing non-surgical brain stimulation modalities ranked high-dose iTBS and bilateral rTMS among the more effective options behind ECT for major depressive episodes Mutz et al. 2019. Versus pharmacology, the effect sizes for treatment-resistant depression compare favorably to a third antidepressant trial after two failures — STAR*D showed remission rates dropping from ~37% at step 1 to ~14% by step 3 of sequential pharmacology Rush et al. 2006; TMS's ~30% remission in this population is therefore a meaningful step up where pharmacology is hitting diminishing returns.

Practice. CANMAT 2016 lists rTMS as a first-line treatment for adults who fail at least one antidepressant trial Milev et al. 2016. APA-aligned consensus recommendations (McClintock et al. 2018) endorse rTMS for treatment-resistant depression and provide protocol detail; broad European guidelines (Lefaucheur et al. 2020) give Level A evidence for left DLPFC high-frequency stimulation in depression McClintock et al. 2018, Lefaucheur et al. 2020. Medicare and most US commercial payers cover TMS for major depression after documentation of two failed antidepressant trials (typically from different classes) at adequate dose and duration, plus typically one failed psychotherapy course.

Community. Patient-reported outcomes from large clinical networks (Carpenter 2012, Dunner 2014) show satisfaction and effect-size figures consistent with RCT data. Online communities (depression forums, NAMI accounts) report a pattern of "tried everything else first, then TMS worked" — consistent with the indicated population. A counter-thread reports relapse within 6–12 months without maintenance, also consistent with the literature.

Evidence (OCD)

Science. Earlier rTMS-for-OCD trials targeting DLPFC were inconclusive (Mantovani et al. 2010 showed promise in treatment-resistant OCD with low-frequency right DLPFC, n=21) Mantovani et al. 2010. The 2018 FDA OCD clearance was based on Carmi et al. 2019, a multicenter randomized double-blind sham-controlled trial (n=99) of deep TMS using the H7 coil targeting medial prefrontal/anterior cingulate cortex, delivered with personalized symptom provocation immediately before stimulation. Response (≥30% Y-BOCS reduction) was 38% active vs 11% sham; mean Y-BOCS reduction ~6 points vs ~3 points Carmi et al. 2019, FDA 2018. Protocol: 29 daily sessions over ~6 weeks plus a maintenance phase. Effect sizes are smaller than for depression and remission is uncommon; the typical outcome is partial response sustained for months in roughly a third of treated patients.

Practice. OCD TMS access is much more limited than depression TMS — fewer clinics offer it, and insurance coverage is patchy and often requires demonstration of multiple failed SSRI trials plus exposure-response prevention (ERP) therapy. ERP remains the first-line psychotherapy; TMS is positioned as augmentation for treatment-resistant OCD, not a replacement.

Evidence (smoking cessation)

Science. The pivotal multicenter sham-controlled RCT (Zangen et al. 2021, n=262 chronic smokers ≥10 cigarettes/day with prior failed quit attempts) delivered 18 daily deep TMS sessions over 6 weeks (H4-coil bilateral lateral prefrontal and insula targeting, with smoking-cue provocation immediately before each session), then weekly maintenance for 12 weeks. The primary endpoint — 4-week continuous quit rate during weeks 15–18 — was 28% active vs 11.7% sham (p<0.01). At 6 months from start, 19.4% active vs 8.7% sham were continuously abstinent Zangen et al. 2021, FDA 2020. Effect sizes are clinically meaningful — comparable to or exceeding varenicline in this hard-to-treat population. Real-world deployment of smoking-cessation TMS is still very limited; very few US clinics offer it as of mid-decade.

Evidence (other indications)

Not FDA-cleared but actively studied: PTSD (Lefaucheur 2020 Level B), generalized anxiety, chronic neuropathic pain (Level A for motor cortex stimulation), substance use beyond nicotine, post-stroke aphasia and motor rehabilitation, tinnitus. None of these rise to standard-of-care; many showed early-trial signal that didn't replicate cleanly. Out of scope for this entry — flagged as candidates for future entries.

