ObjectiveTo determine whether children born to women who use anti-seizure medications (ASMs) during pregnancy have higher risk of autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD) independent of confounding factors.MethodsWe used Swedish-register data (n = 14,614 children born 1996–2011 and followed through 2013) to examine associations in children of women with epilepsy, using the largest sample to date and adjusting for a range of measured confounders. We examined maternal-reported first-trimester use of any ASM (22.7%); and the 3 most commonly reported individual drugs (valproic acid, 4.8%; lamotrigine, 6.8%; and carbamazepine, 9.7%). We identified ASD with ICD-10 diagnoses and ADHD with ICD-10 diagnoses or filled prescriptions of ADHD medication.ResultsExamination of individual drugs revealed that after adjustment for confounding, use of valproic acid was associated with ASD (Hazard Ratio = 2.30, 95% CI = 1.53–3.47) and ADHD (HR = 1.74, 95% CI = 1.28–2.38). Whereas a small, non-statistically significant association with ASD (HR = 1.25, 95% CI = 0.88–1.79) and ADHD (HR = 1.18, 95% CI = 0.91–1.52) remained for reported use of carbamazepine, confounding explained all of the associations with lamotrigine (HRASD = 0.86, 95% CI = 0.67–1.53; HRADHD = 1.01, 95% CI = 0.67–1.53).ConclusionsWe found no evidence of risk related to exposure to lamotrigine, whereas we observed elevated risk of ASD and ADHD related to maternal use of valproic acid. Associations with carbamazepine were weak and not statistically significant. Our findings add to a growing body of evidence that suggest that certain ASMs may be safer than others in pregnancy.

BackgroundPublished research on prescribed opioid analgesic (POA) use during pregnancy and birth outcomes is limited in scope and has not adequately adjusted for potential confounding factors. To help address these gaps, we estimated associations between maternal POAs during pregnancy and two adverse birth outcomes using a large population-based dataset, multiple definitions of POA exposure, and several methods to evaluate the influence of both measured and unmeasured confounding factors.Methods and findingsWe obtained data by linking information from several Swedish registers and conducted a retrospective cohort study on a population-based sample of 620,458 Swedish births occurring between 2007 and 2013 (48.6% female; 44.4% firstborn). We evaluated associations between prenatal POA exposure and risk for preterm birth (PTB; <37 gestational weeks) and small for gestational age (SGA; birth weight 2 standard deviations below the expected weight for gestational age or lower). We evaluated the influence of confounding by adjusting for a wide range of measured covariates while comparing exposed and unexposed infants. Additionally, we adjusted for unmeasured confounding factors by using several advanced epidemiological designs. Infants exposed to POAs anytime during pregnancy were at increased risk for PTB compared with unexposed infants (6.4% exposed versus 4.4% unexposed; adjusted odds ratio [OR] = 1.38, 95% confidence interval [CI] 1.31–1.45, p < 0.001). This association was attenuated when we compared POA-exposed infants with acetaminophen-exposed infants (OR = 1.18, 95% CI 1.07–1.30, p < 0.001), infants born to women who used POAs before pregnancy only (OR = 1.05, 95% CI 0.96–1.14, p = 0.27), and unexposed siblings (OR = 0.99, 95% CI 0.85–1.14, p = 0.92). We also evaluated associations with short-term versus persistent POA use during pregnancy and observed a similar pattern of results, although the magnitudes of associations with persistent exposure were larger than associations with any use or short-term use. Although short-term use was not associated with SGA (adjusted ORsingle-trimester = 0.95, 95% CI 0.87–1.04, p = 0.29), persistent use was associated with increased risk for SGA (adjusted ORmultiple-trimester = 1.40, 95% CI 1.17–1.67, p < 0.001) compared with unexposed infants. The association with persistent exposure was attenuated when we used alternative comparison groups (e.g., sibling comparison OR = 1.22, 95% CI 0.60–2.48, p = 0.58). Of note, our study had limitations, including potential bias from exposure misclassification, an inability to adjust for all sources of confounding, and uncertainty regarding generalizability to countries outside of Sweden.ConclusionsOur results suggested that observed associations between POA use during pregnancy and risk of PTB and SGA were largely due to unmeasured confounding factors, although we could not rule out small independent associations, particularly for persistent POA use during pregnancy.