by Alison Bernstein
***Editor’s note: This guest post is by Alison Bernstein, a mom and scientist also known as “Mommy, PhD.” Here, she breaks down a recent study that has spurred headlines implying that taking antidepressants during pregnancy causes autism.***
A new study “Antidepressant Use During Pregnancy and the Risk of Autism Spectrum Disorder in Children” published in JAMA Pediatrics reports that use of antidepressants (ADs) during the second and third trimester of pregnancy nearly doubles the risk of autism. Does this study mean that pregnant women should toss out their Zoloft, Paxil and Celexa? Not so fast.
Letting depression go untreated poses many serious risks including preeclampsia, miscarriage, growth defects, and preterm birth; in cases of severe depression, suicide becomes a risk if depression is untreated. In the UK, maternal suicide is one of the top four causes of maternal mortality. Any decision about medication use during pregnancy is a delicate balance of risks and benefits that women should discuss with their doctors and loved ones.
The American College of Obstetricians and Gynecologists and the American Psychiatric Association have written recommendations and guidelines to assist in this complicated process. While this study may slightly shift the balance of risk and benefit, the study is far from a slam dunk for not using antidepressants during pregnancy. Dr. Bryan King, a pediatrician at Seattle Children’s Hospital, writes in an editorial accompanying this paper: “It makes no more sense to suggest that ADs should always be avoided than to say that they should never be stopped.”
Putting the risk in perspective. The study reports an 87% increase in risk of an autism diagnosis in children whose mothers took any type or amount of antidepressants during the second or third trimester of pregnancy. This means that a child whose mom filled a prescription is 0.87 times more likely to have a diagnosis of ASD. In real numbers, 4724 infants in the study were exposed to antidepressants, with 2532 exposed in the second or third trimester. Of those 2532, 31 were diagnosed this autism. This means that 12 more children were diagnosed with autism than expected if there was no association. That’s a lot less scary than an 87% increase or a doubling of risk.
To put this amount of risk in perspective, in the same dataset, advanced maternal age (35 and older) was associated with a 60% increase in risk and chronic or gestational hypertension was associated with a 58% increase in risk. A 2013 paper in the BMJ, with a much larger dataset (over half a million children), was large enough was able to consider depression separately from other diagnoses, comparing women who were depressed and took antidepressants with women who were depressed and did not, and even comparing two classes of antidepressant drugs. This study found a 50% increase in risk associated with maternal depression (whether antidepressants were taken or not) and a 3-fold increase in risk with taking antidepressants. However, they also calculated that this association accounted for only 0.6% of cases of autism. Overall, this is a small increased risk from antidepressants. While it remains important for individual women to discuss with their doctors, the role of antidepressant exposure in the larger context of autism prevalence is quite small.
Correlation does not automatically imply causation. This study is an observational study that, by design, is unable to tell us anything about causality. While it is not at all implausible that taking a neuroactive medication that can cross the placenta would affect neurodevelopment, the data in this study does not demonstrate causation. An important limitation of this study is that they were unable to actually consider if women actually took the medication; merely filling a prescription was used as a proxy for taking the medication. Unfortunately, this is a limitation of the type of database used for this study and is often the best data that scientists can get.
This study accounted for confounding factors that may also be associated with autism, including maternal socioeconomic characteristics, history of maternal chronic physical and psychiatric conditions, and infant characteristics. However, what this and other studies have been unable to address is that a woman with severe depression is more likely to stay on antidepressants during pregnancy. Even in the much larger BMJ paper discussed above, they were unable to quantify the severity of depression or the course of symptoms throughout pregnancy. Depression itself is a known risk factor for autism so this is an important piece of information that is missing from many of these studies. In an interview published on Research Gate, the study’s senior author, Dr. Anick Bérard, acknowledged this limitation.
“We could not measure maternal depression severity. However, we found that mothers who took more than one class of antidepressants during the 2nd/3rd trimester were more than four times more likely to have a child with autism – this group could be seen as more severe than those using only one class during the same window.”
This means that the association with autism could be due to the more severe depression, rather than the medication itself.
Considering that the risk of autism is largely genetic with considerable overlap of the genetic architecture of diagnoses like depression, bipolar, schizophrenia and ASD, another possible explanation is that there is shared genetic risk for severe depression and ASDs that is independent of antidepressants themselves. In short, detangling the factors at play in understanding the association between depression, antidepressant use and autism is extraordinarily difficult. Answering this question would require a randomized control trial, which is not ethical in this situation.
Statistical funny business. The final paragraph of the paper contains this sentence: “No adjustment was made for multiple comparisons; hence, we cannot rule out chance findings given the number of comparisons made.” This is a statistical no-no. If the analysis does not account for making multiple comparisons, statistical significance appears to be greater than it is. This does not necessarily mean the association is not real, but there is not enough information provided to know if the association would remain significant if multiple comparisons were accounted for. Given that there are simple, standard statistical tests to account for multiple comparisons, it is puzzling that the authors mentioned it but did not fix the statistics prior to publication.
Perpetuating misconceptions about autism. There is a persistent misconception that we are in the midst of an autism epidemic and that autism prevalence is on the rise. This creates room for the fear mongering about causes of this rise in autism and gives ammunition to the anti-vaccination movement and others who promote dangerous and ineffective treatments for autism. Reporting of this story and even the JAMA editorial perpetuates this myth. However, epidemiological studies show that while autism diagnosis has gone up in the past twenty years, this is largely due to changes in diagnostic criteria, awareness and the recategorization of intellectual disability. While autism may have gone unrecognized in the past, this does not indicate that the trait itself is increasing.
Putting this in context. Despite alarmist media reporting, this finding does not change much for the current guidelines for doctors treating pregnant women. Antidepressants are already in category C for pregnancy (animal studies show adverse effects on the fetus, but human data is unclear; potential benefits may outweigh these risks). The inherent limitations of this and similar studies make it unlikely that this study will change how antidepressants are categorized. While this study adds another small piece to our understanding of the risks of depression and antidepressant use during pregnancy, it does not warrant a massive change in clinical practice. Antidepressant use remains a complicated decision for which a woman must weigh the risks and benefits with her doctors and loved ones.