Epilepsy is a common chronic neurologic condition, affecting 1–3 % of the general population with roughly equal male-to-female predominance. In the United States, estimates are that approximately three million people in the United States have epilepsy, with perhaps 500,000 being women of childbearing age [1]. As three to five births per 1000 are believed to be to a woman with epilepsy, epilepsy is one of the most common neurological conditions to be encountered in pregnancy [1]. The majority of people with epilepsy are well controlled, requiring one anticonvulsant to remain seizure-free. Many women with epilepsy have babies, but there are circumstances that are unique to women with epilepsy that obstetricians, neurologists, and other physicians involved in pregnancy care should be aware of.

Women with epilepsy have lower fertility rates than women without epilepsy. In part, this is related to social factors, such as reduced marriage rates seen in women with epilepsy. However, hormonal imbalances also play a part [2]. Women with epilepsy have higher rates of anovulatory cycles; this is more pronounced in women with temporal lobe onset seizures than in those with primary generalized epilepsies [2]. Furthermore, there are additional neurohormonal factors that lead to diminished libido in women with epilepsy, further decreasing sexual behavior and decreasing the possibility of conception [2]. Anticonvulsants themselves can further disrupt the neuroendocrine axis. For example, valproate is associated with the development of polycystic ovarian syndrome, leading to oligomenorrhea and anovulatory cycles [2].

As there may be a risk to the developing fetus from exposure to anticonvulsants in utero, some women may wish to consider a medication taper prior to conception. With a limited number of syndromic diagnoses, such as childhood absence epilepsy, the risk of relapse in adulthood after medication withdrawal is small, but with the majority of adult epilepsies, there is no foolproof way to determine whether or not a patient will remain seizure-free after withdrawal of anticonvulsant therapy. Thus, women must be willing to accept this uncertainty if they choose to come off of antiseizure medications. The risks accruing to the mother with breakthrough seizures can include physical injury due to trauma as well as social or employment harm due to loss of driving privileges. However, the risks to the developing fetus can also be significant, such as higher chances of preterm delivery and for being small for gestational age [3].

An area of active investigation is the effect of fetal anticonvulsant exposure, both on major congenital malformations and, equally importantly, on long-term development. Evaluating the structural risk of fetal anticonvulsant exposure is problematic for a number of reasons. Dose ranges tested on animals may exceed dose equivalencies in human, and malformations seen in animals may not predict human teratogenicity; for example, topiramate causes limb agenesis in rats but is known to be associated with an increased risk of orofacial clefts in humans.

Folate supplementation is a recommendation for all women of childbearing age. Women with epilepsy should not be exceptions to this rule and should receive supplemental folic acid prior to conception [4]. Women with epilepsy taking phenobarbital or phenytoin have lowered serum and/or red cell folate levels, which are inversely correlated with their anticonvulsant levels [5]. There is no data to suggest that folate administration is harmful to women with epilepsy.

While fetal anticonvulsant syndrome has been described for many decades, given the varied causes of epilepsy, it would be important to verify that it is the presence of anticonvulsants and not the epilepsy that is associated with the increased risk of congenital malformations. Holmes et al. looked at a large cohort of women at the time of delivery and placed their babies in one of three categories: those born to women taking anticonvulsant medications during pregnancy, those born to women who had a history of seizures but were not taking anticonvulsants during pregnancy, and those born to women without epilepsy who were not taking anticonvulsant therapy during pregnancy [6]. The babies were given careful physical examinations to identify major congenital malformations and other signs of presumed fetal anticonvulsant exposure such as distal digit and midface hypoplasia. They concluded that only the infants exposed to anticonvulsants in utero had an increased risk of major congenital malformations; those infants born to women with epilepsy not taking anticonvulsants had no greater frequency of malformations than those infants born to women without epilepsy and not taking anticonvulsants [6].

In reviewing the available data related to congenital malformations in anticonvulsant exposure, it appears to be conclusive that birth defects occur more often in children born to women taking anticonvulsants than to women who do not, with the absolute risk in the range of 4.6 % versus 2.8 % in one study [7]. When analyzed separately, the risk is higher with valproate as monotherapy or included as polytherapy. This finding has been replicated in multiple studies, particularly for the risk of valproate-associated midline craniosacral anomalies such as spina bifida, with this being a dose-dependent phenomenon [8]. Lowering effective serum concentrations, particularly in the first trimester, may decrease this risk. Carbamazepine appears also to be associated with spina bifida, though the absolute risk may be somewhat less than for valproate [9]. The other older anticonvulsants, phenytoin and phenobarbital, carry an increased risk of other midline defects including heart defects and craniosacral and facial anomalies [10].

With the newer agents, lamotrigine has been reported to have a 2.9 % incidence of major congenital malformations when used in monotherapy and 2.2 % for levetiracetam [11, 12]. However, topiramate carries a higher than expected risk of oral clefts with an absolute rate of 0.36 % in neonates exposed to topiramate compared with 0.07 % in children born to women without exposure to topiramate [13].

In the absence of major and obvious structural malformations, do children exposed to anticonvulsants in utero suffer from cognitive or behavioral aberrancies that may be related to the early exposure? Thanks to new cohort studies, we now know that the answer with certain anticonvulsants may be yes. The Neurodevelopmental Effects of Antiepileptic Drugs (NEAD) study is a prospective study monitoring a cohort of children born to women with epilepsy who took one of four anticonvulsants in monotherapy during pregnancy: phenytoin, carbamazepine, valproate, or lamotrigine. All mothers enrolled in the study had IQ testing performed, as this is a major predictor of child IQ, and data was maintained on their epilepsy types and presence or absence of seizures during pregnancy. The children were tested at regular intervals (ages 2, 3, and 4.5 years). The most significant finding, borne out over time, is that children exposed to valproate in utero had lower IQs than children exposed to the other medications, and this finding is dose dependent, suggesting that women who must take valproate for seizure control should be maintained on the lowest dose possible for seizure control [14]. Additionally, verbal skills were further impaired across all exposures compared to nonverbal skills, although the significance of this finding is unclear as there are additional reasons why this may have been present [14]. When examined through use of a retrospective questionnaire study, children exposed to valproate in monotherapy or as polypharmacy when in utero have a higher likelihood of requiring additional educational supports during school, perhaps related to these cognitive effects [15].

Other neurological syndromes may also be seen in association with anticonvulsant use, particularly with valproate. Exposure to this medication in utero has been shown to be associated with an absolute risk of autism of 2.95 % in children exposed to valproate in utero compared with 1.02 % in children not exposed to valproate [16]. Similarly, when looking at the absolute risk of autism spectrum disorder, which may include Asperger’s syndrome and other pervasive developmental disorders, the risk is 4.15 % versus 2.44 % [16].

Seizure control in the context of pregnancy can be variable and can be difficult to assess. Most of the data reported on seizures in pregnancy come from large cohort studies, many of which are observational and lack control groups. While there are a number of pregnancy registries around the world, some collect data both retrospectively and prospectively, while others only enroll women when they are pregnant and collect data prospectively. The large, multinational EURAP registry collects data from over 30 countries worldwide, so is not a population-based study. Some studies collect patients from very specialized epilepsy centers, where patients may be more refractory and thus require the services of a tertiary referral program, whereas other studies may collect a broader representation of epilepsy patients. These methodological factors lead to the wide differences in reported rates of, for example, seizure freedom in pregnancy.