The alarm has been sounded: Antidepressants cause autism! Or so one could easily think based on a new study in JAMA Pediatrics. Four researchers in Quebec conclude that “the use of antidepressants, specifically selective serotonin reuptake inhibitors [SSRIs], during the second and/or third trimester increases the risk of [autism spectrum disorder] in children.” In a ResearchGate interview, study senior author and perinatal pharmacoepidemiologist Anick Bérard of the Université de Montréal and the CHU Sainte-Justine Research Centre firmly advocated avoiding antidepressant use during pregnancy: “Depression needs to be treated during pregnancy but with something other than antidepressants in the majority of cases. The risk/benefit ratio is clearly leaning towards no use.”
If true, this could be a discovery with implications for the much-debated “autism epidemic” and require serious reconsideration of depression treatment for pregnant women. Has a missing link been found?
Autism is such a hot-button issue that it’s dangerously easy to frighten and mislead people. Infamously, a fraudulent 1998 study purportedly linking autism to the measles-mumps-rubella vaccine stoked a mass panic that resulted in lower vaccination rates, lower herd immunity, and the return of measles. I recently investigated how the debunked pseudoscience of facilitated communication persists in large part by feeding on the desperate hopes of autistic children’s parents, some of whom are quietly shepherding such junk science into public schools. So it is not surprising that other researchers have stressed a note of caution about the new antidepressant study and its methods. Yet even these dissenting voices do not, I think, go far enough.
Study co-author Bérard, it turns out, has been criticized by a federal judge for cherry-picking results to link antidepressants to birth defects. The press should treat such studies with skepticism rather than leading with their findings. Sober pieces in Science, Wired, and NPR rightly questioned whether the study was significant and whether Bérard’s advocacy for stopping antidepressant usage during pregnancy was justified. In particular, Emily Underwood in Science wisely led off by writing “Many epidemiologists and psychiatrists say the study, published today in JAMA Pediatrics, is flawed and will cause unnecessary panic,” which is the most important point to make about this study. But too many journalists failed to make this point, and with autism research, such credulity is downright dangerous.
Researchers have already questioned the significance of the observed correlation. Aaron Carroll of Indiana University’s Center for Health Policy and Professionalism Research wrote at his blog, the Incidental Economist, in a post titled “The Panic Du Jour”:
You have to place things in context. … It’s only significant for SSRIs and in the second and/or third trimester, which is 22 kids total. The absolute risk increase was only 0.5%. There are limitations to the study, and other studies have found different results. My take home would be that this deserves more work and attention, and that any potential harms from the antidepressants should be weighed against the known benefits for these pregnant moms.
Echoing his caution, Bryan King, program director of the autism center at Seattle Children’s Hospital, told NPR’s Jon Hamilton, “My biggest concern is that it will be over-interpreted,” pointing out that even the increased risk was still small overall. (JAMA Pediatrics published a skeptical editorial by King about the study in the same issue.) Autism Speaks Director of Medical Research Paul Wang said, “This study suggests that there is some kind of connection between depression, prenatal development, and autism. But it doesn't prove that antidepressants cause autism. Antidepressants appear to decrease the chance of prematurity, which is itself a risk factor for autism.” And NPR cites a number of studies showing conflicting results over whether there is a link between antidepressants and autism, none of them conclusive. Oxford psychiatric epidemiologist Amir Sariaslan pointed out to me the real possibility of genetic confounding: “We know that both depression and autism are considerably heritable, and that the same genes are involved in the development of most common psychiatric disorders. If the same genes are involved with the development of maternal depression and offspring [autism spectrum disorder], and we neglect to account for them, we will observe an association between the two that is artificial, or confounded.”
In response to such critiques, Bérard told NPR, “We have to be vigilant even if the risk is small,” and told me in an email: “You need to consider other treatment options such as exercise or psychotherapy,” and again emphasized the supposed risk. This is misleading, however, since treatment is a matter of balancing competing risks, and the study runs the risk of playing up one unproven danger to the exclusion of far more established dangers—such as the impact of untreated or insufficiently treated depression. Bérard’s position smacks too much of banning liquids on airplanes and making people take off their shoes in airport lines—but with far worse potential consequences if women are persuaded to stop antidepressants that they genuinely need. Exercise and psychotherapy might be effective substitutes for some, but expectant mothers should make that decision without the unjustified specter of autism hanging over them.