Protocol

Standard 10 Hz depression protocol (NeuroStar and similar figure-8 systems): one session daily, Monday–Friday, for 4–6 weeks (typically 30–36 sessions total). Each session ~37.5 minutes. Patient sits in a reclining chair, awake, no anesthesia. Motor threshold is determined first (the pulse intensity that produces a visible thumb twitch when applied over motor cortex); treatment intensity is 120% of motor threshold. Coil placed over left DLPFC (typically located using the "5 cm rule" anterior to motor hotspot, or via 10–20 EEG F3, or via fMRI/MRI-guided neuronavigation in research centers). 3000 pulses per session typical. After-session: patient drives self home, returns to work.

iTBS depression protocol: identical schedule (daily, 4–6 weeks) but sessions are ~3 minutes — 600 pulses delivered as 2-second trains every 10 seconds. Demonstrated non-inferior to 10 Hz Blumberger et al. 2018; clinics increasingly default to iTBS for throughput reasons.

SAINT/SNT accelerated protocol: 10 iTBS sessions per day separated by ~50 minutes, for 5 consecutive days (50 sessions total in one week). fMRI-guided targeting of the area of left DLPFC most functionally anticorrelated with sgACC, individualized per patient. Approved for treatment-resistant depression 2022; access still limited to a small number of centers Cole et al. 2020, Cole et al. 2022.

OCD protocol: 29 daily deep TMS sessions over ~6 weeks. Patient first elicits OCD symptoms via personalized provocation (e.g., contamination thought, ordering disturbance) for ~3 minutes; the stimulation then runs ~18 minutes over medial prefrontal/anterior cingulate via H7 coil Carmi et al. 2019.

Smoking protocol: 18 daily deep TMS sessions over 6 weeks, with smoking-cue provocation before each session (showing the patient smoking imagery for ~3 minutes), then weekly maintenance for 12 weeks. H4 coil Zangen et al. 2021.

Contraindications

Science / Practice. The dominant absolute contraindication is implanted ferromagnetic or active medical devices in the head/neck region — cochlear implants, deep brain stimulators, vagal nerve stimulators, aneurysm clips, certain ocular implants, cranial metal that could move or heat. Cardiac pacemakers and ICDs are not absolute contraindications for standard rTMS (the field falls off rapidly with distance) but require coordination with cardiology. Pregnancy: deep TMS labels caution; standard TMS is broadly considered safe across pregnancy but data is limited and most clinicians use shared decision-making Rossi et al. 2021.

The headline risk is seizure. Modern estimates put incidence at roughly 1 per 30,000 sessions or 1 per 1,000 patients across a treatment course — markedly lower than early estimates because the 2009 Rossi safety guidelines codified pulse-frequency / intensity / inter-train interval limits that prevent kindling. The 2021 update reports that with adherence to current dosing, seizures are rare and typically occur in patients with additional risk factors (alcohol withdrawal, sleep deprivation, certain medications that lower seizure threshold like bupropion at high doses, clozapine) Rossi et al. 2021. History of epilepsy is a relative contraindication, not absolute; many epilepsy patients have safely received TMS.

Common nuisance side effects: scalp discomfort at the stimulation site (30–40% of patients), headache (30%), facial muscle twitching (transient, during stimulation), lightheadedness. These typically attenuate over the first week. Notably absent: cognitive side effects (no memory impairment, in clear contrast to ECT), sedation, weight gain, sexual dysfunction.

Misconceptions

The most common misconception is conflation with ECT (electroconvulsive therapy). The two share "we apply electricity to your brain for depression" at surface level and nothing else operationally: ECT requires general anesthesia, induces a controlled seizure, causes acute and (some) persistent cognitive side effects, has roughly twice the acute remission rate, and is the standard of care for severe / catatonic / suicidal depression. TMS requires no anesthesia, deliberately stays sub-seizure-threshold, has no cognitive side effects, lower acute remission, and is the standard for treatment-resistant outpatient depression that isn't urgent Mutz et al. 2019.

Second misconception: that TMS is experimental or unproven. It carries FDA clearance for three indications, multiple guideline endorsements, and is reimbursed by Medicare and most major insurers for depression — this is established medicine. Confusion stems from limited public awareness and from many clinicians (primary care, even general psychiatrists) not yet routinely discussing it as an option.

Third: that the patient is unconscious or sedated during sessions. Patients are fully awake, can talk, read, watch video, and drive themselves home. The session feels like rhythmic tapping on the scalp.

Fourth (community): that newer-is-always-better and accelerated SAINT-style protocols dominate standard. The accelerated approach is promising but the high remission rates come from a small early trial with intensive screening; standard daily iTBS or 10 Hz remains the well-replicated workhorse, with accelerated protocols a reasonable choice when timing matters (acute suicidality, patient cannot do a 6-week course).

