A new study of almost 150,000 Canadian children is sure to further worry parents. Does taking antidepressants during pregnancy really increase the risk of your baby developing autism by 87%? As usual, the details matter. And the truth is never quite as clear-cut as the press release.
First: what we already know. Autism is a complex disorder of social behaviors and language. It’s unlikely that we’ll ever know “The Cause” of autism, because (like cancer and obesity and many other things) there’s probably not just One Cause. Certainly, genetics plays a huge role—siblings and other relatives of people with autism are at an increased risk for autism spectrum disorders. And we know that many prenatal factors (like premature delivery or maternal health problems) increase the risk, too. Some children with autism have a known genetic variant or marker, and the proportion of those kids seems to rise every year as our testing gets better. Still, there’s a whole lot we don’t know about other influences on how the brain develops, and on what other factors might contribute.
Previous studies on antidepressants taken during pregnancy have had mixed results—some gave shown an increased risk of autism, and others have not. One of the difficulties in doing such a study is that we know maternal depression, itself, is a risk factor for autism. So is it the antidepressant medicines that cause the risk, or the depression itself, or some other factor? For instance, depressed moms may get less sleep, or are perhaps more likely to drink alcohol, or may just feel more stressed—any of these, or any combination of these, could be what really confers the risk to the unborn baby.
In the current study, the authors tried to control for these effects by looking at multiple covariates. There’s some heavy-duty math behind these statistical techniques, but the idea is to isolate one independent variable (taking antidepressants) from every other variable, so you only “see” the effect of what you’re looking for. Of course, you have to know all of those other variables beforehand, and how to measure them, and how they might contribute to your end-point—and that’s why it’s tricky, and that’s why it’s easier to write about a press release than it is to read an actual study and muck your way through the details. In this study, the authors “controlled” for maternal health and mental health issues (though not specifically for maternal autism), substance abuse, income, education level, whether the family received welfare, and many other variables.
The authors used registries of births along with child and maternal health records from Quebec, looking at all babies born in 1998-2009. They excluded premature babies and twins (because these are already known predictors of autism), along with babies with other health problems. Then they looked at the child and maternal health records. To see if mom took antidepressants during pregnancy, pharmacy records were reviewed for filled prescriptions.
The numbers, themselves, are interesting. Of 145,456 live-born babies eligible, 1054 (.7%) were diagnosed with autism during the study period (typically within the first 6 years of life). Of those 145,456 babies, 2532 had moms who took antidepressants during the 2nd or 3rd trimester, including 31 who developed autism. When all of the math was done and covariates accounted for, the proportion of autism worked out to be 87% higher among the moms who took antidepressants after the 1st trimester (there was no increased risk during the 1st trimester, only during the 2nd and 3rd, or the last 6 months of pregnancy.)
That 87% increased risk—that pops, and that’s what you’ll see in the headline. But there are some other things to notice about the study that are more subtle, but just as important. Of the babies whose moms took antidepressants, 1.2% developed autism—and 98.8% did not. The relative increase in risk looks high, but the absolute, actual risk is still quite low. And the whole conclusion of the study rests on those 31 of the 150,000 babies who developed autism after exposure to antidepressants. That’s not a very big number.
I also wonder about controlling for known maternal risks. Perhaps more-severe depression itself confers a higher risk of autism than mild depression—and if more severely depressed moms were more likely than mildly depressed moms to take medication, that would explain the different observed rates. And maternal autism, itself—a friend of mine pointed out that the authors didn’t control for that, and didn’t even try to measure it. But if a higher proportion of moms with autism were depressed (which, logically, makes sense), that would also explain this association, without implicating the antidepressant medication itself.
What should parents do about this? Maternal depression can be debilitating, and needs to be treated—but perhaps non-medical therapy should be the best, first option, especially in the last 2/3rds of pregnancy. But if medication is needed to help mom, I think it’s still a reasonable option, given the small risks that could be implied by this study. It’s also clear from this study that antidepressants in the beginning of pregnancy, through the first trimester, are not associated with increased risks. Women on antidepressants probably don’t need to stop taking them while trying to become pregnant, or early on in the pregnancy.
If your child was exposed to these kinds of medications prior to birth, it would also be sensible to keep tabs on Junior’s development, and refer early on if there’s a concern of autism. Early therapy can be very effective.
People want a clear answer –“This” causes autism. It’s not that simple. This study supports the idea that early influences during pregnancy (such as maternal depression, or maternal medications) can be at least one risk factor. But there’s a whole lot more to learn.
Note: After I wrote this, someone sent me this link to NPR’s review of the same study – they did a good job, and provided links to previous research. They also made the point that the lead author of this study has worked with plaintiff’s attorneys who are suing antidepressant manufacturers. Should that color how we look at results like these? You decide.
This blog was originally posted on The Pediatric Insider.
© 2015 Roy Benaroch, MD
