Is This the Final Word on Prenatal Antidepressants and Autism?

William T. Basco, Jr., MD, MS


June 27, 2017

Fetal Exposure to Antidepressants and Autism

One hypothesis for the origin of autism spectrum disorder (ASD) is fetal exposure to excess serotonin, a theory that has some supportive evidence in mouse models.[1] Human cohort studies have raised the possibility that maternal use of serotonergic antidepressants in the first trimester may be associated with greater risk for ASD in offspring, but the findings are mixed.[1] One large study which used a sibling design was able to control for a number of potentially confounding factors but failed to show an association.[2] Two recent studies[1,3] examined the possible link between antidepressants in pregnancy and ASD in offspring.

Prenatal Serotonergic Use and ASD

Brown and colleagues[1] sought to address concerns of confounding in previous studies of serotonergic exposure during pregnancy and ASD. The data were from Ontario, Canada, where multiple medical and social databases were linked. Study children were all singleton births from 2002 to 2010. They were followed until 2014 to identify those diagnosed with ASD. The investigators used 500 covariates (eg, previous medical diagnoses, previous medical visits and hospitalizations, maternal- and child-specific demographic and socioeconomic variables) to calculate propensity scores for mothers' use of serotonergic treatments, allowing adjustment for the non-random assignment of mothers to serotonergic treatment or no serotonergic treatment.

The analyses included discordant siblings, in which one sibling was exposed to serotonergic antidepressants because of maternal use during pregnancy and the other was not. The analysis included data on more than 35,000 children, of whom 50.4% were male. Among mothers, 7.9% took a serotonergic drug at some point during pregnancy. Those who took a serotonergic medication had more severe previous psychiatric histories. The investigators were able to identify 620 sibling pairs in which one child was exposed and the other was not.

Among all children, 1.1% were diagnosed with ASD after a mean follow-up of 4.9 years. In multivariable analyses, there appeared to be an association between serotonergic medication exposure and ASD, with a hazard ratio (HR) of 1.59, with a 95% confidence interval (CI) of 1.17-2.17. However, this association was not significant after the full propensity score adjusted model was applied (HR 1.61; 95% CI, 0.997-2.59).

With respect to siblings, the incidence was 3.4/1000 person-years in the exposed siblings compared with 2.05/1000 person-years in unexposed siblings, but this difference was not statistically significant (95% CI for the difference between groups, 0.69-3.74) in fully adjusted models. The analysis also found that the offspring of mothers treated with serotonergic antidepressants in the year before pregnancy (but not during pregnancy) were more likely to have ASD (HR 1.85; 95% CI, 1.37-2.51), suggesting that confounding by indication is the cause of noted associations.

The investigators concluded that the previously reported associations may be explained by other factors, and their analysis using propensity scoring did not show an association between exposure to serotonergic agents during pregnancy and the development of ASD.

Maternal Antidepressants and Infant Outcomes

Sujan and colleagues[3] reported their findings from a cohort of Swedish children born in 1996 to 2012, comparing multiple developmental and behavioral outcomes among those whose mothers took antidepressants during the first trimester with the nonexposed general public and siblings. They merged multiple national databases to link pregnancy, birth, and childhood outcomes to produce the dataset. The analysis was conducted using "any antidepressant" exposure, but the investigators also performed sub-analyses focusing specifically on serotonergic antidepressants. The analyses controlled for year of birth, parity, maternal age at birth, educational level, parental criminal conviction, severe psychiatric illness, or a history of suicide attempts. The data included more than 22,000 offspring who were exposed to antidepressants.

Across the population, the rate of ASD diagnosis in the offspring of mothers who took antidepressants in the first trimester (5.28%) was more than twice as high as that found in nonexposed children (2.14%). These offspring also were more than twice as likely to have attention-deficit/hyperactivity disorder (ADHD). In adjusted models, first-trimester antidepressant use was associated with ASD (HR, 1.64; 95% CI, 1.46-1.83) and ADHD (HR, 1.58; 95% CI, 1.46-1.71). However, in the comparisons of exposed and not-exposed siblings, there was no difference in the risk for ASD (HR, 0.82; 95% CI, 0.62-1.13) or ADHD (HR, 0.99; 95% CI, 0.79-1.25). An increased risk of similar magnitude was observed among the offspring of women treated with antidepressants before pregnancy but not during pregnancy.

Finally, paternal use of antidepressants during the mother's first trimester was also associated with greater frequencies of the measured outcomes, again suggesting that an environmental influence rather than direct drug effects may be responsible for these associations.

The investigators concluded that their findings do not point to an increased risk for ASD or ADHD associated with antidepressant use. Rather, the risk is likely associated with genetic factors and the child's familial environment, suggesting that confounding is responsible for the apparent population-level association between antidepressant use and ASD or ADHD.


Both studies used sibling analyses and comparisons of mothers treated before pregnancy with untreated mothers to try to isolate the effect of the drug from genetic and environmental effects. Although the Swedish study did not use propensity scoring, the authors of the Canadian study commented that propensity scoring, although extensively used in their study, still can't account for uncontrolled, unknown factors. An accompanying editorial[4] reviews just how difficult it is to isolate the effects on childhood outcomes of a prenatal drug versus the maternal, paternal, and environmental confounders associated with depression. However, these studies, based primarily on evaluations of exposed- and nonexposed siblings, should provide reassurance. Instead we should focus more on the total picture of what maternal depression during pregnancy may bode for offspring.[4]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: