Autism and Antidepressants: The biggest news isn't the study itself

Taking antidepressants in pregnant raises a woman's risk of having a child diagnosed with autism according to a new study published in JAMA. Huge study of 145,456 pregnancies in Quebec. Women who took anti-depressants during the second or third trimesters of pregnancy were 87% more likely to have a child diagnosed with autism. Those who used a specific class of antidepressants, selective serotonin re-uptake inhibitors (SSRIs) had more than twice the risk. The children of women who took more than one drug were four times as likely to have autism. A few other recent studies have found a link between antidepressants and autism, this is the largest one so far. Some have not shown a link. The evidence is mixed. 

Enough with that. On to the big news.

This is a big study, not for the results per se, but for how the media approached the sucker - with unusual caution and my goodness, a whole lot of much-needed nuance.

Just look at these risks in useful, simple terms:

The overall risk is low – less than 1 percent of the nearly 150,000 babies in the study were diagnosed with autism by age six or seven.

Only 31 babies, or 1 percent of the group whose mothers took antidepressants in later pregnancy, were later diagnosed with autism.

In the study, the overall rate of autism was 0.7%; that rate rose to 1.2% among women who took antidepressants in the second or third trimester. USA Today

In the population as a whole, 0.7% of the registered babies (1,054) later received an autism diagnosis. Among the 2,532 babies whose mothers took an SSRI during her second and/or third trimester of pregnancy, 31 infants (or 1.2%) would be diagnosed with autism some time in his or her first six years of life. LA Times

I even heard the words “relative risk” and “absolute risk” on the Today Show. The Today Show! This might be unprecedented. We are not often told how low risks are in absolute terms. It doesn't add to the drama.

Then there were reminders that correlation is not causation. From practically everyone. In strong language.

Does the new study prove antidepressants cause autism? No. Correlation is not causation, and science is complicated.” Boing Boing

“...correlation does not prove causation.” Natalie Azar, The Today Show.

Gail Saltz, a psychiatrist at The New York Presbyterian Hospital Weill-Cornell School of Medicine reported “correlation is not causation” on the NBC national broadcast last night.

She and others pointed out other factors might explain why antidepressant use in pregnancy is related to later autism risk.

In other words, why we might not want to assume anti-depressants caused autism:

"For example women who took antidepressants were more likely to have a history of psychiatric and other illness, were older, and we're more likely to have had another child with autism."

She wasn’t finished.

"Each of these factors could increase the risk of autism. So the study says there was a difference but can't say if the drug caused the difference or whether these differences are what increased the risk." NBC News

Nor was she alone. Others explained why we should be cautious in interpreting the study:

One shortcoming of the study is that it didn’t control for the severity of maternal depression, making it difficult to assess whether the increased autism risk might be tied to the underlying disease instead of the drugs used for treatment. It also didn’t account for the doses women were taking. Scientific American

One of the primary limitations of observational studies like these…no matter how carefully researchers aim to evaluate each risk factor, it’s often extremely difficult to tease them apart — particularly in this case in which autism spectrum disorder and depression share some of the same environmental and genetic risk factors. New York Times

The media nearly put out a primer on interpreting studies. The New York Times Motherlode article could have been pulled from a lecture in Research Methods 101:

And when it comes to studying pregnancy and antidepressants, observational research is the best we’ve got. Though randomized controlled trials are considered the gold standard in determining cause and effect, they are not an option in this field. It would be unethical for researchers to assign pregnant women randomly to one group or another that could potentially harm a baby. When controlled trials are unavailable (as they often are in issues of health) experts in this arena must rely on observational studies and hope that once enough of them show similar associations time after time, causal effects will start to emerge. New York Times’ Motherlode

I can practically see the nervous teaching assistant hovering at the back of the classroom.

