The White House recently issued a press release with links to scientific studies to back up Trump’s claim that use of acetaminophen, commonly referred to as Tylenol, during pregnancy causes autism, but those studies provided only “weak” and “inconclusive”, evidence, according to physicians with expertise in reviewing medical research who spoke to the Guardian.
Jeffrey Singer, a surgeon and senior fellow at the Cato Institute who has written about the Tylenol/autism claims, said that the links in the White House press release showed that the claims contained a political spin.
The release creates the appearance of extra evidence by linking to the same studies multiple times, Singer said. The most recent study cited – a review paper published this August by authors affiliated with Mount Sinai and Harvard – is made to look like two separate studies, one from Harvard and one from Mount Sinai. A Boston Birth Cohort study authored by Johns Hopkins researchers is also linked twice in the release, as though it is two separate studies.
“So you say a Harvard study found the same thing as the Hopkins study and the Mount Sinai study. It can give the appearance that [the same results have] been achieved in one study after another, all from very prestigious centers,” Singer said.
Jake Scott, a physician and member of the Vaccine Clinical Advisory Committee at Stanford, said the press release “makes me sick in my stomach, to be honest”, because “they’re presenting association as causation, for one thing”.
There is a much higher bar for proving causation than association. In this case proving causation would mean showing that eliminating exposure to Tylenol would actually reduce the number of autism cases.
For example, rising ice cream sales might be associated with heat stroke, simply because both are things that happen in the summer, but banning the sale of ice cream would not reduce the number of heat strokes. To prove causation, it’s not enough to show that one occurrence follows another – these results must show up again and again in a variety of study designs, and scientists also need to find a plausible explanation as to how the two factors are related.
None of the studies linked in the press release claim to have proven causation.
In fact, all of them acknowledge they cannot definitively establish causation, although some use more cautious wording than others. One way to quickly find the scope of a study’s claims is to search key terms like “causal”, “causation”, “association”, “limitations”, “caution”, and “further investigation”, according to experts.
Once you actually start reading the studies, more of the statements in the press release start to fall apart, the experts said. For instance, the release states: “Scientists have proposed biological mechanisms linking prenatal acetaminophen exposure to altered brain development and adverse birth outcomes.” Biological mechanisms explaining how an exposure leads to an outcome are an important way to establish causation. It took decades for scientists to establish that cigarettes cause cancer through the biological mechanism of introducing carcinogens into the lungs, which create cancerous mutations in the cells, for example.
But the study linked in the release to illustrate a biological mechanism doesn’t actually include a biological mechanism, but the opposite. The authors write: “Mechanisms underlying the associations of prenatal acetaminophen exposure with adverse birth outcomes remain unknown.” Additionally, the study did not investigate “altered brain development” at all, as the press release indicated, but instead only looked at outcomes related to birth weight and gestational age.
Singer and Scott agree that another way to fact check the Trump administration’s scientific claims is to look at what research is being left out of the conversation.
For example, the press release left out a Swedish study published in 2024. “That’s a study where they followed nearly two and a half million births … they used sibling controls,” said Singer, “So what they found was that acetaminophen [the generic name for Tylenol] use during pregnancy showed a slight association [with autism].” But that association disappears when you compare siblings, who have many more genetic and environmental similarities than comparison groups in other studies, Singer explained.
The authors of the Swedish study wrote: “This suggests that associations observed in other models may have been attributable to confounding,” meaning that it was likely not the Tylenol itself, but other factors connected to Tylenol use (which could include, for example, more fevers during pregnancy) that created the risk.
“That’s a pretty impressive study,” Singer said. “I’m not saying they’re right either. What I’m saying as a doctor who reads the research literature, this question has not yet been answered.”
Singer says there’s likely a reason why the White House included some studies and not others: “Secretary Kennedy, I think he’s under pressure, because remember, he was going to tell us the cause of autism by September.”
The Tylenol claims are reminiscent of arguments against vaccines, according to Scott, who recently testified before Congress to argue against claims that vaccines cause chronic illnesses – claims based on a study that was never submitted for publication due to flaws in its design.
“It’s all part of a pattern of exploiting really weak evidence that’s based on really poor quality studies,” Scott said.
