Article Text

Download PDFPDF
Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism
  1. Victoria Smith1,
  2. Nick Brown2,3
  1. 1Department of Paediatrics, The Queen Elizabeth Hospital Kings Lynn NHS Trust, Kings Lynn, UK
  2. 2Department of Paediatrics, Salisbury NHS Foundation Trust, Salisbury, UK
  3. 3Aga Khan University, Karachi, Pakistan

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Design: Population-based cohort study.

Setting: National register of children born in Denmark from 1996 to 2006.

Patients: 655 615 children with an estimated time of conception over 11 years to December 2006; 2644 children exposed to antiepileptic drugs during pregnancy and 508 exposed to valproate. Follow up to autism spectrum diagnosis; death; emigration; or until 31 December 2010. Excluded Those with likely errors or missing values for gestational age, adopted children, and death <1 year old.

Risk factors: Parental age at conception, parental psychiatric history, gestational age, birth weight, sex, congenital malformations and parity.

Outcomes: Absolute risk (cumulative incidence) and the HR of autism spectrum disorder and childhood autism in children after exposure to valproate in pregnancy.

Main results

Of 655 615, 5437 were identified with autism spectrum disorder, including 2067 with childhood autism. The estimated absolute risk after 14 years of follow-up was 1.53% (95% CI 1.47% to 1.51%) for autism spectrum disorder and 0.48% (95% CI 0.46% to 0.51%) for childhood autism.

Overall, the 508 children exposed to valproate had an absolute risk of 4.42% (95% CI 2.59% to 7.46%) for autism spectrum disorder (adjusted HR 2.9 (95% CI 1.7 to 4.9)) and an absolute risk of 2.50% (95% CI 1.30% to 4.81%) for childhood autism (adjusted HR, 5.2 (95% CI 2.7 to 10.0)) (table 1).

Table 1

Association between any maternal use of valproate during pregnancy and autism spectrum disorder and childhood autism


Maternal use of valproate during pregnancy was associated with a significantly increased risk of autism in offspring, even after adjustment for parental psychiatric disease and epilepsy.

Abstracted from: OpenUrlCrossRefPubMedWeb of Science


This is an impressive analysis of a cohort drawing on robust national Danish datasets for valproate exposure (prescription register) and outcome (Psychiatric register). The survival model was appropriately adjusted for epilepsy and parental psychiatric illness, and all a number of perinatal variables. The authors were unable to exclude residual confounding by non-prescribed folic acid intake, but felt that this would have been likely to reduce rather than increase the observed effect size. They were also unable to rule out potential ascertainment bias in that children of women who had taken valproate in pregnancy were more likely to have been closely examined as a result of its known other side effects.

With these provisos, of particular interest was the constancy of HR by age of the children and the lack of risk associated with other antiepileptic drugs (AED). The effect was strongest for women taking valproate at conception but, though it diminished with time between cessation and pregnancy, remained significant even if stopped 6 months beforehand (HR 2.3, 95% CI 1.4 to 3.9).

The mechanism of action of valproate (a fatty acid derivative unlike other AEDs) is poorly understood. There is no doubt that valproate is an excellent anticonvulsant for a number of seizure types in children and, appropriately, used very widely. It does, however, have a poor record in pregnancy with regards to both teratogenicity and cognitive development in offspring. A recent major review showed consistent and significantly higher rates of malformation with valproate than other AEDs, and concludes that safer alternatives, such as lamotrigine or levetiracetam, should be used instead.1 More recent work suggests risk to neurodevelopment and cognition which may be independent of dose. A smaller study from Northern Ireland found an OR of 26.1 (95% CI 1.4 to 43.1, p<0.01) with valproate exposure, and poor neurodevelopment.2

Whatever the mechanism, this paper adds another argument for cessation of valproate well in advance of any potential conception. One could put forward a strong case for switching from valproate if at all possible in any girl above the age of 10 years.



  • Competing interests None.

  • Provenance and peer review Commissioned; internally peer reviewed.