Prospective assessment of autism traits in children exposed to antiepileptic drugs during pregnancy

Amanda Wood, Caroline Nadebaum, Vicki A Anderson, David C Reutens, Terence J. O'brien, Frank Vajda

Research output: Contribution to journalArticlepeer-review

51 Citations (Scopus)
325 Downloads (Pure)


The association between autism spectrum disorders (ASDs) and prenatal anticonvulsant exposure is increasingly investigated, but comprehensive, blinded assessment using a validated instrument for autism within a well-characterized prospective cohort has not been conducted. Thus, existing studies may represent an underestimate of the true risk. Herein we present a prospective cohort study in children exposed to anticonvulsants during pregnancy, with all assessments conducted by examiners who were blinded to drug-exposure status.

Participants were 105 Australian children aged 6–8 years who were recruited via the Australian Pregnancy Register for Women on Antiepileptic Medication. Maternal epilepsy, pregnancy, and medical history data were obtained prospectively. Autism traits were assessed using the Childhood Autism Rating Scale (CARS).

Eleven children (10.5%) had elevated CARS scores. Two were exposed to valproate monotherapy (2/26; 7.7%), two to carbamazepine monotherapy (2/34; 5.9%), and seven to valproate in polytherapy (7/15; 46.7%). Linear regression analysis showed that the mean valproate dose during pregnancy was a significant predictor of CARS scores after controlling for polytherapy, mean carbamazepine dose, folic acid use, seizures during pregnancy, tobacco and marijuana use, maternal intelligence quotient (IQ), and socioeconomic status. First trimester folic acid supplementation and marijuana use were also significant predictors of CARS scores.

Using direct assessment of children in our prospective study, we found an elevated rate of autism traits across the sample. The most important determinant of association with autistic traits was higher doses of sodium valproate exposure. The use of valproate in women who may become pregnant is now generally avoided; however, there are insufficient data regarding the risk of ASD with low-dose valproate. If this risk is no greater than with other antiepileptic drugs (AED)s, it may enable women with genetic generalized epilepsy to retain optimal seizure control as well as minimize harm to their unborn child.
Original languageEnglish
Pages (from-to)1047–1055
Issue number7
Early online date11 May 2015
Publication statusPublished - Jul 2015


Dive into the research topics of 'Prospective assessment of autism traits in children exposed to antiepileptic drugs during pregnancy'. Together they form a unique fingerprint.

Cite this