Epilepsy

References & Evidence Notes

Each numbered entry below is either a source you can follow or a note setting out what the evidence does and does not support. Both are numbered together so the markers in the text line up.

Last reviewed 27 August 2026. Supplement entries are cross-checked against the NIH National Center for Complementary and Integrative Health and the Linus Pauling Institute Micronutrient Information Center.⁸ Two things matter more than anything else here. Never stop or change anti-seizure medication without your neurologist: abrupt withdrawal is a leading cause of status epilepticus, which is a life-threatening emergency. And a seizure lasting more than five minutes, or repeated seizures without recovery between them, means calling emergency services now. This page is unusual in one respect: the ketogenic diet is a genuine, long-established medical therapy for drug-resistant epilepsy, not an adjunct. That is exactly why it belongs with a ketogenic diet team and not a website, because it needs monitoring for growth, lipids, kidney stones, bone density and micronutrient status, and it interacts with medication. Anyone who also has diabetes needs that team from the outset: sustained ketosis alongside some diabetes medicines can tip into ketoacidosis at normal blood-sugar readings.

1. On the ketogenic diet as therapy: the classic ketogenic diet has been used for drug-resistant epilepsy since the 1920s, and randomized and controlled evidence supports it, particularly in children. Roughly half of children achieve a 50% or greater seizure reduction and a minority become seizure-free. The Cochrane review of ketogenic diets for drug-resistant epilepsy covers 13 studies and 932 participants, 711 of them children, and rates every included study at high risk of performance and detection bias because blinding is not possible with a diet (Martin-McGill KJ, et al. Cochrane Database Syst Rev. 2020;6:CD001903. PubMed 32588435). The modified Atkins diet and low glycaemic index treatment are less restrictive with broadly comparable results and are more practical in adults. These are prescribed and supervised therapies, initiated with a specialist team, and are not self-started. ↩
2. On withdrawal and status epilepticus. Abrupt discontinuation or missed doses of anti-seizure medication is among the commonest precipitants of status epilepticus. Status epilepticus is defined operationally by the ILAE at five minutes of continuous seizure activity, with a second threshold at thirty minutes beyond which long-term consequences including neuronal death become likely, PubMed 26336950, and outcome depends on how quickly it is treated. Planned withdrawal, when appropriate after a seizure-free period, is done slowly and under neurological supervision. ↩
3. On SUDEP: sudden unexpected death in epilepsy is a real risk, associated with poorly controlled generalised tonic-clonic seizures and with medication non-adherence. Discussing it is now recommended practice, because the main modifiable factor is seizure control itself. This is the strongest argument against experimenting with medication. ↩
4. On seizure triggers with a dietary or lifestyle component: sleep deprivation, alcohol, particularly withdrawal after binge drinking, and missed meals with hypoglycaemia are recognised precipitants, and a prospective study measuring sleep by diary and adherence by drug monitoring found these factors interact strongly rather than acting alone, PubMed 37413785. Photosensitivity affects a minority. Managing these is genuinely useful and requires no supplement. ↩
5. On supplements and interactions, where care is needed: enzyme-inducing anti-seizure medications reduce active vitamin D, which raises parathyroid hormone and increases bone resorption, and several of these drugs raise fracture risk, so fracture risk assessment belongs in routine care, PubMed 40889081. Valproate reduces active vitamin D by that route and by others, and carries a teratogenic risk now managed formally when prescribing to girls and women of childbearing age, PubMed 41562644, which is why folate status matters here. Ginkgo biloba has been associated with lowering the seizure threshold, and several herbal products interact with anti-seizure drug levels, including St John's wort, a strong enzyme inducer, so caution applies both when starting it and when stopping it, PubMed 32361205. The effect runs drug by drug rather than across the board: carbamazepine, which already induces its own metabolism, does not appear to be significantly affected, PubMed 22606944. Tell the neurology team before starting or stopping any herbal product. ↩
6. On vitamin B6: pyridoxine-dependent epilepsy is a rare genetic condition treated with pharmacological doses under specialist care and is not a reason for general B6 supplementation. High-dose B6 taken long-term causes a predominantly sensory, axonal peripheral neuropathy, PubMed 37447150. That review found patients report improvement after stopping, but the improvement is self-reported and the neurotoxicity at high intakes is not in doubt, so this is a supplement with a real ceiling. ↩
7. On pregnancy: anti-seizure medication in pregnancy is a specialist balance between fetal risk and seizure risk, and the drugs are not interchangeable on this point. In a population cohort, valproate and zonisamide were associated with several neurodevelopmental outcomes, hazard ratios ranging from 1.26 to 4.50, while levetiracetam and phenytoin were not associated with an increased risk of any of them, PubMed 41813017. Folic acid before conception is standard, but the dose is now genuinely contested and belongs with your neurologist rather than being decided alone. In a Nordic cohort of 3.4 million children, among mothers with epilepsy, prenatal exposure to high-dose folic acid was associated with cancer in the child, a hazard ratio of 2.7, with absolute risk rising from 0.6% to 1.4%, PubMed 36156660. No such signal appeared in children of mothers without epilepsy. A later study found a smaller increase in cancer risk in the women themselves, PubMed 39540679, and the finding has been disputed in the literature, PubMed 37452793. None of this is a reason to take no folic acid. It is a reason to agree the dose with the team managing your epilepsy. Stopping medication on discovering a pregnancy is dangerous and is not the answer. ↩
8. National Center for Complementary and Integrative Health (NIH), nccih.nih.gov, and the Linus Pauling Institute Micronutrient Information Center, lpi.oregonstate.edu/mic.