Epilepsy

Epilepsy is a common neurological disorder of recurrent seizures. The Cochrane review of acupuncture for epilepsy reports that about 70% of people with it become seizure free on antiepileptic drugs and up to 30% go on having seizures despite adequate doses, which makes their epilepsy drug resistant. The research on needling for epilepsy falls into two groups. One is Chinese trials of needle acupuncture and of catgut implantation, in which absorbable thread is embedded at the points, mostly compared with antiepileptic drugs or added to them. The other is trials of transcutaneous auricular vagus nerve stimulation (taVNS), in which a clip electrode on the skin of the ear conch stimulates the branch of the vagus nerve that supplies it, compared with sham or low-level stimulation. taVNS is distinct from implanted vagus nerve stimulation (VNS), in which a surgeon wraps an electrode around the vagus nerve in the neck and places a pulse generator under the skin of the chest. The trials measure seizure frequency from patient diaries, the share of patients whose seizures fell by at least half (the responder rate), and seizure freedom.

The Cochrane review

The latest version of the Cochrane review, by Cheuk and Wong, was published in May 2014 with searches to June 2013; the Cochrane Library lists no later version. It included 17 randomised trials with 1,538 participants, mostly with generalised epilepsy, treated for 7.5 weeks to a year. Sixteen trials were run in China and published in Chinese, and one in Norway; every trial had a high risk of bias and short follow-up. The comparisons were kept apart:

• Against sham acupuncture (one trial, adults with intractable epilepsy): median seizure frequency fell 45% with acupuncture and 20% with sham, with no confidence interval reported; quality of life did not differ. GRADE low.
• Added to Chinese herbs, against the herbs alone (two trials, children): no significant difference in the share with seizures halved. GRADE moderate.
• Added to valproate, against valproate alone (two trials): no difference in seizure freedom or in seizures halved, and a better quality-of-life score in one trial. GRADE low.
• Acupuncture alone, against phenytoin or valproate: no significant difference against phenytoin; against valproate, no difference in seizure freedom and more patients with seizures halved (risk ratio 1.32, two trials). GRADE low.
• Catgut implantation: added to antiepileptic drugs, more patients with seizures halved (risk ratio 1.42, five trials) but no difference in seizure freedom; alone against valproate, more patients seizure free (risk ratio 2.82, four trials) but no difference in seizures halved. GRADE low.

The authors concluded that the trials were small, heterogeneous and at high risk of bias, and that the current evidence does not support acupuncture for treating epilepsy. They asked for larger trials with controls that isolate the effect of acupuncture, separate trials in children and adults, and longer follow-up, because epilepsy waxes and wanes with or without treatment. Thread embedding describes catgut implantation.

The one sham-controlled trial was run at the National Center for Epilepsy in Norway. Kloster and colleagues reported that 29 patients completed it, 15 given classical acupuncture by two Chinese professors of acupuncture and 14 given sham. Seizure frequency fell in both groups without reaching statistical significance, and seizure-free weeks increased in both, significantly so in the sham group. A companion paper by Stavem and colleagues, on 34 patients who completed 20 treatments over eight weeks, found no difference between the groups on the 89-item Quality of Life in Epilepsy questionnaire. The sham needled three points outside the traditional channels with smaller needles. Sham acupuncture sets out what such a control does and does not rule out.

Later reviews of needling

Deng and colleagues, in Chinese with an English abstract, reported higher efficacy with acupuncture, alone or with drugs, than with conventional drugs (odds ratio 3.94), and wrote that the conclusion needed verification because they found very few high-quality, well-designed trials.

Xue and colleagues pooled 17 trials in which acupuncture was added to antiseizure drugs. They reported a higher effective rate, defined as seizures falling by at least half (odds ratio 4.28), lower seizure frequency, fewer epileptiform discharges on the EEG, better quality-of-life scores and fewer adverse events than with the drugs alone. They described the evidence as low to moderate and said the conclusions should be viewed with caution.

Su and colleagues pooled 10 trials in children aged 1 to 17, all conducted in China and published between 2001 and 2019, in which acupuncture and a herbal formula together were compared with antiepileptic drugs. They reported a lower risk of treatment failure (risk ratio 0.3) and of adverse events (risk ratio 0.27), at moderate certainty, and no difference in abnormal EEGs after treatment, at very low certainty. They named deficiencies in randomisation, allocation concealment and blinding among the trials' limitations. Because acupuncture and herbs were given together, the review cannot separate the two.

Stimulating the vagus nerve through the ear

Three meta-analyses have pooled the taVNS trials. Wu and colleagues in 2020 found three trials with 280 patients and reported lower seizure frequency with taVNS but no significant difference in responder rate, with headache the most common adverse effect. Moro and colleagues in 2025 found four randomised trials with 368 patients with drug-resistant epilepsy, all against sham stimulation. They reported about three fewer seizures a month with taVNS and a percentage seizure reduction 17.6 points larger, while the responder rate (odds ratio 2.00), seizure freedom, depression and quality of life did not differ significantly; they called for better-designed trials because of methodological concerns in some of the included studies. Makkawi and colleagues, searching to December 2023 for trials in people aged 14 and over, found four trials with 417 participants and reported lower seizure frequency at several time points, no significant difference in responder rates, and a high risk of bias. None of the three abstracts reports a GRADE rating.

