Women and Epilepsy: Valproate, Contraception and Pregnancy

Women and epilepsy: valproate, contraception and the Australian decision not to follow Britain

In 2024 the United Kingdom tightened its rules on sodium valproate so far that anybody under 55, male or female, now needs two specialists to independently agree before it can be prescribed or continued.

Australia looked at the same evidence and declined to follow.

That is not an oversight or a lag. The Therapeutic Goods Administration’s Advisory Committee on Medicines considered the matter and advised that the risk profile following maternal exposure had not materially changed since its 2018 review, and that the evidence was insufficient to justify prescribing restrictions. The TGA is not pursuing regulatory action to restrict prescribing.

And Australian neurologists have been unusually blunt in defending that position. Professor Ernie Somerville of Prince of Wales Hospital in Sydney called the UK restriction “extremely inappropriate,” and added: “It’s actually dangerous.” His argument is that the study triggering the UK change was not peer reviewed and has been found to contain errors, that later reviews did not reach the same conclusion, that the guidelines ignore how heavily malformation risk depends on dose, and that some patients could die without the only drug that works for them. Lisa Todd of Epilepsy Action Australia has opposed blanket bans on similar grounds, noting that switching medications can worsen seizures and contribute to drug-resistant epilepsy.

So Australian women with epilepsy are in an unusual position: a drug that is heavily restricted in Britain remains available here, for reasons that are genuinely defensible, in a situation where the risks are real and dose-dependent and the alternative — uncontrolled seizures — is also dangerous.

This guide sets out what is actually known, with the numbers, so that you can have the conversation properly. It covers valproate in pregnancy, the paternal exposure question that triggered the UK change, contraception interactions that catch people out, pregnancy planning, and the other issues specific to women with epilepsy.

Nothing here is a reason to stop or change a medication. Stopping valproate abruptly can cause status epilepticus. Every decision in this guide belongs with a neurologist.

Valproate in pregnancy: the actual numbers

The reason this drug is contentious is that it is both unusually effective and unusually teratogenic. Both halves are true and most coverage picks one.

Major congenital malformations, and the dose dependency that gets left out:

That is close to a tenfold difference across a dose threshold, and it is the single most important fact in this guide. A blanket statement that “valproate causes birth defects in about 10 per cent of pregnancies” conceals a risk that is modest at lower doses and extremely high at higher ones. It is also why Australian clinicians object to guidelines that treat the drug as uniformly dangerous regardless of dose.

No dose has been shown to be completely free of risk. The malformations reported include neural tube defects such as spina bifida and anencephaly, craniofacial defects, cardiac defects, limb defects and hypospadias.

Neurodevelopmental effects following maternal exposure, expressed as odds ratios:

• Intellectual disability: 2.4 to 4.5
• Autism spectrum disorder: 1.7 to 4.4
• ADHD: 1.4 to 1.8

These are relative risks rather than absolute ones, and the ranges are wide because the underlying studies differ in method and population. They are nonetheless substantial, and the neurodevelopmental risk is the part of the valproate picture that has strengthened most over the past decade.

And the folic acid correction, which matters because a great many people believe the opposite: folic acid has not been shown to reduce valproate-associated malformations. It is still recommended — at least 0.4 mg daily before conception and through pregnancy — for general reasons, but it does not neutralise valproate risk, and anybody who has been reassured on that basis has been misinformed.

The Australian regulatory position

Since 2018, the TGA and overseas regulators have recommended against valproate in women of childbearing potential unless other drugs are ineffective or not tolerated. Where it is prescribed, it should be initiated by a specialist experienced in epilepsy or bipolar disorder, and co-prescribed with effective contraception.

What Australia has not done is adopt the UK’s pregnancy prevention programme requirements or the 2024 two-specialist sign-off rule.

Where that leaves a woman with epilepsy in Australia. Valproate is available, and if you are of childbearing potential it should not be your first-line drug unless the alternatives have been tried and failed. Lamotrigine and levetiracetam are the preferred options, and both are PBS-listed for initial epilepsy treatment in this group — so cost is not a barrier to using them first.

