Antidepressants in Pregnancy: Autism Risk?

On December 14, 2015, Boukhris, Sheehy, Mottron, and Berard published a paper in JAMA Pediatrics which described their  study of the association between autism-spectrum disorders and exposure in utero to antidepressants.

They looked at all women who had pregnancies in Quebec between 1998 and 2009.  These 186 165 women have been followed prospectively in the Quebec Pregnancy Cohort (QPC).  The authors looked only at those infants born at full term; all antidepressant exposure was recorded, and there was up to 11 years of follow-up. 

They found that 3.2% of all infants born to this cohort were exposed to antidepressants in utero.  Of those infants who had this exposure, 46 later received a diagnosis of an autistic-spectrum disorder (1 %).  More specifically, among those exposed in the second or third trimester, there were 31 autistic-spectrum diagnoses (1.2 %).   Among those infants with no in utero antidepressant exposure, 1008 later received an autistic spectrum disorder diagnosis (0.7 %).

When the diagnoses were restricted to those made by a neurologist or psychiatrist, the findings remained positive, but with reduced statistical confidence.


Other factors, such as maternal history of psychiatric disorders, mother living alone, maternal gestational diabetes or hypertension, were also positively correlated with the infant later having an autistic-spectrum disorder diagnosis.

A major weakness of the study was that it did not have a detailed analysis or discussion of autistic-spectrum symptoms (even subsyndromally) in the parents or other relatives of the infants.  They state "although our sample size is large, the size decreased substantially in stratified analysis on family history of ASD, which led to decreased statistical power."  And they observe that "those using [antidepressants] were...more likely to have had another child with ASD than those not using ADs", but did not expand on this finding.      The most likely contributing factor towards autism-spectrum phenomena would be the presence of these same phenomena in the family, which would be heritable.   It is possible that the presence of this hereditary factor could have been an underlying cause for the heightened risk which they observed.   This same risk factor could theoretically have contributed to the mothers using antidepressants more frequently during the pregnancy.

Another weakness of the study concerns the use of the "autism spectrum disorder" diagnostic label.  The use of this label is more frequent nowadays.  The degree to which it is appropriately considered a "disorder" could be subject to debate about criteria or severity.  For example, with the "autism spectrum quotient" questionnaire,  criteria such as "I don't like reading fiction,"  "I am fascinated by numbers," and "I would rather go to a library than a party" increase the score towards an autism diagnosis!  It seems much more important to restrict such a label to some kind of marked social, behavioural, or communicative problem, rather than intellectual or recreational preferences.   Enjoyment of libraries should not lead to a DSM label! 

Another related confounding factor could be that someone who has taken antidepressants might be more likely to have their children assessed by someone able to "diagnose."   It could be that if an autism questionnaire was administered to every child in the cohort, then 1.2% (rather than 0.7%) of the entire cohort could meet some "autism spectrum" threshold.  Perhaps the increased incidence in the group who had taken antidepressants is simply due to these mothers having a higher likelihood of asking for their children to be assessed.   In order to determine whether this is true, we would have to ensure that every child in the cohort received the same type of assessment, including the same questionnaires (such as the "autism spectrum quotient").   But in this study, this was certainly not the case. 

While the authors find that SSRIs in particular were associated with increased ASD incidence, with minimal associations from other antidepressant classes, it is notable that there were far too few cases of  non-SSRI antidepressant exposure to make any reliable statement at all about non-SSRI antidepressants.   Therefore, they should remove the implied message that only SSRIs are involved with the association. 

Nevertheless, it is an important study with a very large cohort.  We must be vigilant about the possibility of risks for giving any medications, particularly in pregnancy.

If, in future studies, this difference in autism incidence is found to be directly caused by antidepressant exposure, the evidence here shows that the risk increases from 0.7% to 1.2%.  Posed differently, the probability of not having an autism-spectrum diagnosis changes from 99.3% to 98.8%.


In some cases, the risk of severe depression in pregnancy could outweigh the risk of a treatment causing harm, but this would need to be carefully evaluated and discussed.   A question to ask in this situation would be whether the antidepressant is specifically required for treatment of the depression.  In some cases, it might indeed be very helpful, with a much higher likelihood of depressive symptoms, and various adverse outcomes for mother and child, without it.  But in other cases, despite high depressive severity, the antidepressant might not clearly be an imperative component of the therapy.  Perhaps in some cases other treatment modalities could be sufficient, at least during the pregnancy.  It depends on the specific case or situation.  

I am bothered by the author's concluding remarks in their abstract; the remarks assert causation, despite the findings really being associative:  they say "use of antidepressants...increases the risk of ASD in children."  It would be more appropriate for them to have said, as they were more careful to do elsewhere in the body of the paper, that "there is an association between antidepressant exposure in utero and subsequent ASD diagnosis."

In any case, it is not controversial to assert that all possible non-medication strategies should be optimized in the treatment of depression, particularly in pregnancy.  This includes promotion of healthy lifestyle factors, careful attention to social and community support, and psychotherapy.