TMS Protocol Evidence: A Guide to Comparators, Targeting, and Durability

TMS evidence is easiest to misread when several different questions are compressed into one success rate. Does stimulation outperform a convincing sham? Can a shorter session preserve the benefit of a longer established protocol? Does an intensive schedule work in a selected research population? Does improvement last? Each question needs a different study design. This guide connects those questions without treating all magnetic stimulation as one interchangeable treatment.

The practical purpose is to improve conversations with clinicians, not to choose stimulation settings independently. A protocol includes the device and coil, stimulation pattern, target, intensity, number of pulses, session spacing, and overall course. Evidence applies most directly when those features and the patient population resemble the study. Sharing a technology name is not enough to establish that two services deliver equivalent care.

Begin with the question a trial can answer

A sham-controlled trial estimates whether an active intervention adds benefit beyond the comparison experience under the study conditions. Randomization helps balance known and unknown influences between groups. Masking reduces the effect of expectations, although convincing sham stimulation is challenging because treatment makes noise and produces scalp sensations. Neither feature makes a trial perfect; both help distinguish treatment effects from improvement that might otherwise have occurred.

An active-comparator trial asks a different question. It may compare two established approaches without measuring either against sham in that study. A noninferiority design asks whether the newer approach avoids being worse than the comparator by more than a prespecified margin. That is not proof that the treatments are identical for every patient, outcome, or follow-up period. Look for the margin and the investigators' stated primary endpoint before reading a headline as equivalence.

What conventional rTMS established

The George and colleagues sham-controlled prefrontal rTMS trial reported an intention-to-treat analysis involving 190 participants. During its masked phase, remission occurred in 14.1% of the active-treatment group and 5.1% of the sham group. This is a useful demonstration of why an active-group percentage needs a comparator: some participants improve even without receiving the active intervention.

Those figures describe a particular outpatient, nonpsychotic unipolar depression sample and a historical protocol. They are not a universal estimate for modern services, every diagnosis, or every stage of treatment resistance. Different eligibility rules, outcome definitions, and course lengths can change the observed rate. A clinic should not select an unusually favorable figure from another population while dismissing less favorable evidence solely because it is older.

What shorter theta-burst sessions establish

The THREE-D randomized noninferiority trial randomized 414 participants and compared intermittent theta-burst stimulation with 10-Hz rTMS. It supported noninferiority of the studied iTBS protocol on the depression outcome. Its practical importance includes shorter stimulation sessions, which may reduce the burden of individual visits. Because both groups received active treatment, it was not a new active-versus-sham test.

Shorter stimulation time does not automatically mean fewer visits or a shorter overall course. Nor does it establish that more sessions each day, a different pulse total, or another target will retain the same outcomes. Our theta-burst versus 10-Hz comparison separates these claims. Ask whether a proposed service matches the studied protocol or combines the result with additional changes requiring their own evidence.

Why accelerated treatment needs its own evidence

The Cole and colleagues Stanford Neuromodulation Therapy randomized trial studied a specific intensive treatment package, with 29 treated participants. Individualized functional-connectivity targeting was bundled with accelerated, high-dose iTBS. The primary depression outcome was measured four weeks after treatment. This provides evidence about that package in its studied setting, not permission to generalize the findings to anything advertised as accelerated TMS.

The small sample makes precision and replication important. A promising average effect cannot establish how every subgroup will fare or resolve uncommon safety questions. It also cannot isolate which component caused the benefit. The individualized target, dose, spacing, and other elements were not separately randomized. Ask for evidence supporting the exact schedule being offered, using our accelerated-schedule checklist to organize the comparison.

Separate targeting accuracy from clinical effectiveness

Navigation can help a team reproduce a chosen coil position. An MRI-based approach may provide anatomical or connectivity information used to select that position. These are different technical functions. Showing that a coil reaches the intended location more consistently does not, by itself, prove that the location produces better depression outcomes. Clinical benefit needs patient outcome data, not only an image showing accurate placement.