Audience

The indicated and reimbursable population for depression is adults (18+) with major depressive disorder who have failed at least one (insurance often requires two) prior antidepressant trial at adequate dose and duration. Children and adolescents: emerging evidence; small trials in adolescents (Croarkin 2021) suggest safety and effect signal but with smaller samples; coverage is generally not available for under-18 Croarkin et al. 2021. Older adults respond — though with somewhat lower response rates, particularly above age 60 — and modern protocols (higher dose, longer course) have narrowed but not eliminated this gap Lisanby et al. 2009. OCD audience: adults with treatment-resistant OCD who have failed ERP + SSRI trials. Smoking cessation audience: adults ≥22 (per FDA label) with chronic smoking and prior failed quit attempts.

Alternatives

For treatment-resistant depression, the realistic ladder once two or more antidepressants have failed:

• Augmentation pharmacology — lithium, atypical antipsychotic (aripiprazole, brexpiprazole, quetiapine), T3 thyroid hormone, esketamine. Each adds metabolic / endocrine / dissociative side-effect burden.
• Ketamine / esketamine (Spravato) — rapid onset (hours-days), modest durability without re-dosing, dissociation during administration, abuse liability concerns. Network meta-analyses suggest comparable acute effect size to TMS in TRD.
• ECT — highest acute efficacy (~60–70% remission) but the cognitive cost, anesthesia, and stigma push it to later in the ladder for most patients Mutz et al. 2019.
• Psychotherapy intensification — CBT, behavioral activation; lower acute effect than biological options but durable.

Versus all four, TMS occupies a particular slot: comparable or somewhat smaller acute effect than ketamine and ECT, larger than another antidepressant trial, no anesthesia, no cognitive side effects, no abuse liability, the inconvenience cost is daily clinic attendance for 4–6 weeks. For OCD: ERP first, SSRI second, clomipramine third, augmentation fourth, TMS fifth, surgery (anterior capsulotomy, DBS) sixth. For smoking: varenicline, NRT combinations, bupropion, behavioral therapy first; TMS sits at the back of the ladder.

Failure-modes

Common pattern: response without remission (symptoms drop 30–50% but don't reach the diagnostic threshold for remission), followed by gradual return of symptoms over months. The literature shows roughly 50% of acute responders relapse within 12 months without any maintenance strategy Dunner et al. 2014, Janicak et al. 2010. Re-induction (a fresh course) is the typical response and most insurers cover re-induction after documented relapse. Maintenance strategies — periodic single sessions, taper schedules — have less consistent evidence; clinical practice varies. Mantovani et al. 2013 reported that responders maintained on continuation pharmacology had ~50% durable response at 6 months Mantovani et al. 2013. A second failure mode is the under-dosed course: insurance approval for 20 sessions when the patient actually needs 30; clinics that taper sessions when symptoms improve mid-course rather than completing the full course. Real-world data shows clearer outcomes with longer courses (Carpenter 2012) Carpenter et al. 2012.

Third failure mode: targeting / coil-position drift over a 6-week course. Standard "5 cm rule" targeting can miss the optimal DLPFC site by 1–2 cm in patients with atypical neuroanatomy. fMRI-guided neuronavigation reduces this variance but is not yet standard outside research-affiliated clinics. Daily re-positioning by technician introduces some session-to-session variability that may explain why naturalistic effects don't always match best-RCT effects.

Practicalities

Cost. A full depression course of ~36 standard-10-Hz sessions billed at ~$300–500/session (varies by region; iTBS sometimes billed lower because shorter) is roughly $10,000–18,000 sticker price. With insurance coverage for treatment-resistant depression, patient out-of-pocket is typically copays only — for many patients, $0–$2,000 total. Insurance approval requires documented adequate trials of 2+ antidepressants (different classes), a clinician letter, and often one failed psychotherapy course. OCD coverage exists but is more variable; smoking cessation TMS is largely cash-pay.

Time. Standard 10 Hz course: ~37.5 min/session + travel + waiting, daily M-F for 4–6 weeks. Realistically, this consumes about an hour of a working day. iTBS course: ~3 min in the chair, but still daily clinic visits — overall time burden ~30–45 min/day including travel. SAINT/SNT: 5 days totally devoted to the protocol, ~10 hours/day at the clinic.

Geography. Concentrated in mid-to-large urban centers with academic psychiatry or larger private practices. Rural and small-city access remains thin; even modest metros may have only 1–2 providers, creating wait-list bottlenecks.

