This population based study of 655,615 Danish children identified 5,437 as having autism spectrum disorder, and 2,067 with childhood autism. When examining the relationship of foetal valproate exposure to developmental outcome, increased risks of autism spectrum disorder (absolute risk, 4.42) and childhood autism (absolute risk, 2.50) were observed among the 508 children exposed to valproate.
These findings may be limited by inadequate information about use of folate, alcohol or illicit drugs during pregnancy, and possible missed psychiatric diagnoses in the parents. There was no adjustment in the analyses for prescriptions of other drug types. Furthermore, insufficient number of women took valproate only in late pregnancy to determine risk across trimesters. With regards to valporate dosing, there was no information on actual doses taken nor were there any measures of medication adherence or serum lelevs of valproate.
However, strengths of the study include that it was population-based, provided follow-up for a very large cohort for 14 years, had less than 3% loss to follow-up, and adjusted for confounding risk factors, thus reducing the risk of selection bias. This risk of autism adds to the other risks of foetal valproate exposure such as major congenital malformations and it is still unclear if there is a safe dose. In a related editorial, the author raises concerns that women of childbearing potential do not fully understand the treatment risks associated with valproate use in pregnancy and need to be made aware of alternative therapies.