While the pre-eminent causes of autism are genetic, environmental factors are still likely to be influential. In 2013 a large Danish population-based study reported that the cumulative incidence of all autism spectrum disorders over a 10-year period in children born to mothers taking valproate during pregnancy was significantly increased, from 1.53% to 4.42%.[1] In mothers with epilepsy the risk was increased from 2.44% to 4.15%. Analyses of the subgroup of children with childhood autism showed a similarly increased risk. There was no comparable increase in children born to mothers taking other anticonvulsants, or to those who discontinued valproate before pregnancy. The effect of valproate was unrelated to the dose, polytherapy, whether it was taken throughout pregnancy or started during pregnancy (but numbers of the latter were low), and remained after adjustments for congenital malformations and parental psychiatric disorders.

The study relied on national databases, so precise details about therapy such as dosage, other drugs including alcohol, and independent corroboration of the diagnosis were not available. Like many similar studies, the absolute numbers in some subgroups were still quite low. Experience in the UK National Childhood Encephalopathy Study showed that rechecking the clinical diagnoses in these small groups can be important. In addition, the prevalence by birth year increased steadily from 0.2% to 1.2% with age, so the comparisons had to take this into account. Furthermore, IQ was not assessed as a possible confounding factor, which could be important in view of the suggested link with learning difficulties. Nonetheless, it is the most robust data set so far and supports previous studies with lower levels of evidence that described the same risk association.

In contrast, there is also some evidence that folate started before conception could have a protective effect. In 2012 a case-control study in the USA found that greater than 600 μg daily in the first month of pregnancy was associated with a reduced risk of autism spectrum disorder in mothers or children with certain MTHFR genotypes.[2] In 2013 the Norwegian Mother and Child Cohort Study[2] looked at a national group with formally confirmed diagnoses of autism and found that childhood autism occurred in 0.21% of those whose mothers did not take folate, significantly more than the 0.10% in those who took 400 μg or more daily over the 4 weeks before conception and 8 weeks after.[3] This effect was not present for folate use in mid-pregnancy, nor for fish oil consumption. There was no association with other autism spectrum disorders but there was more limited power to detect an effect in these groups. While the cohort is not fully representative of the Norwegian population, analysis of a nationwide data file showed the same association. Nonetheless, effects from confounders such as socio-economic status, and non-response bias, could not be completely ruled out. Again, IQ was not assessed. The same group previously reported an association between peri-conception folate treatment and a reduced risk of severe language delay at the age of 3 years. However, it is unclear why the benefit should occur so early in fetal development. In addition there is no obvious difference in autism prevalence between countries with and without routine folate supplementation of food, or before and after its introduction, even though the latter appears to reduce the risk of neural tube defects. Dosage may be important, since in the USA the effective mean daily intake from fortified food is around 100 μg.[2]

This intriguing data only reveal associations, not proof of cause. In Denmark 85% of mothers taking valproate had epilepsy but in other countries such as the USA the majority of valproate treatment is for other indications, such as migraine or bipolar disorder. When considering valproate therapy in adult and adolescent females of child-bearing age, or younger females who are likely to need to continue treatment into their child-bearing years, it seems best to advise that there is a clear association with some congenital malformations, especially neural tube defects, and some evidence of links with learning difficulties and now autism in their offspring. Hence, if possible, an equally effective alternative would be preferable. The next questions are whether younger females who are already stably maintained on valproate should be switched to an alternative medication once they pass through puberty, or if valproate is prescribed, should folate supplementation be routinely given as well and if so, in what dose? As usual, more studies are needed before we can give our patients clear guidance.


  1. Top of page
  2. References
  • 1
    Christensen J, Grønborg TK, Sørensen MJ, et al. Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism. JAMA 2013; 309: 1696703.
  • 2
    Schmidt RJ, Tancredi DJ, Ozonoff S, et al. Maternal periconceptional folic acid intake and risk of autism spectrum disorders and developmental delay in the CHARGE (Childhood Autism Risks from Genetics and Environment) case–control study. Am J Clin Nutr 2012; 96: 809.
  • 3
    Suren P, Roth C, Bresnahan M, et al. Association between maternal use of folic acid supplements and risk of autism spectrum disorders in children. JAMA 2013; 309: 5707.