Tylenol has entered the chat: RFK Jr. wants to link it to autism, even though the best evidence says otherwise.

It's not pseudoscience. But it's not settled either, nor is it "safe" to assume otherwise.

Left: Getty images. Right: Chip Somodevilla/Getty Images.

The Wall Street Journal reported that Secretary Kennedy’s forthcoming report on what causes autism will pin some of the blame on prenatal exposure to Tylenol.

Would that conclusion be accurate? Unlikely. Is it a problem if this becomes our national stance on the issue? Yes. Let’s go Inside Medicine…

Not pseudoscience, but nowhere near conclusive.

Unlike many Kennedy community claims, a link between autism and Tylenol is not complete garbage. (Note: I’ll be using Tylenol here, but this all applies to the generic versions too, known as acetaminophen in the US and paracetamol in some other countries.) In fact, researchers, including at Harvard and Yale, have published epidemiological studies pointing towards an association. But a 2024 study from Sweden published in the Journal of the American Medical Association provided what looks to have been a devastating rebuttal to the thought. The Swedish team elegantly cut through the noise by simply comparing outcomes among siblings. That was a clever natural experiment, and it was novel to this question. Using that approach inherently removed many (most?) of the confounding variables, such as the genetics of the mothers and other risk factors. The Tylenol and autism link fell apart.

While even that study was not perfect, there’s convincing evidence that its methods were adequate to the task. That paper doesn’t definitively end the debate (though, honestly, it might), but it certainly says that the research showing a link between Tylenol and autism has a much steeper hill to climb before we’d accept it—let alone make national policy. For one thing, the Swedish researchers were able to replicate the older studies’ findings—and this is key—when the sibling control method was not deployed. That means that for whatever limitations the Swedish study had (including that it only covered Tylenol that doctors prescribed, rather than over-the-counter use), its methods were similar enough to detect the association that previous researchers had reported. So, had they not deployed the sibling control method, the Swedish team also would have found the link. I find that to be compelling.

Why invoking the precautionary principle doesn’t fly.

Let’s say the Swedish study is inadequate—that is, neither incorrect, nor sufficient to slam the door shut on the issue. Why not tell pregnant women to avoid Tylenol “just to be on the safe side”—the so-called “precautionary principle”? The answer is two-fold. First, crying wolf is inherently bad. (You know how that story ends.) Second, it’s not even clear that this advice would be erring on the safe side. There’s some data to suggest that fevers during pregnancy are linked to autism down the road and, get this, using fever-reducing medications like Tylenol appears to break that link. If we believe that, the guidance stemming from the precautionary principle would be reversed.

Other reasons why this is bad.

There are at least three reasons why it would be bad national policy to link autism and prenatal Tylenol use based on what we know now. One is that this inappropriately blames mothers (which may be the point). But the other two reasons are more dangerous. First, it’s likely that the HHS report on autism will also blame some vaccines. We know this because Secretary Kennedy had a pathetic meltdown on X/Twitter when a recent Danish study yet again debunked any link between vaccines and an array of neurological conditions. By throwing in some non-vaccine scapegoats, it may look like the Secretary is being “even-handed,” rather than just taking on vaccines. Second, we seem to be entering a new phase of American science wherein cherry-picked studies are used to “prove” something, while others (often better science, by the way) are ignored or wrongly dismissed. That is not gold-standard science. Rather, as my colleague Dr. Bill Hanage says, in the place of evidence-based policymaking, it’s policy-based evidence making.

In a way, what worries me is that this is not pseudoscience. It’s just not ready for primetime science and, as above, if we get this backwards, we could cause as much or more harm. This throws the precautionary principle out the window.

Tylenol deserves scrutiny, though.

None of this negates another inconvenient truth, by the way: Tylenol (all acetaminophen) is probably the single most dangerous over-the-counter medication on the market. (Aspirin may be more dangerous, but it’s used less these days.) While millions of people routinely use the medication safely, every year there are thousands of people who develop liver damage, and hundreds who die from overdoses, including many unintentional ones. I refer you to the deeply reported and riveting new book, No More Tears, by Gardiner Harris. You’ll be absolutely shocked by Johnson & Johnson’s behavior on the issue of Tylenol safety (and others) over the years.

Should there be a more prominent warning to patients on the box? I’d say yes. Does that mean that we should accept some strange bedfellows and passively accept this purported link between autism and prenatal Tylenol use to achieve that goal? No. Despite what this administration does—despite what it mistakenly believes is “gold-standard science”—genuinely rigorous scientific processes still matter

---

If you have information about any of the unfolding stories we are following, please email me or find me on Signal at InsideMedicine.88.