CLINICAL QUESTION

What are the associated rates of autism diagnosis after prenatal exposure to topiramate, valproate, or lamotrigine?

BOTTOM LINE

The use of lamotrigine to treat seizure disorders in pregnancy is not associated with an increase in the diagnosis of autism among offspring compared with no treatment during pregnancy. Topiramate is associated with an increase of about 2 percentage points, and valproate is associated with an increase of more than 6 percentage points. (Level of Evidence = 2b)

SYNOPSIS

The authors of this study identified pregnancy cohorts in the Medicaid Analytic eXtract–Transformed Medicaid Statistical Information System Files 2000–2018 and the Merative Market-Scan Commercial Claims and Encounters Database 2003–2020. These databases contain demographic information, diagnoses, medical procedures performed, and outpatient medications dispensed. The study cohort (N = 28,952) was restricted to pregnant patients with an epilepsy diagnosis who were linked with their offspring. The authors excluded children with chromosomal anomalies and major congenital malformations. The key period of exposure to these medications was in the second half of pregnancy (19 weeks to birth) when synaptogenesis is high. The unexposed group (8,815 patients) had no dispensing of seizure medications from 90 days before their last menstrual period through delivery. Clinical diagnosis of autism spectrum disorder by 8 years of age was based on claims data. Among individuals given an epilepsy diagnosis and at least one medication included in the study, 1,030 received topiramate, 800 valproate, and 4,205 lamotrigine. The cumulative incidence of autism was 1.9% for children with no exposure to antiseizure medications in the full population. For children whose mother had a diagnosis of a seizure disorder, the rate of autism diagnosis was 4.2% in unexposed children, 4.1% in those exposed to lamotrigine, 6.2% in those exposed to topiramate, and 10.5% in those exposed to valproate. Secondary analyses regarding topiramate showed no gradient based on dosage or duration of exposure or analysis of any diagnosis other than seizure disorder (similar to lamotrigine). Valproate did show a dose-response gradient, although without precise estimates.