The control in these trials is usually an active one. In the first randomised trial, by Rong and colleagues in China, 144 patients received taVNS or stimulation of a part of the ear without vagal supply; after eight weeks seizure frequency had fallen 42.6% with taVNS and 11.5% with the control, and the control group was then switched to taVNS. In the multicentre cMPsE02 trial, Bauer and colleagues compared 25 Hz stimulation (39 patients) with 1 Hz stimulation (37) over 20 weeks. Mean seizure frequency fell 23.4% at 25 Hz and rose 2.9% at 1 Hz by the end of treatment, a difference that was not significant, and responder rates were similar; the authors attributed the result partly to the higher stimulation intensity in the 1 Hz group.

The four-centre Chinese trial by Yang and colleagues, published in 2023, used the same comparison: 100 patients received 25 Hz stimulation at the highest intensity they could tolerate, two hours a day, and 50 received 1 Hz stimulation at the lowest intensity they could feel, with their drug doses held constant. At 20 weeks 44.7% of the active group and 16.7% of the control group had their seizures halved, a significant difference; 121 of 150 patients finished treatment. In a 2026 crossover pilot at Samsung Medical Center in Seoul, Hong and colleagues gave 14 patients with drug-resistant focal epilepsy eight weeks of active and eight weeks of sham taVNS in random order, and found mean seizure reductions of 46% with active and 49% with sham stimulation. Ear acupuncture covers needling at the same part of the ear.

A sham-controlled trial of thread embedding

A 2023 trial by Nguyen and colleagues, from the University of Medicine and Pharmacy at Ho Chi Minh City and Anhui University of Chinese Medicine, randomised 54 outpatients with drug-resistant epilepsy to four sessions of thread embedding or sham embedding four weeks apart, with patients and the assessing neurologist blinded. The threads were placed at GV20 (Baihui), GV14 (Dazhui), BL15 (Xinshu), BL18 (Ganshu), ST40 (Fenglong) and GB34 (Yanglingquan). At the follow-up visit four weeks after the last session, seizures had fallen by at least half in 10 of 27 patients (37%) in the embedding group and in 1 of 27 (3.7%) with sham. The authors asked for longer follow-up to confirm the result.

Seizures and harms reported during treatment

The Cochrane review found no excess of adverse events with acupuncture over the control treatments, but several of its comparisons rested on trials that did not report adverse effects, and the authors wrote that it remains uncertain whether acupuncture is effective and safe in epilepsy. In the cMPsE02 taVNS trial, adverse events were mostly mild or moderate (headache, ear pain, skin redness under the electrode, vertigo, fatigue and nausea), and four serious adverse events included one sudden unexpected death in epilepsy (SUDEP) in the 1 Hz group, which the investigators assessed as unrelated to treatment. Yang's trial reported pain, sleep disturbance, flu-like symptoms and local skin discomfort, no severe adverse events in either group, and excluded anyone who had had status epilepticus, a prolonged or repeated seizure without recovery between, in the previous year. In the thread-embedding trial, every patient in both groups had mild pain at the needle sites that passed within 15 minutes, and seven had bruising that resolved within five days.

What guidelines say

NICE's guideline NG217 on the epilepsies in children, young people and adults, published in April 2022 and last updated in August 2026, does not mention acupuncture or transcutaneous stimulation. Its chapter on non-drug treatment covers the ketogenic diet, resective surgery and implanted VNS, which it recommends considering as an add-on to antiseizure medication when surgery is not suitable, while noting that there was no long-term robust evidence on VNS.

The International League Against Epilepsy (ILAE) list of guidelines and reports held 161 entries in October 2026, none with acupuncture, complementary therapy or vagus nerve stimulation in its title. The ILAE Surgical Therapies Commission's 2022 review of neurostimulation, by Touma and colleagues, covered implanted VNS, responsive neurostimulation and deep brain stimulation. Clinical guidelines compares how guideline bodies treat acupuncture across conditions.

What the reviews could not settle

The needling reviews rest almost entirely on Chinese trials against drugs or added to them, so they cannot separate the needling from the attention and expectation that come with it; the only sham-controlled needle trial in the Cochrane review reported on 29 to 34 patients, and the thread-embedding trial 54. Seizure counts came from patient diaries, which the authors of Yang's trial named as a source of inaccuracy, and follow-up was short for a condition that waxes and wanes. In the taVNS trials the control usually delivers some current, which the cMPsE02 authors suggested may explain why their groups did not differ. Yang and colleagues asked for trials comparing taVNS with implanted VNS, and for longer treatment periods to test whether long use brings more adverse effects.

AcuiQ's epilepsy page lists the protocols individual studies prescribed, each with its citation.