Reviews/Epilepsy Management/Vagus nerve stimulation reduces seizure frequency
Surgical Dre · EP-SRG-02 · v2
Borderline: within 4 points of the Established/Likely boundary, so a re-grade could land either side.
Across many separate studies in different countries, in both children and adults and across many underlying causes of epilepsy, seizures became less frequent after patients received vagus nerve stimulation for epilepsy that had not responded to drugs. The pattern was the same in every study: seizures went down, never up. One study that also followed a comparison group treated with medication alone found a clearly higher response rate in the stimulation group. The weakness is that almost none of these studies used a control or sham group, and seizure counts were reported by patients or caregivers, so the exact size of the benefit is hard to pin down. The panel treated the direction of the effect as well supported while flagging that the magnitude remains uncertain.
Twenty-two primary studies consistently report reduced seizure frequency after VNS in drug-resistant epilepsy, with 12-month median reductions of 70.4% [1], 73.9% [8] and 82.5% [20], and responder rates spanning roughly 40-70% [4] up to 90.5% [22]. The single study carrying a medication-only comparator reported VNS response of 65-75% versus 31-36%, with relative risks of response of 1.81, 2.21 and 2.42 at 1, 2 and 3 years [23]. Benefit appears across pediatric and adult cohorts, focal and generalized seizures, and diverse etiologies including structural [3], monogenic [2] and tuberous sclerosis complex [23]. Several cohorts also show the effect strengthening with longer follow-up [9][15][17]. Direction is unanimous across the corpus. The contested point: The division is over magnitude rather than direction: responder rates swing from 23% at six months [15] to 90.5% at last follow-up [22], and one real-world cohort found a decreasing seizure trend in only 27.5% of patients, placing VNS efficacy 'within the efficacy range of recently introduced medical treatments' [13]. Because every figure except [23] comes from open-label, self-reported counts, the true effect size stays unsettled.
All 25 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Epilepsy Management, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
The entire included corpus is open-label and uncontrolled except [23]; no RCT or sham-controlled trial is present, and seizure counts are patient/caregiver-reported, so regression-to-the-mean and expectation effects can inflate every reported magnitude.
[23], the only study with a comparator, was not randomized — allocation to VNS vs medication-only reflected actual treatment choice, exposing it to indication and selection bias.
Where it remains uncertain. No RCT or sham control is present and seizure counts are patient- or caregiver-reported, while responder rates range from 23% at six months [15] to 90.5% [22].
Aleatory
Seizure frequency fluctuates naturally and is patient/caregiver-reported; high-baseline patients can regress toward the mean regardless of treatment, as the open-label designs cannot separate.
Epistemic
The corpus contains no RCT or sham-controlled study; all efficacy figures except [23] are open-label and uncontrolled, so the magnitude of benefit is not fixed by this evidence.
Inferential
[23], the only study with a medication-only arm, allocated by actual treatment choice rather than randomization, so the 65-75% versus 31-36% difference may partly reflect indication and selection bias.
The load-bearing papers from a corpus of 25, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Primary quantified seizure-frequency reduction in focal and generalized DRE.
“After 12 months of μVNS, the mean seizure frequency reduction for all seizures was 61.3% (median reduction: 70.4%; 90% CI of median: 48.9%-83.3%).”
Panel auditInternal validity: low · scope of inference: Adults/children with drug-resistant focal or generalized epilepsy receiving investigational microburst VNS.Key limitations: no control arm · small feasibility sample · investigational non-standard VNS
Propensity-matched responder estimate in a large pediatric sample.
“The percentage of matched patients (total n = 220: n = 179 traditional VNS, n = 41 closed loop VNS) who responded to VNS after one year of therapy was 43% for traditional VNS and 39% for closed loop VNS (p = 0.64).”
Panel auditInternal validity: medium · scope of inference: Pediatric drug-resistant epilepsy, comparing VNS device types.Key limitations: no non-VNS control · retrospective · selection bias
Prospective multicenter evidence of large reduction in the most dangerous seizure type.
“Following 12 months of VNS, participants experienced a median GTCS reduction of 73.9%, with 37% of participants reporting seizure freedom from GTCS for the 3 months prior to the 12-month follow-up.”
Panel auditInternal validity: medium · scope of inference: DRE patients with generalized tonic-clonic seizures receiving adjunctive VNS.Key limitations: no control arm · GTCS subset only · patient-reported seizure counts
The reported seizure reduction is attributable to the RNS add-on, so VNS's independent contribution to reduction cannot be read from this abstract.
Panel auditInternal validity: medium · scope of inference: Adults with drug-resistant focal epilepsy on concurrent VNS and RNS.Key limitations: VNS effect not isolated · retrospective · no VNS-only comparator
4 of 25 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
Unanimous, independent, multinational observational evidence plus one comparator-controlled cohort firmly establish the directional reduction, while the absence of any RCT/sham in this corpus holds confidence to the low end of Established.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
No randomized or sham-controlled evidence. Every efficacy figure except [23] is open-label and uncontrolled, so the magnitude of seizure reduction cannot be separated from regression-to-the-mean and expectation effects. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A randomized, sham-stimulation-controlled trial in drug-resistant epilepsy with blinded, diary-plus-objective seizure counting.
Open question: How much of the reported seizure-frequency reduction survives blinding against sham stimulation? Why it is still open: The corpus contains no sham-controlled study, so expectation effects on self-reported counts cannot be excluded.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19