Challenges of Studying Tylenol Use During Pregnancy
Studies that establish causal relationships between medications and adverse effects often involve randomized controlled trials. In this case, that might mean one group of pregnant women being assigned to take Tylenol for fevers that might arise, and the other not, and then comparing the results in their children up to 10 years later. Such a study, however, would be challenging to design and conduct in an ethical and feasible way.
Observational studies, however, involve “simply watching people and seeing what happens to the children of mothers who do take acetaminophen and comparing that against children who are born to mothers who don’t use acetaminophen,” says Brian Lee, PhD ’09, MHS ’08, an epidemiology professor at Drexel University. Lee, using health records from Sweden’s national health system, conducted one of the largest studies to assess whether acetaminophen use might be associated with autism and other neurodevelopmental outcomes.
Obstacles to Understanding Cause and Effect
The challenge of understanding cause and effect is that there may be other factors at play that influence both the exposure of interest (in this case, taking Tylenol) and the outcome (in this case, autism).
Ice cream offers a helpful analogy for understanding these confounding factors. Looking at data alone, there’s a statistical association between eating ice cream and drowning, says Lee. “But of course, it’s not a causal one. The confounder here is hot weather, which increases both your ice cream consumption and your propensity to go swimming—hence drowning.”
To account for the weather factor, you might look only at data from hot days—“and that association between ice cream eating and drowning completely goes away,” Lee explains.
When it comes to studying Tylenol and autism, one confounding factor is that women may take the painkiller during pregnancy to address another health condition that may be linked to autism, such as infection or fever. “It might look like there’s an association with acetaminophen, but really the association could just be reflecting the underlying infection that the person has,” Lee said in the September 29 episode of Public Health On Call.
And then there’s another major consideration: Genetics, “that big elephant in the room that causes a large portion of autism,” says Lee.
“So then the question is, how well does the study really adjust for those differences … so that we can try to separate the effect we care about from all these other relationships?” Elizabeth Stuart, PhD, AM, chair of Biostatistics, told the podcast.
A Major Study on Tylenol and Autism
Lee’s study, conducted in Sweden, did not have access to genetic data. But it did have a massive trove of health records of nearly 2.5 million children born in Sweden between 1995 and 2019, as well as the health records of entire families. As a result, it was able to account for genetics by comparing siblings, who have significant genetic similarities—something that previous studies linking Tylenol and autism did not.
Like studies that preceded it, Lee’s research initially found a small association between Tylenol use and autism when they compared members of the general population. But when looking at siblings within the same family where the mom took acetaminophen for one pregnancy and not another, that increased risk “completely disappeared,” says Lee. “There was no evidence to say that there was any effect whatsoever.”
The sibling comparison was a major advantage to the design of this study. But,, like all studies, there were limitations, says Stuart.
For example, mothers may not accurately report their Tylenol use. And “we don’t necessarily have direct measures and certainty around what the women were taking”—such as how much Tylenol they took and how often, Stuart notes. To account for inevitable limitations, researchers like Lee also try to anticipate potential issues with their data and assess how much impact they may have.
Making Sense of Conflicting Findings
On the podcast, Lee addressed why many studies may have found an association between autism and acetaminophen use during pregnancy. In his view, it’s because those studies did not utilize a sibling control to account for familial factors. Recently, a Japanese study of some 200,000 pregnancies came to the same conclusion as the Swedish study: It suggested a slight increased risk of autism from acetaminophen use during pregnancy, but when they analyzed siblings with comparable genetic makeup, there was no link.
Lee noted that early studies linking coffee intake with increased risk of cancer did not account for the fact that many coffee drinkers were also smokers. “And in the later studies that did, these associations went away. And I think the science here for acetaminophen and autism is headed in that direction,” says Lee.
Building a Body of Evidence on Tylenol and Autism
Lee and Stuart both said that no one study on Tylenol and autism can establish cause and effect. “It’s really about building a body of evidence and being transparent about the strengths and weaknesses of any individual study,” says Stuart.
“Usually for epidemiology, we never get a 100% definitive case closed on any topic,” says Lee. Rather, as more and more studies are done, “there’s an evidence needle pointing in a certain direction that is shifting toward, this causes this, or this doesn’t.”
Right now, in his view, “the needle is pointing strongly toward there being no causal effect of acetaminophen use during pregnancy on autism.”
The Next Step Toward Understanding the Tylenol-Autism Issue
It’s essential to be clear and transparent about where the evidence stands, Stuart says.
To help achieve that, one next step might be for a government agency, a nonprofit, or academic or professional organization to bring together people with different types of expertise on the topic and study design to assess and communicate where the evidence stands—and helping the general public process that uncertainty, says Stuart.
“Sometimes science communication makes it seem more certain than it is. In these sorts of questions, we just don’t have that certainty.” As people navigate questions around Tylenol and autism, it’s key to give people “a little bit more understanding of why we don’t have certainty in these questions, and a little bit more comfort with that.”
Even if the research is not fully settled, the practical recommendations about how to use Tylenol during pregnancy remain unchanged, says Lee: “Even if you don’t have any reason to believe that a medication is harmful, you probably shouldn’t be taking it … at levels that exceed the levels recommended by a physician.”
link