The session itself. Patient sits in a chair similar to a dental chair. Coil positioned and held in place. Stimulation begins — rhythmic tapping/clicking sound (around 100 dB; earplugs are provided) and a tapping sensation on the scalp. Sessions are loud but otherwise undemanding. Patients can listen to podcasts or music with bone-conduction headphones or low-volume earbuds outside the stimulation window.

History

The motor-evoked-potential paradigm (single-pulse TMS to map motor cortex) was demonstrated by Anthony Barker's group in Sheffield in 1985 — single pulses, no clinical aim. Repetitive TMS emerged in the early 1990s. The first FDA clearance came in 2008 (NeuroStar, depression) FDA 2008. Deep TMS (Brainsway H-coils) received first depression clearance in 2013; OCD in 2018 FDA 2018; smoking cessation in 2020 FDA 2020. Theta-burst protocols moved from research to clinical use after the Blumberger 2018 non-inferiority trial. Accelerated SAINT/SNT cleared FDA in 2022.

Stakes and payoff

Stakes: in the indicated population (treatment-resistant depression specifically), continuing the ladder of medication trials at this point is a low-yield game — STAR*D's step-3 and step-4 remission rates dropped to ~14% and ~13% respectively Rush et al. 2006. The reader who has been on three antidepressants for several years has a substantial probability of still being depressed in another two years. Payoff: durable remission in ~30–40% of those treated, partial response in another ~20–30%; for the subset who reaches remission, the magnitude of the change is the difference between "function is broken" and "function is intact." OCD and smoking payoffs are smaller in magnitude but real — for smoking, ~20% continuous abstinence at 6 months in a population with prior failed quit attempts is a meaningful additional shot.

Credibility range

Optimist case. TMS is settled medicine for treatment-resistant depression. It has FDA clearance from 2008; multiple positive sham-controlled RCTs; large naturalistic series showing real-world effectiveness; CANMAT, European, and APA-aligned consensus endorsements; broad insurance coverage; a clean side-effect profile that compares favorably to all alternatives (no anesthesia, no cognitive cost, no metabolic burden, no abuse liability, no sexual side effects). Acute remission ~30% in patients for whom each subsequent antidepressant trial has ~14% remission means TMS is the highest-yield option remaining in the ladder before ECT — which has higher remission but a much steeper side-effect price. The OCD signal (Carmi 2019) is replicated in a properly powered sham-controlled multicenter trial. The smoking signal (Zangen 2021) is the same. The SAINT/SNT accelerated protocol points toward dramatically higher remission rates in less time when targeting is individualized. The mechanism — modulation of fronto-limbic networks via cortical stimulation — has direct fMRI confirmation. The dominant "loss" to a patient who tries TMS is time and inconvenience; the dominant "gain" if they're a responder is a categorically different baseline.

Skeptic case. Sham TMS is hard. The clicking sound and scalp sensation of real stimulation are not perfectly mimicked by sham coils; blinding has been imperfect in several pivotal trials, which inflates apparent effect via expectancy. The placebo response in depression trials is large in general (sham response rates 25–35% are routine), making the active-sham gap modest in absolute terms. Naturalistic effectiveness (~58% response in Carpenter 2012) is much higher than RCT effectiveness (~29% response in Berlim meta-analysis), suggesting selection effects and regression to mean in real-world data. Response durability is poor: ~50% relapse within a year without maintenance, and what "maintenance" looks like is not well evidence-based. The SAINT remission figures (~79%) come from very small samples (n=29 in the 2022 sham-controlled trial) and are likely to attenuate with replication; independent replications are still scarce. For OCD the effect-size is small in absolute terms (Y-BOCS reduction ~6 points, a Y-BOCS responder typically still has significant OCD); for smoking, effect sizes are comparable to varenicline but with vastly higher access friction. Cost-effectiveness is debated: $10,000+ per course, with high relapse, may compare unfavorably to repeated pharmacology cycles. Major confounders in observational data: TMS patients are highly motivated (they show up daily for 6 weeks), educated, insured — none of this generalizes cleanly to broader populations.

Author's call. Land on the optimist side for treatment-resistant depression specifically — the evidence is solid (multiple positive RCTs, regulatory clearance, guideline endorsement, replicated effect sizes), the mechanism is plausible and partially verified, the side-effect profile is genuinely better than every alternative at this point in the ladder, and the population (treatment-resistant adults with diminishing returns from further pharmacology) is precisely the population for which the modest effect size is most valuable. For OCD: cautiously positive — real signal in a properly designed pivotal trial, but smaller effect size and less robust replication outside the registration trial; positioned as augmentation, not primary. For smoking cessation: real signal but limited access; honest framing is "another tool, not a game-changer." The skeptic case rightly tempers the SAINT-protocol hype and the durability picture — both should be flagged in the article. Evidence: 4 (multiple positive RCTs, broad guideline endorsement, regulatory clearance — but with the small remission-vs-sham gap honestly named). Controversy: 2 (real disagreement on accelerated protocols, durability, and effect-size inflation in naturalistic data — but no foundational dispute about whether it works).