Where valproate remains genuinely first-line, and this is the part that gets lost in the controversy: the genetic generalised epilepsies — absence seizures, myoclonic seizures and generalised tonic-clonic seizures. For some people with these syndromes, valproate is substantially more effective than the alternatives, and juvenile myoclonic epilepsy in particular can be difficult to control on anything else. That is the clinical situation the Australian objection to blanket restrictions is really about.

What to do with all of this. If you are a woman of childbearing potential taking valproate, the questions are: has lamotrigine or levetiracetam been tried; what dose of valproate am I on and where does it sit relative to 1,100 mg a day; what is my contraception and does it interact; and if I wanted to become pregnant, what would the plan be. Those four questions are a reasonable agenda for one appointment, and none of them require you to decide anything on the spot.

The paternal exposure question

This is what triggered the UK’s 2024 extension to men, and the Australian response to it is instructive.

The study. A Nordic retrospective cohort using birth registry data, published in 2025, found a higher rate of neurodevelopmental disorders in children of fathers who had taken valproate in the three months before conception, compared with lamotrigine or levetiracetam. The cumulative risk was estimated at 5 per cent for valproate versus 3 per cent for the comparators.

The Australian assessment. The TGA and the Epilepsy Society of Australia considered the findings unsubstantiated because of methodological problems. Later studies did not reproduce them. The European Medicines Agency has opened a signal procedure to resolve the conflicting results. The Australian Department of Health, Disability and Ageing described the manufacturer’s Product Information update as “precautionary” and noted there have been no reported events in Australia.

What the TGA nonetheless advises, on a precautionary basis, from 2025:

• Inform male patients of the potential risk
• Use effective contraception — for both partners — during treatment and for three months after stopping
• Consult a doctor before stopping contraception, if pregnancy occurs, or when fatherhood is planned
• Do not donate sperm during treatment or for three months after stopping

Reading this honestly. The paternal signal is contested, the Australian regulator thinks the evidence is weak, and the advice is precautionary rather than restrictive. If you are a man taking valproate and planning to father a child, raise it with your neurologist — and do not stop the drug on the basis of a news headline, because an uncontrolled seizure carries its own serious risks.

For context on the Australian adverse event picture: the Database of Adverse Event Notifications lists 170 reports from 1980 to August 2025 linking valproate or valproic acid with the term “congenital, familial and genetic disorders.” The Department notes that a report does not establish causation.

Contraception, which catches people out in both directions

This is the most practically consequential section for many women with epilepsy and it is routinely not explained.

Interactions run both ways, and that is what makes it confusing.

Enzyme-inducing antiseizure medications can reduce contraceptive effectiveness. Several commonly used drugs — carbamazepine, phenytoin, phenobarbitone, primidone, and topiramate at higher doses — accelerate the metabolism of hormonal contraceptives, which can make the pill, the patch, the ring and the implant less reliable. This is a genuine failure risk and many women are never told.

Valproate is different and runs the other way. Oestrogen-containing contraceptives may increase valproate clearance and lower plasma valproate levels — which means starting the pill can reduce seizure control in somebody stable on valproate. The Australian guidance is to monitor clinical response when an oestrogen-containing contraceptive is started. Intrauterine devices or progesterone-only pills are alternatives, depending on preference.

And lamotrigine has its own interaction. Oestrogen-containing contraceptives can substantially lower lamotrigine levels, and levels can rise again in the pill-free week, which produces a cycle of fluctuating seizure control and side effects. Many women on lamotrigine do better on a continuous regimen or a non-hormonal method, and this is worth raising specifically.

What to actually do:

Tell whoever prescribes your contraception which antiseizure medication you take, and tell your neurologist which contraception you use. These are frequently different clinicians and neither assumes the other has asked.

Ask directly: “does my epilepsy medication affect this contraception, and does this contraception affect my epilepsy medication?” Both halves.

Consider the options that sidestep the problem. A copper IUD has no hormonal interaction at all. Hormonal IUDs and the depot injection are less affected by enzyme induction than oral methods. These are reasonable first choices for a woman on an enzyme-inducing drug.