Then there was the advice for pregnant women suffering from depression. There was much talk about the seriousness of depression. Many articles advised women not to ditch their antidepressants just yet. Usually this real world perspective (what the heck does this have to do with real people? what should we do now?) gets tacked on at the end of the article. It’s almost a throwaway line or two at the end with the standard "consult with your caregiver" bit. Not this time. No, the advice got extra special treatment:

There is no evidence to suggest a women should change her medication at this time, but it’s important to understand the risks and benefits of any medication, especially during pregnancy.” Autism Speaks

(Really? No evidence?  Wanna edit that remark?)

WebMD couldn’t wait and threw the advice into the subtitle.

But the risk is low, and it's important to treat depression in pregnant women, experts say 

Newsweek told women what to do in the first line:

Women who take antidepressants should seek out the advice of a physician if they wish to become pregnant, because new research suggests taking one popular class of the drugs—selective serotonin reuptake inhibitors (SSRIs)—during pregnancy can significantly increase a child’s risk for autism.

I'm nearly dumbfounded. For a couple reasons. One, this highly cautious tone is rare. It's as if we aren’t supposed to put much faith in this study! The media nearly bent over backwards telling us why we shouldn’t believe autism is caused by anti-depressants. Especially the smarter newsies like Science,  Slate or Wired.  We're told over and over not to worry, not to think too much about this study. And reminded that even if antidepressants were a trigger for autism, they only play a minimal role in comparison to a complex array of more significant factors, particularly genetic ones. 

It's not often we're reminded that a factor, whether a medication, breastfeeding or exercise, for example, is one of many and in reality, might only contribute a modest effect. When is the last time you heard breastfeeding only played a minor role in a child's overall health? That its effects were modest by comparison? Um, hardly ever. I could go on and on about the lack of control (of confounds) in the breastfeeding literature - a body of research that save some lab experiments with breastmilk in a petri dish is based on observational and thus correlational data. This study got taken to task on that front. 

I can't remember the last time I read a new study that I was urged to treat with so much caution.  Especially not a new study that also brought out experts on national television talking as if they were lecturing on statistics. From a scientific perspective, it is important not to place too much emphasis on a single study, even a large one, but view it in the context of other studies. So this caution is not misplaced, but coming from the media, it is surprising and I wonder how readers will interpret this study. It's an okay but not great observational study adding to a small yet growing pile of evidence linking antidepressants and autism (that needs further detangling). It has some limitations beyond what was described above like low incidence of autism compared to US and only single, full-term births. 

Oh, and there were some questions raised about the researchers' potential bias. See NPR, Slate, Science or Wired, they pretty much debunk this one. Slate calls it pseudoscience - come on, the data is from a national database. Everyone has come out of the labs, the press room and their home offices to tell why this study shouldn't merit so much attention. This doesn't happen every day! 

I've often said scientific studies should come with a warning label. Hey, folks, we finally got one! The esteemed Science even warned us in the headline: Reality check: Taking antidepressants while pregnant unlikely to double autism risk in children. Frankly that title makes it sound like this was a study providing evidence antidepressants don't increase the risk. That is not this study. If I were merely reading the headline I'd have come away with an entirely different study in mind. I wasn't as confused by Wired's still confusing headline: Taking Antidepressants While Pregnant May Raise Autism Risks, But Its Complicated. 

Oh yes, it is complicated. How true. 

Second, the amount of detail, the specific information about the risks, the limitations, the confounding factors, the nature of scientific studies and evidence –  it is really rare, perhaps unprecedented in the news coverage of a single freshly published study focused on children. So yes this one is ground breaking as Science Daily reported, but not for its findings but its coverage. 

I can’t think of another study that received this much attention and this much nuance and frankly debunking. It’s as if the media developed super critical scientific skills overnight. 

How did this happen? 

The parenting community is on edge after that infamous retracted autism study. It was the Perfect Parenting Storm. Does this explain the skepticism here? Was there a memo attached to this study urging caution in the media? There was an editorial accompanying the study and yet experts have urged caution in the past to little avail. If only this caution, this critical stance would spread to other children’s health and parenting studies. Let's see if it applies to other autism studies if not other discoveries. 

Okay, media, I know you can do this, please try it on the next study. Keep it up. Puleeeze.

No comments: