The review included 51 studies. Participant numbers ranged from one to 2.5 million; the total number included was unclear as some studies cited only selected populations. Of the AED studies, 22 cohort studies reported on malformations and 16 studies (n=10,309 participants) quoted malformation rates for comparison between drugs. Ten cohort studies reported on pregnancy and neonatal outcomes. Thirteen studies (one case series and 12 cohort studies, published from 2004 onwards) reported on neurodevelopmental outcomes. Twelve studies (one case report, four case series and seven cohort studies) reported on outcomes with lithium carbonate. Most studies of AEDs were cohort studies. Studies on lithium carbonate included one case report, four case series and seven cohort studies (six were mentioned in the text, but the table showed seven). More than half of the studies were prospective cohort studies with a control group.
In studies that allowed comparison between drugs, there was a consistent finding of higher malformation rates with sodium valproate (4.2% to 20.3%, 12 studies) than with carbamazepine (zero to 8.2%, 11 studies) and lamotrigine (1% to 3.2%, four studies). In four studies, risk of malformation with sodium valproate was further increased with doses higher than 1,000mg.
Consistent findings for perinatal outcomes included reduced head circumference with carbamazepine (four studies) as well as reduced birth weight and length (two studies). A significant association between sodium valproate and neonatal hypoglycaemia was found in one study.
Eleven studies examined the effect of sodium valproate on child development and all found an association with poorer developmental outcomes (such as global reduction in IQ, lower verbal IQ, autism, memory and attention). In studies of dose effects, doses of sodium valproate greater than 1,000mg and polytherapy were associated with poorer neurodevelopmental outcomes (two studies). No association was seen between carbamazepine or lamotrigine and developmental outcomes.
Evidence on effects of lithium exposure in pregnancy was more limited. There was a trend of increase in cardiovascular malformations with lithium, specifically with an increase in Epstein’s anomaly (seven studies). Data from the table also suggest a high rate of stillbirths (two studies). Lithium exposure was associated with prematurity (one study) and increased birth weight (two studies). Information on developmental outcomes was limited.