Stakeholder and incentive map

• Commercial. Device manufacturers (Neuronetics/NeuroStar, Brainsway, MagVenture, MagStim, Nexstim). Strong incentive to push protocol expansion and indication broadening. TMS clinic chains (NeuroStim TMS, Greenbrook TMS Therapy) operate franchise-style. Both have an incentive to advertise the optimistic effect-size and downplay durability problems.
• Professional. Academic neurostimulation societies (Clinical TMS Society, International Federation of Clinical Neurophysiology). Generally aligned with conservative interpretation of evidence; produce the safety and clinical-application guidelines that anchor practice Rossi et al. 2021, Lefaucheur et al. 2020.
• Regulatory / payer. FDA, Medicare/CMS, private insurers. The FDA has been cautious — depression clearance 2008, OCD 2018, smoking 2020 each required dedicated multicenter sham-controlled trials. Medicare coverage came in 2013, after considerable delay. Coverage criteria (two failed antidepressants, adequate dose/duration, prior psychotherapy) limit access but also discipline use to the indicated population.
• Competing therapies. Ketamine clinics, ECT programs, the Spravato (intranasal esketamine) supply chain — all compete for the treatment-resistant-depression patient. Each pushes its own efficacy/safety narrative.
• Skeptic / counter. Some psychotherapy-oriented clinicians frame TMS as an "easy out" that bypasses upstream psychosocial work; some biological psychiatrists are skeptical of effect size relative to a fresh pharmacology trial. Methodologists have raised concerns about sham-blinding integrity, the placebo magnitude, and the gap between trial and naturalistic data.

Population variability

• Severity. Better response in moderate-severity treatment-resistant depression than in severe melancholic/psychotic presentations (the latter remain ECT territory). Inadequate-but-not-failed responders to medications also respond to TMS adjunct.
• Age. Adolescent and young-adult trials show signal but smaller evidence base Croarkin et al. 2021. Older adults (60+) historically had lower response rates because of cortical atrophy increasing scalp-to-cortex distance; this has been partly addressed by higher-intensity protocols and adjusted coil positioning.
• Prior treatment history. Patients with fewer prior failed antidepressant trials (Stage 1–2 resistance) respond more robustly than patients with multiple prior failures (Stage 4+) Lisanby et al. 2009. Voigt et al. 2019 in non-treatment-resistant patients also showed signal, supporting earlier deployment, but this isn't where current US coverage is positioned Voigt et al. 2019.
• Comorbid anxiety. Doesn't appreciably reduce depression response.
• Bipolar. Bipolar depression has signal but with theoretical mania-induction risk; protocols typically use low-frequency right DLPFC instead of high-frequency left to mitigate.
• Smokers / heavy alcohol. Higher seizure risk; smoking cessation TMS is paradoxically the indication that recruits smokers, with no excess seizures reported in the registration trial.

Knowledge gaps

Long-term durability beyond 12 months is poorly characterized — most maintenance studies run 6 months or less. Predictors of response remain investigational (genetic markers, fMRI connectivity patterns, EEG signatures); we cannot yet tell a given patient with confidence whether they will respond. Optimal maintenance protocol — frequency, density, taper — has no clear evidence-based answer. The SAINT/SNT remission figures need independent replication beyond Stanford. Sham-blinding integrity is methodologically still imperfect; better sham designs (low-intensity active stimulation matched for scalp sensation) are being developed but not yet standard. The translation of single-pulse cortical effects to multi-session network-level clinical effects is incompletely understood; mechanism research lags clinical adoption. Comparative-effectiveness against ketamine/esketamine in head-to-head trials is sparse; most patients receive both in sequence rather than randomly assigned. Generalizability beyond the indicated populations (depression, OCD, smoking) is open: PTSD and chronic pain show signal but neither has FDA clearance and trial data is heterogeneous.

