Antidepressants during pregnancy and autism in offspring: population based cohort study

Objectives To study the association between maternal use of antidepressants during pregnancy and autism spectrum disorder (ASD) in offspring. Design Observational prospective cohort study with regression methods, propensity score matching, sibling controls, and negative control comparison. Setting Stockholm County, Sweden. Participants 254 610 individuals aged 4-17, including 5378 with autism, living in Stockholm County in 2001-11 who were born to mothers who did not take antidepressants and did not have any psychiatric disorder, mothers who took antidepressants during pregnancy, or mothers with psychiatric disorders who did not take antidepressants during pregnancy. Maternal antidepressant use was recorded during first antenatal interview or determined from prescription records. Main outcome measure Offspring diagnosis of autism spectrum disorder, with and without intellectual disability. Results Of the 3342 children exposed to antidepressants during pregnancy, 4.1% (n=136) had a diagnosis of autism compared with a 2.9% prevalence (n=353) in 12 325 children not exposed to antidepressants whose mothers had a history of a psychiatric disorder (adjusted odds ratio 1.45, 95% confidence interval 1.13 to 1.85). Propensity score analysis led to similar results. The results of a sibling control analysis were in the same direction, although with wider confidence intervals. In a negative control comparison, there was no evidence of any increased risk of autism in children whose fathers were prescribed antidepressants during the mothers’ pregnancy (1.13, 0.68 to 1.88). In all analyses, the risk increase concerned only autism without intellectual disability. Conclusions The association between antidepressant use during pregnancy and autism, particularly autism without intellectual disability, might not solely be a byproduct of confounding. Study of the potential underlying biological mechanisms could help the understanding of modifiable mechanisms in the aetiology of autism. Importantly, the absolute risk of autism was small, and, hypothetically, if no pregnant women took antidepressants, the number of cases that could potentially be prevented would be small.

Outcome N, total N, matched exposed:unexposed N, with outcome N, exposed with outcome Balance was assessed using the standardized mean distance, and overall matching success was assessed using the ASAM (average standardized absolute mean distance), where the closer to 0 the ASAM is, the better the covariate balance is. The ASAM before matching was 0.093. The ASAM after matching was 0.022. No individual covariate had a standardized difference greater than 0.10. A graphical display of the balance of each covariate in the overall matched sample of N = 6,426 is provided below.

Sibling matching
Sibling matched analyses for ASD, ASD without ID, and ASD with ID were based on the following, respectively: 3,038 ASD cases, of whom 66 were exposed, and 94 sibling sets with discordant outcomes and discordant exposures; 2,408 ASD without ID cases, of whom 60 were exposed, and 81 sibling sets with discordant outcomes and discordant exposures; 630 ASD with ID cases, of whom 6 were exposed; 14 sibling sets with discordant outcomes and discordant exposures.

Missing data
In the study sample N = 254,610, there were missing data for the following characteristics: maternal age: N=5 (0.002%); paternal age: N=2284 (0.9%); family income quintile: N=117 (0.05%); maternal education: N=1395 (0.55%). These data were not imputed due to the small number missing. In Analysis 1, Model 3, missingness on these characteristics resulted in a sample loss of 1.5%. Missingness on these characteristics did not influence Analysis 2 because the propensity score estimation technique was able to incorporate the missingness in prediction. The sibling models in Analysis 3 did not adjust for these characteristics. In Analysis 4, missingness on these characteristics resulted in a sample loss of 1.0%. Therefore, missingness on these characteristics was not considered substantial.
Missingness was also observed for maternal smoking: N=34,249 (13.4%) and maternal BMI: N=52,893 (20.8%). Because smoking and BMI were shown to be not associated with ASD in our previous investigations in our study sample [4 5], these covariates were only included in the propensity score model. The boosted CART model as a non-parametric technique is able to incorporate missing values in prediction and as a result no sample loss due to missingness on smoking/BMI occurred in the propensity score analyses. Furthermore, since previously identified predictors of smoking/BMI missingness (e.g., birthyear) were also incorporated in the propensity score model; and because of the null associations of those characteristics with ASD, the missingness of these two characteristics is unlikely to meaningfully influence estimates. Table A: Regression-estimated odds ratios and 95% CI for associations of antidepressant use during pregnancy and ASD when comparing children exposed prenatally to antidepressants prescribed for a known psychiatric condition with children exposed to maternal psychiatric disorders but no antidepressants.
Outcome (N exposed cases)  * assuming that the elevated risk of ASD due to U is consistent in both the exposed and the unexposed ** the original Table 3 propensity-score matched estimate of the OR and 95% CI for exposure to antidepressants during pregnancy and risk of ASD without intellectual disability Odds ratios and 95% confidence intervals are calculated based on formulae provided by Lin et al., Biometrika 1998 http://www.jstor.org/stable/2533848