And if you are on an enzyme-inducing drug and using the pill, treat that as a conversation to have soon rather than at some point.

Planning a pregnancy

Plan it, if you possibly can. This is the single most useful sentence in the section. A pregnancy planned six months ahead allows medication to be reviewed and changed with the seizures kept stable; a pregnancy discovered at eight weeks does not, because the critical period for malformation has already passed and changing drugs mid-pregnancy carries its own risks.

What a pre-pregnancy review involves:

Reviewing the medication. The Australian guidance where somebody on valproate becomes pregnant is to consider switching to levetiracetam or lamotrigine, and if no alternative controls the seizures, to use the lowest dose that prevents them. Done in advance, that switch can be made slowly with monitoring. Done at eight weeks, it is a scramble.

Folic acid, at least 0.4 mg daily before conception and during pregnancy, with the honest caveat that it has not been shown to reduce valproate-specific malformations.

Establishing good seizure control before conception, because seizure freedom in the year before pregnancy is the best predictor of seizure freedom during it.

Discussing what happens if seizures worsen, which they do in a minority of pregnancies, partly because blood volume changes alter drug levels. Lamotrigine levels in particular can fall substantially during pregnancy and may need dose adjustment with monitoring.

And discussing the risks of untreated epilepsy in pregnancy honestly, because the framing is frequently one-sided. A generalised tonic-clonic seizure in pregnancy carries risks to both the pregnant person and the fetus, status epilepticus is dangerous, and SUDEP risk does not disappear during pregnancy. The decision is not between a risky drug and a safe alternative; it is between two sets of risks, and the right answer depends on the syndrome, the dose and the person.

For women with epilepsy and an intellectual disability, this conversation requires particular care around capacity, consent and the right to make reproductive decisions, and it should not be had about somebody rather than with them. See epilepsy and intellectual disability.

The other issues specific to women

Polycystic ovary syndrome. PCOS occurs at almost twice the rate in women taking valproate for epilepsy. If it is suspected, the Australian guidance is to consider an alternative medication and refer to an endocrinologist or gynaecologist. This is a significant and under-discussed consequence — weight gain, menstrual irregularity, hirsutism and fertility effects — and it is reversible in some women on changing drugs.

Bone health. Valproate therapy over more than 10 years is associated with lower bone mineral density and osteoporosis risk. Enzyme-inducing antiseizure medications also affect vitamin D metabolism and bone density. For a woman who will take these drugs for decades and then go through menopause, this compounds — and bone density monitoring and vitamin D are reasonable requests that are rarely offered.

Catamenial epilepsy. Seizures that cluster around the menstrual cycle, which is a recognised pattern affecting a meaningful minority of women. It is worth tracking seizures against the cycle, because where a pattern exists there are specific management options, and because a woman whose seizures cluster premenstrually has frequently been told it is coincidence.

Menopause, which can change seizure patterns in either direction and which is almost entirely unresearched in epilepsy. Worth raising rather than enduring.

Other valproate monitoring, for completeness: hepatotoxicity occurs in about 1 in 20,000 patients, with 95 per cent of cases in the first six months and higher risk in children aged three and under; asymptomatic liver enzyme elevation occurs in about 10 per cent, warranting a dose reduction consideration and repeat liver function tests after a month; and thrombocytopenia occurs in 12 to 18 per cent, usually mild and reversible on dose reduction.

And the mental health picture, which affects women with epilepsy disproportionately. An Epilepsy Action Australia survey found around 20 per cent of respondents reporting depression — roughly double the general population rate — about one in six reporting anxiety, and more than half rating their quality of life as low-to-average. Valproate is specifically not recommended for behavioural symptoms in dementia, where it does not help and increases serious adverse effects, but for bipolar disorder it remains a maintenance option where lithium cannot be used — and in pregnancy planning for bipolar disorder the calculus is different again, because stopping valproate can precipitate a relapse that raises self-harm and suicide risk. That decision needs psychiatry involved, not just neurology.