Most women with pre-existing epilepsy have no change in the frequency of their seizures during pregnancy, but some 30 per cent have more frequent seizures. These are often women whose epilepsy has been hard to control at other times. Anticonvulsant plasma levels tend to fall in the later stages of pregnancy through increased volumes of distribution and rates of elimination: consensus guidelines recommend that plasma levels are monitored routinely and dosage adjusted to keep these steady. This approach is not entirely foolproof because most laboratories are unable to measure the changes in protein binding that also occur during pregnancy, increasing the availability of free drug. If the dosage is increased prophylactically during pregnancy then it is important to remember to reduce the dosage again in the postpartum period.

An equally important reason for deteriorating control of seizures is lack of adherence to anticonvulsant therapy because of fear of teratogenic effects. This is best addressed by counselling well before pregnancy,which should include a discussion of the risks of uncontrolled epilepsy to both mother and fetus, although it must be admitted that this is made difficult by the lack of quantitative data. As always, the aim of treatment is to control the epilepsy using a single anticonvulsant in the lowest effective dosage, and it is reasonable to try to reduce or withdraw anticonvulsants prior to pregnancy if there is a chance that the epilepsy may have remitted. It is unwise to attempt this during pregnancy itself.

All the anticonvulsant drugs are either known teratogens or of unknown safety in pregnancy. Epilepsy roughly doubles the risk of fetal malformation and most of this risk seems to be due to the treatment rather than the epilepsy or its cause. This translates into an absolute risk in the region of 5 to 10 per cent, although many of these malformations are minor. The risk is greatest in women taking two or more anticonvulsants. In the case of more serious abnormalities, phenytoin, barbiturates, carbamazepine, and sodium valproate all appear to be associated with facial clefts, cardiac septal defects, and a pattern of craniofacial and digital dysmorphism known as the fetal anticonvulsant syndrome. Sodium valproate and to a lesser extent carbamazepine appear to be associated with neural tube defects. Some of these defects may be secondary to drug-induced folate deficiency and it is good practice to offer folate supplements (5 mg daily) routinely to all potentially fertile women who are taking anticonvulsants, especially in the 3 months prior to and the first trimester of any planned pregnancy. Carbamazepine is traditionally regarded as the safest of the anticonvulsants for which we have reasonable data, but the teratogenic effects of sodium valproate appear to be dose dependent and dosages of less than 1 g daily may be equally safe. This is important when treating women with conditions such as juvenile myoclonic epilepsy which respond much better to valproate than carbamazepine. Several registers are currently collecting prospective data on the safety of the newer anticonvulsants (lamotrigine, gabapentin, vigabatrin, topiramate, tiagabine, etc.) but, at the time of writing, these must all be regarded as of unknown safety in pregnancy.

In addition to worries about teratogenic effects (which arise in the first 8 weeks of gestation), there are growing concerns about the potential for anticonvulsants to produce more subtle adverse effects on brain development and the subsequent behaviour and intelligence of the child. Such effects have been demonstrated in relation to anticonvulsant polytherapy including barbiturates, and it is possible that they may occur with other drugs. Prospective studies are in progress, but at the moment there is no clear guidance for women with epilepsy or their doctors in relation to the magnitude of these risks.

Carbamazepine, phenytoin, and the barbiturates accelerate vitamin K metabolism and increase haemorrhagic risks. Again it is considered to be good practice to offer the mother vitamin K supplements during the last month of pregnancy and to give the baby vitamin K at birth.

Epilepsy presenting for the first time in pregnancy requires investigation in the same way as adult-onset epilepsy in general. Idiopathic epilepsy that only occurs in pregnancy (so-called gestational epilepsy) is rare. Women presenting with serial seizures or status epilepticus are particularly likely to have an underlying secondary cause such as eclampsia, stroke, tumour, or encephalitis. Epilepsy during labour is usually either iatrogenic (for example, omission of normal anticonvulsant therapy) or again symptomatic of serious intracranial disease. Eclampsia is clearly the first consideration, but other possibilities include amniotic fluid embolism and cerebral venous thrombosis.

All the anticonvulsants pass into breast milk to some extent, but this need not prevent breastfeeding. Only the barbiturates occasionally cause problems with excessive sedation, but this small risk must be balanced against the problems of effectively withdrawing barbiturates by not breastfeeding. This can lead to the baby becoming irritable and jittery; impaired suckling and withdrawal seizures have also been reported.