A new U.S. study suggesting a link between epidural use during childbirth and autism is sparking alarmed criticism from anesthesiologists.
Despite an impressive sample size – more than 147,000 patients over seven years – critics say there are basic research flaws that mean moms looking to science for answers will draw the wrong conclusions.
Nearly three quarters of the births surveyed were from mothers who received a labour epidural for pain control. Diagnoses of autism were rare – less than two per cent in both groups – but rates were slightly higher in children whose mothers received epidurals compared to children whose mothers did not.
“Our findings raise the concern that the short duration of (epidural) exposure may be associated with long-term neurodevelopmental disorders in offspring,” the study authors wrote in their JAMA Pediatrics article.
But this has drawn sharp criticism from numerous anesthesiologists who administer epidurals.
“Labour and delivery can be extremely political, emotional and personal,” says Dolores McKeen, an obstetric-anesthesiologist and President of the Canadian Anesthesiologists’ Society (CAS). “These stories can really instill a lot of fear and guilt in moms and families… that are not necessary and not founded, particularly in a time that’s already so stressful.”
The CAS recently joined numerous U.S. societies for anesthesiologists, obstetricians and pediatricians, as well as the UK-based Royal College of Anaesthetists, in publicly denouncing this study.
“I think even for the average physician who does research, identifying issues with this paper is not super straightforward,” says Dan McIsaac, an anesthesiologist and researcher with the University of Ottawa. “You need people with both the clinical experience to understand how these decisions are being made as well as the research insight into how big data studies are done.
Critics say flaws in the study begin with the data’s diagnoses of autism, which they say are unreliable.
“Just because the top of a column on a database table says ‘autism’ doesn’t mean that’s actually an autism diagnosis made by a clinician,” says McIsaac. “In the studies that are available to date, they suggest that, in fact, those diagnostic codes for autism (used in the study) are not accurate enough to line up with a true clinical diagnosis of autism.”
McIsaac explains that this inaccuracy can happen in big health-data studies when information is extracted out of context from past records rather than recorded in the present with specific research goals already in mind.
For example, a child may have been diagnosed with autism during a checkup but the diagnosis may have been omitted from a database because other health concerns were more pressing. Or, conversely, a doctor may have decided a child did not have autism but still recorded ‘autism’ as the reason for the visit, leading to incorrect interpretation years later.
In these situations, “there are factors being considered that aren’t documented,” says McIsaac. “At the end of the day, if the data isn’t high enough quality, then we don’t have confidence at all.”
Critics also flag the study’s observational design, which was unable to ensure both groups of patients shared the same racial compositions, income, education or number of previous pregnancies. This means that any of those factors, rather than epidurals, could have accounted for the less than one per cent difference in rates of autism. Moreover, those same factors also affect a woman’s likelihood of choosing an epidural in the first place.
“Considering the way healthcare is structured in the States (where the study was done), those who have insurance and means are probably more likely to consult a doctor for anything related to autism,” says Valerie Zaphiratos, an obstetric-anesthesiologist in Montreal and chair of the CAS Obstetric Section. “If you can pay for healthcare, you may find a diagnosis of autism earlier compared to a woman who has less means and did not get an epidural because she wasn’t covered for it.”
Women who received epidurals may also have had medical conditions during pregnancy that made it more likely for them to require an epidural, such as diabetes, high blood pressure or obesity. These medical conditions, rather than epidurals, could have also accounted for the difference in autism, says McIsaac.
“If this is something women want an answer to, we need to do a randomized-controlled trial. That’s the only way to get a definitive answer and I have a strong suspicion that we would find no association,” he says.
If so, it wouldn’t be the first time an observational study is corrected by a randomized-controlled trial. For more than a decade, there was debate over whether general anesthesia used on children for surgeries is associated with altered brain development. A randomized-controlled trial last year finally yielded a definitive answer: No, it’s not.
“That caused a lot of concern for parents. You can imagine – your kid needs surgery but you’re really worried that the anesthetic will throw their development off. You’re feeling guilty, stressed and everything else,” says McIsaac.
“Now, we’ve got women coming in with labour – a tremendously painful, emotionally draining process… and we’re throwing probably inaccurate data into the fray to potentially confuse things and stress people out even more. That’s why there’s a lot of concern around this paper from people who take care of women every day.”
Anny Xiang, one of the study authors, told U.S. News, “I don’t think people should be panicked.” However, when further attempts were made to contact the researchers for an interview, a spokesperson said they were unavailable to discuss their work.
McKeen says it’s important to reassure women that they should not feel guilty about wanting pain relief.
“I’m very sad that JAMA Pediatrics did not use their experts and resources to address this fear-mongering and creation of a lot of unnecessary anxiety,” she says. “I think having a well-hashed out discussion and allowing obstetric-anesthesiologists to respond in an editorial would have been important.”
Zaphiratos says she does not understand the motivation for the study. “It’s never been a fear in the past.”
Because the study has only been out for a few weeks, both McKeen and Zaphiratos say they haven’t yet seen it make a big splash among moms-to-be.
So, there’s still time to ensure pregnant women receive the right take-home message.
“If my wife were going into labour, this study would have absolutely no impact on advice I would give her on whether or not to get an epidural,” says McIsaac.

The comments section is closed.
First there was the measles vaccine-autism connection, now this. We are our own enemy! How can ordinary people interpret this? And then we wonder why the public distrusts or doesn’t understand science. Thanks for sharing, Ben.
This paper is data dredging: put enough data in and by chance you will find some positive relationships )Ioannidis JP. Why most discovered true associations are inflated. Epidemiology. 2008;19(5):640-648. doi:10.1097/EDE.0b013e31818131e7). It is doubtful that it is even worth running a trial of this hypothesis, until similar relationships have been confirmed by other database studies. JAMA Pediatrics seems to publish such destructive papers: as they did for the very dubious paper about fluoride last year. (Green R, Lanphear B, Hornung R, et al. Association between maternal fluoride exposure during pregnancy and IQ scores in offspring in Canada [published online August 19, 2019]. JAMA Pediatr. doi:10.1001/jamapediatrics.2019.1729) So be cautious about what a single paper like this shows: very large population sizes enable trivial associations to be “significant”, even though not large enough to be important.