Guidelines for AEDs during pregnancy emphasise dangers of polytherapy and sodium valproate. I evaluate 1651 articles on AEDs and pregnancy since 1970, identified through a PubMed search. All studies are observational, none prospective, and conclusions are uncertain. Fetal malformations range from 0% to 4% in controls; 0% to 8% of untreated women with epilepsy; 0% to 30% on AEDs and 0% to 30% on sodium valproate. There is no separation of direct effects of epilepsy and co-morbidity; from AED use. Epilepsy severity, AED choice and dosage are not independent variables. There are significant contradictions: such as opposing results for monotherapy vs polytherapy in various trials; and for lamotrigine and facial clefts. Studies of cognition in offspring of mothers taking AEDs are flawed by ascertainment and expectation bias. Results are inappropriately pooled; despite variable study design, recruitment, definitions, ascertainment of malformations and cognition. Resources spent on pregnancy registers are not matched with monitoring of maternal safety. Seizure freedom in pregnant women is up to 20% lower than in others with epilepsy. There are reports of injury and SUDEP in women changing medications to reduce teratogenicity, but no systematic studies. The ratio of publications on AED teratogenesis to maternal morbidity of pregnant women is 100:1. This imbalance in information may result in pregnant women with epilepsy stopping, or using less effective, AEDs; with resultant morbidity and mortality. Without information on maternal safety, and with the uncertainties of published data on fetal outcome, it is doubtful that truly informed consent and sound decision-making for women is possible.
Statistics from Altmetric.com