Research News

Valproate in pregnancy linked to autism in children

BMJ 2013; 346 doi: (Published 24 April 2013) Cite this as: BMJ 2013;346:f2602


More than 600 000 babies were born alive in Denmark between 1996 and 2006. The 508 babies whose mothers took valproate while pregnant had a higher risk of autism than others in analyses adjusted for maternal age, paternal age, parental psychiatric history, gestational age, birth weight, sex, congenital malformations, and parity. Most of their mothers had taken valproate in the first trimester (441/508) and most had epilepsy (85%; 432/508).

Over 14 years of follow-up, 4.4% of exposed children developed autism spectrum disorder (compared with 1.5% for the whole cohort; adjusted hazard ratio 2.9, 95% CI 1.7 to 4.9) and 2.5% of exposed children developed childhood autism (compared with 0.5% for the whole cohort; 5.2, 2.7 to 10.0). The association was evident in separate analyses confined to mothers with and without epilepsy. Other antiepileptic drugs were not associated with autism in this study, but there were only a handful of exposed cases in some of these analyses and confidence intervals were wide.

Valproate is a known teratogen that carries an absolute risk of congenital malformations of around 11%, says a linked editorial (p 1730). Previous studies have also reported an association between antenatal exposure and cognitive impairment in children. We should now add autism to the list of risks and counsel women accordingly, whether or not they are planning a pregnancy. Doctors should look harder for an alternative to valproate for all girls and women of childbearing age and prescribe the lowest effective dose if valproate is the only option. Pregnancies are often unplanned, so discussion of the risks should start early.


Cite this as: BMJ 2013;346:f2602