Taking a closer look at the study itself, it overstates its case and does not justify its flat conclusion that “depression should be treated with other options (other than antidepressants) during this critical period.” Most concerning is Bérard’s polemical tone against antidepressant use in pregnancy. Bérard sloppily conflates SSRIs with antidepressants in general, in order to make a less justified leap away from pharmacological treatment of depression altogether. As Bérard told Wired’s Sarah Zhang, “The thinking is often that treatment only involves drugs and cannot involve anything else. I think the time has come to rethink this whole process.” There are legitimate concerns about medication overuse, but such comments also raise the question of whether her research paper is objective as it needs to be to warrant raising such a drastic conclusion as avoiding antidepressant (or even merely SSRI) use during pregnancy. It is premature to speculate on the cause for a relationship when the relationship has not yet even been established. It is especially irresponsible given the propensity toward panic around autism. Bérard slips too easily into an alarmist tone, which may very well exacerbate the autism-related panic that has already resulted in a new surge of measles cases and the embrace of pseudoscience.
In Science, Emily Underwood reports that Bérard “serves as a consultant for plaintiffs in litigations involving antidepressants and birth defects,” suggesting that she might not have approached the study with a disinterested attitude toward antidepressants. In 2014, she served as a plaintiffs’ expert witness in the Pennsylvania Zoloft birth defects lawsuit against Pfizer, until her testimony was excluded by Judge Cynthia M. Rufe on the grounds that Bérard’s methods were “not scientifically sound.” In her ruling, Rufe excoriates Bérard for 25 pages, writing:
The Court holds that Dr. Bérard’s opinion is not grounded in the methods and principles of science … [the report’s] methodology is not reliable or scientifically sound. … Dr. Bérard takes a position in this litigation which is contrary to the opinion she has expressed to her peers in the past, relies upon research that her peers do not recognize as supportive of her litigation opinion, and uses principles and methods which are not recognized by the relevant scientific community and are not subject to scientific verification. … Dr. Bérard’s opinions regarding Zoloft are only made possible by her departure from use of well-established epidemiological methods.
(According to the Massachusetts General Hospital Center for Women’s Mental Health, Paxil is the only antidepressant around that tentative concerns around birth defects have led to a Food and Drug Administration reclassification, while Prozac, an SSRI, is the best characterized antidepressant, and it shows no link to an increase in major birth defects based on available research.)
That’s most definitely not to say that concerns about possible pharmaceutical links to autism are illegitimate. The full extent of the effects of antidepressants during pregnancy are not sufficiently understood and should be further studied. Ben Goldacre’s Bad Pharma points out many serious concerns over the quality and objectivity of research supporting drug use. That’s why Bérard’s work is so frustrating and counterproductive, because it increases the amount of noise rather than signal. By injecting unreliable and tendentious commentary (and fear) into the conversation, she makes it harder to get to the clarity we desperately need around the impact of antidepressants during pregnancy.
Even if Bérard’s past involvement in lawsuits against antidepressant makers does not constitute a conflict of interest, Rufe’s conclusions paint Bérard as less a disinterested researcher and more as a polemical crusader. In an email, Bérard contested Rufe’s opinion: “Her report was based solely on the expert report that was given to her before we underwent trial. I believe that she had a bias towards the treatment of depression with antidepressants and considered that if depression was not treated with such medications, it was not treated at all.” While it is possible that Rufe made the wrong call, Bérard’s overreaching claims and alarmist tone tend to support Rufe’s view. Rufe’s concerns around Bérard’s testimony mandate, at the least, far closer scrutiny and skepticism of the autism study’s methods and conclusions.
Bérard also invokes the horrors of thalidomide-related birth defects, an unjustified and emotionally manipulative comparison. Her statement that “I would always be very cautious about saying that anything is ‘safe’ during pregnancy. We have to remember that thalidomide was labeled as ‘safe’ for use during pregnancy,” is a cynical recipe for panic. If nothing is safe, everything is dangerous. The question instead ought to involve assessing which dangers are real and significant. Eastern Michigan psychologist James Todd cautioned against giving too much weight to the study, telling me, “Autism prevalence studies have a half-life like artificial elements.”
In proposing a conclusive link with autism and advising diminished use of antidepressants during pregnancy, Bérard et al. are not just acting prematurely but irresponsibly, and some of the press has abetted them. In contrast to the skeptical reporting at Science, NPR, and Wired, pieces like those in the Washington Post, the Los Angeles Times, and Newsweek were far too credulous toward this single study and buried the caveats far into the text. In an age when many people don’t even get past the headline, this is not good enough. Leading with the debunking, as Science and NPR do, is a wiser move than leading with the scaremongering. I call out these other outlets by name because the lesson of anti-vaccination hysteria and facilitated communication pseudoscience is that fear around autism can be incredibly dangerous, and we need to do better. For now the most likely hypothesis remains what it was before the study was published: Antidepressants do not cause autism.