Scope coverage vs the brief. The brief named treatment-resistant depression, OCD, smoking cessation, side-effect profile, response durability, and access considerations. All six are covered end to end in the article — TRD is the centre of gravity (it's where the evidence and the indicated audience are strongest), OCD and smoking get smaller but proportionate treatment in evidence/protocol/audience/payoff/out-of-scope. No silent narrowing.

What was excluded and why.

• Off-label and emerging indications (PTSD, generalized anxiety, addiction beyond nicotine, chronic neuropathic pain, post-stroke rehabilitation, tinnitus, adolescent depression specifics). Real research base for several but no FDA clearance, no consistent guideline endorsement, and the evidence is too heterogeneous to do honestly inside this entry. Flagged for separate entries as the literature matures.
• Detailed mechanism neurochemistry (BDNF, GABAergic tone, dopamine release in striatal targets) — present in the research dossier but compressed in the article to "modulates downstream emotion circuits" because the molecular detail isn't load-bearing for any reader decision.
• Specific device-brand comparisons (NeuroStar vs Brainsway vs MagVenture vs MagStim) — declined to name preferences. The clinical literature doesn't show meaningful between-device effect differences for the same protocol, and naming preferences would read as endorsement.

Hard scoring calls.

• Evidence: 4 not 5. Multiple positive sham-controlled RCTs, three FDA clearances, broad guideline endorsement (Milev 2016 CANMAT, Lefaucheur 2020, McClintock 2018). What kept it off 5: sham-blinding integrity is methodologically imperfect (real-vs-sham scalp sensation differs), absolute effect sizes are modest, and the OCD / smoking literatures rest on single pivotal trials rather than the multi-RCT base depression has. The 5-tier anchor ("multiple large RCTs or equivalent (Cochrane-level), consistent") is met for depression specifically; averaging the substance holistically lands at 4.
• health_short_term: 3. Defensible because the substance produces a substantial functional change in the indicated population (TRD responders); the score is per the substance, not per the median reader. Could be argued at 2 if one weighted the non-responder subset more heavily.
• longevity and sleep: 0. Considered 1 for both (indirect via depression remission) and decided against. TMS-specific longevity outcomes aren't measured, and any sleep effect is downstream of mood remission rather than a direct sleep-architecture effect. Honest call per meta §5a — don't score from a plausible-mechanism prior when the dossier hasn't earned it.
• controversy: 2. Real but bounded — disagreement on accelerated SAINT/SNT protocols, durability picture, and trial-vs-naturalistic effect-size gap, but no foundational dispute that TMS works for the indicated patient.
• cost_burden: 3. The score reflects sticker price; the pitch and article body translate honestly that insurance covers most of it for the indicated patient. Could be argued 2 if scoring post-insurance reality.

Contraindications vocabulary gap. The closed contraindications vocabulary lacks tokens for the dominant TMS-specific issues — implanted-ferromagnetic-device and seizure-history are absent. Only pregnancy was tagged, which understates the practical screening list. These are captured in the article's warning callout, but the structured tag layer is incomplete. Worth proposing token additions.

Action / cadence call. action: decide rather than do — TMS requires a prescriber, screening, and insurance approval; it's a deliberate decision, not a habit to start. cadence: course reflects the time-limited 4–6 week structure with a defined endpoint.

Category call. Placed in medical (Medical & Healthcare) rather than mental (Mental & Cognitive). TMS is a device-based clinical procedure delivered in outpatient courses; the medical bucket fits how a reader would browse to it. Cross-link from a future mental-category overview of treatment-resistant depression would close the loop.

Future-link candidates (entries to cross-link once they exist):

• ECT (electroconvulsive therapy) — the closest comparator; explicitly contrasted in the misconceptions and out-of-scope sections.
• Ketamine / esketamine (Spravato) for treatment-resistant depression — the other modern non-pharmacology-only option in this slot.
• Varenicline / nicotine replacement / smoking-cessation pharmacotherapy — referenced as the first-line for smokers; smoking-TMS only makes sense in the back of that ladder.
• Exposure-response prevention therapy for OCD — the first-line that TMS-for-OCD augments.
• Treatment-resistant depression as a meta-entry — would benefit from a dedicated overview that this entry links into.

Separate-entry candidates surfaced during research.

• Deep brain stimulation (DBS) for OCD / depression — invasive neurostimulation, distinct enough to warrant its own entry.
• tDCS (transcranial direct-current stimulation) — a related but distinct non-invasive modality with weaker evidence; should not be conflated with TMS.
• The depression treatment ladder as an editorial overview — would make this entry, the antidepressants entries, the ketamine entry, and the ECT entry sit more legibly together.