Breastfeeding on epilepsy drugs didn't harm child development at age 6

Women with epilepsy are routinely steered away from breastfeeding on the theory that the antiseizure medicines in their milk might blunt a child's development. It is an intuitive fear and a consequential one, because it trades a known benefit — breastfeeding — against a hypothetical harm. A prospective cohort study published on 30 September tracked the children to age six and found the trade was based on a risk that did not materialise [s1].

The data come from MONEAD, a multicentre US study that enrolled pregnant women with epilepsy and healthy pregnant women between December 2012 and October 2016, then followed their children [s2]. The contemporary medicines these mothers were taking matter: the regimens were primarily levetiracetam and lamotrigine, the two drugs that have largely displaced the older agents whose reputations drove the original caution [s1]. This was a test of how today's prescribing plays out, not a rerun of the valproate era.

No gap where the fear predicted one

At age six, the researchers compared 273 children of women with epilepsy against 84 children of healthy women on two primary measures: a verbal index score and an adaptive-behaviour composite [s1]. The adjusted mean verbal scores were 107.93 against 107.18 — a difference of 0.75 (95% CI −2.40 to 3.89; P = .64) [s1]. Adaptive behaviour came out at 102.91 against 102.14, a difference of 0.78 (95% CI −2.52 to 4.07; P = .64) [s1]. Both confidence intervals sit squarely across zero, and both group means land within a couple of points of the test average of 100. On the outcomes the study was built to measure, the children of mothers who breastfed on antiseizure drugs were indistinguishable from the comparison group.

The study then looked within the epilepsy cohort, sorting children by how long they were breastfed: not at all, under six months, or six months and longer. Here the direction of the only significant finding is the opposite of the feared one. Children breastfed for six months or more had higher adaptive-behaviour scores than those not breastfed at all — an estimated 4.68-point advantage (95% CI 1.34 to 8.03; P = .01) — driven by the practical and conceptual subdomains [s1]. Shorter breastfeeding showed no such difference, and crucially, neurodevelopmental outcomes were comparable within each specific drug regimen, whether levetiracetam, lamotrigine, other monotherapy or polytherapy [s1].

Reading the signal honestly

The six-month advantage is the part most likely to be over-read, so it is worth being careful. This is an observational association, not a dose of medicine; mothers who breastfeed longest differ from those who do not in ways — education, support, household circumstances — that also shape a child's development, and no adjustment fully erases that. The finding is consistent with the broad literature that breastfeeding tracks with modestly better outcomes in the general population, and the honest reading is not that the milk improved the children's brains but that taking antiseizure medicine while breastfeeding did not cost them anything measurable. The benefit held in the pooled monotherapy group (a 5.88-point difference, 95% CI 2.21 to 9.55; P = .001) but not in polytherapy, where numbers were smaller [s1].

The limits are the familiar ones for a cohort of this kind. The sample is modest, the follow-up stops at six years, and the medicines studied are the current mainstream ones — so the reassurance does not automatically extend to drugs used rarely or to combinations the cohort could not populate. Epilepsy affects roughly 50 million people worldwide, a large share of them women of reproductive age, which is why a question about breastfeeding and routine medication reaches well beyond any single clinic [s3].

What makes this study more useful than its size suggests is its design. It was registered and run prospectively, enrolling women during pregnancy and following their children forward rather than reconstructing exposures after the fact, which removes a major source of recall bias [s2]. It also compared children against a group born to healthy mothers, so the epilepsy cohort was measured against a real-world benchmark rather than a test manual's average. Those features are why the flat primary result — no detectable gap in language or adaptive behaviour — is harder to wave away as an artefact of a small or selectively assembled sample.

What changes at the bedside

For a woman stable on levetiracetam or lamotrigine, the practical upshot is that the default advice to avoid breastfeeding is not supported by the best longitudinal data available. The decision moves back to where it belongs — the benefits of breastfeeding, the mother's own health and preference, the specifics of her regimen — rather than being pre-empted by a developmental risk that this study, at six years, could not find. As with any observational result, the right posture is confidence tempered by its design: strong enough to counsel with, not strong enough to promise with.

Sources

• Breastfeeding and Neurodevelopmental Outcomes in Children of Women Taking Antiseizure Medications — JAMA Network Open, 30 September 2026
• MONEAD study registration (NCT01730170) — ClinicalTrials.gov
• Epilepsy (fact sheet) — World Health Organization