Similarly, a trial of a bundled treatment cannot establish that its targeting method alone is superior. To answer that question directly, a comparison would need to hold other important treatment features sufficiently constant. Our targeting-methods evidence guide explains the distinction. A useful consultation identifies what the targeting method measures, what evidence links it to outcomes, and what remains uncertain about its added value.

Make the outcome definition visible

Response usually refers to a defined reduction in symptom severity; remission refers to symptoms falling below a specified threshold. These are not interchangeable outcomes. A person can experience meaningful improvement without reaching remission, while a low score at one assessment does not establish permanent recovery. Check the actual scale and threshold instead of assuming that words such as success, relief, or recovery have a standardized meaning.

The denominator matters just as much. A rate among everyone randomized answers a different question from a rate among treatment completers. Excluding people who discontinued because of burden, lack of benefit, or adverse effects can make a service appear more effective than it would be for new starters. Ask how missing assessments were handled and whether the published headline describes the primary analysis or a selected subgroup.

Keep follow-up on the same timeline

An outcome at the end of treatment, four weeks later, and six months later addresses different clinical questions. Early symptom change can be valuable without proving durability. To understand persistence, look for follow-up completeness, additional treatments, and whether participants received further stimulation. If improvement depends on continued care, that is important information rather than evidence that the initial treatment necessarily failed.

The trials discussed here do not create one universal maintenance schedule. An acute-treatment comparison cannot determine the best retreatment interval for every responder. A clinician should explain whether a continuation recommendation rests on directly relevant research, broader clinical experience, or an individualized response pattern. Ask what evidence would trigger reassessment instead of accepting an open-ended promise that more sessions always preserve benefit.

Compare populations before comparing percentages

Treatment resistance is not a perfectly uniform category. Studies may differ in the number and adequacy of prior medication trials, concurrent treatments, episode duration, and exclusion criteria. A study excluding psychotic depression cannot establish effectiveness for psychotic depression. Similarly, evidence in nonpsychotic unipolar depression does not automatically answer questions about bipolar depression, another diagnosis, or a markedly different level of clinical urgency.

These differences also make indirect comparisons unreliable. A high remission rate in a small, selected study and a lower rate in a broader trial do not prove one protocol is superior. A direct randomized comparison would answer that question more cleanly. When direct evidence is unavailable, a responsible discussion acknowledges the uncertainty and weighs feasibility and safety rather than presenting a league table of unrelated percentages.

Treat tolerability and feasibility as outcomes

A shorter session may be valuable even if symptom outcomes are similar, particularly when transport, work, or caregiving makes treatment difficult. Conversely, a compressed course may reduce travel days while creating more demanding treatment days or requiring time away from home. These burdens can affect whether someone completes care. They belong in the evidence conversation alongside depression scores, not in a separate sales discussion after treatment is chosen.

Ask how adverse effects, discontinuations, and missed visits were recorded. A trial too small to detect uncommon harms should not be described as proving that those harms are absent. Protocol-specific safety assessment and trained supervision remain necessary. TMS is not simply a consumer wellness exposure whose settings can be copied from a publication without clinical evaluation and device-specific procedures.

Bring an evidence summary to the consultation

A useful one-page comparison lists the exact proposed protocol, the closest supporting trial, the studied population, comparator, outcome definition, and assessment time. Add the main uncertainty and the burden of completing the course. This prevents a conversation from drifting between incompatible claims, such as citing a conventional-course trial to justify an intensive schedule while quoting a different study's remission rate.

Finally, ask the clinician to explain what the evidence does not establish. Honest limits make shared decisions stronger: no trial guarantees an individual response, and a promising technical innovation does not remove the need for diagnostic assessment and follow-up. The most useful evidence is not necessarily the most dramatic result. It is the result that most closely answers the decision facing this person, with its boundaries kept visible.

Medical disclaimer: This article provides general education, not diagnosis, individualized treatment advice, or emergency care. Evidence and regulatory statements refer to the dated sources cited, not an exhaustive review of subsequent changes.

Sources

1. George et al., sham-controlled prefrontal rTMS trial, 2010
2. Blumberger et al., THREE-D randomized noninferiority trial, 2018
3. Cole et al., Stanford Neuromodulation Therapy randomized trial, 2022