Circumcision, Autism, and the Gish Gallop

By now you’ve probably heard that in a cabinet meeting last Thursday, HHS secretary RFK Jr. criticized an eight-month pregnant woman for “gobbling Tylenol with a baby in her placenta.”

Everyone misspeaks on occasion. I do it all the time. But I wouldn’t refer to babies growing in placentas, any more than I would describe sperm cells as growing in a penis. Nobody makes that kind of mistake. Not unless they’re a slightly confused 10-year-old, or a leading health official with limited knowledge of science.

Moments later, RFK referred to “two studies that show children who are circumcised early have double the rate of autism.”

This was the beginning of a Gish Gallop of misstatements and flawed evidence, culminating in a new claim about acetaminophen (i.e., it causes autism when infants take it) and a reference to a new review.

A Gish Gallop is a debate trick in which you spew so many flawed arguments so rapidly your opponent doesn’t have time to refute all of them. (See Footnote 1.)

Indeed, I don’t have time to address everything RFK Jr. said this week about autism spectrum disorder (ASD). In this newsletter I just want to do two things: Show why circumcision and acetaminophen don’t cause infants to become autistic, and then explain why saying so is important. At the end I’ll share some good news.

Circumcision or acetaminophen?

Right after claiming that circumcision causes autism, RFK Jr. added that “it’s highly likely because they [infant boys] are given Tylenol”, meaning there’s a small chance that circumcision is the culprit, but mostly likely it’s the acetaminophen given for pain.

Early evidence

The earliest of the two studies RFK Jr. may have been referring to, conducted in 2013 by a pair of U Mass researchers, showed higher rates of ASD in countries with (a) greater prenatal acetaminophen use and (b) higher circumcision rates.

However, the researchers only looked at 8 countries for the acetaminophen-autism correlation, and 9 countries for the circumcision-autism correlation. This is not even close to being enough for a trustworthy correlational analysis.

The researchers don’t name those countries, but you can find them on the internet in a supplemental data file. I looked and much of the data seems useless. The numbers for Australia, for instance, come from a study of 78 people. We can’t say anything about country-wide rates of acetaminophen use, or anything else, based on such a tiny sample.

A deeper issue is known as the ecological fallacy, or the misapplication of group data to individuals.

In their samples (8 or 9 countries), the researchers only show that countries with relatively high acetaminophen use or circumcisions also have relatively high numbers of kids with ASD.

This by itself constitutes zero evidence of anything. Rather, it just identifies two characteristics of countries that tend to co-occur. You can play this game all day. For instance, among industrialized nations, wealthier ones tend to have more houses per capita and more heart disease. But houses don’t cause heart disease.

A better approach?

In 2015, a pair of Danish researchers compared rates of ASD among 342.877 boys who were either circumcised or intact. This is the right kind of comparison, as it bypasses the ecological fallacy I mentioned earlier.

The results sound dramatic: Circumcised boys were 46% more likely to have a diagnosis of ASD.

However, once again, sample size is among the reasons the data can’t be trusted. Only 5,033 boys, or 1.5% of the sample received a diagnosis of ASD. Among them, only 57 had been circumcised.

Statistically, the problem with such tiny subgroups is that group differences are easily over-inflated. If, say, only 55 of the boys with ASD had been circumcised, the difference in ASD rates would no longer be significant. (I discuss other problems with the study in Footnote 2.)

The researchers hypothesized that the painfulness of circumcision is what triggers autism. If that’s what you believe, why focus on circumcision? Why not consider any source of acute pain? In particular, why not focus on urinary tract infections? They’re extremely painful, and much more common among uncircumcised infants compared to intact peers.

RFK Jr.’s original reference to “two studies” could’ve been to either or both of the ones I’ve discussed so far. However, in an X post last week, he cited a new paper:

“An August 2025 http://Preprints.org review by Patel et al. directly validates my point that the observed autism correlation in circumcised boys is best explained by acetaminophen exposure...”

The new preprint

“Preprint” does not mean “accepted for publication and waiting on line to be printed in an academic journal”. Rather, it means “just chilling in cyberspace somewhere.”

This particular preprint is so awful I’m almost reluctant to tell you about it. I mean, my golden retriever once had an upset tummy and decided to visit every room of the house while I was out. Do you really want to hear the details?

I’ll share a few anyway, so you know how low our leading health official will go to find evidence supporting his ideological agenda.

The new preprint is a review. The main purpose is to explain why mainstream experts have rejected evidence that acetaminophen triggers most cases of ASD. The authors claim that

“Despite overwhelming evidence, current medical literature in the field adheres to the myth that autism has a complex cause, with multiple factors contributing to the disorder, and with no single factor being critical.”

This is not a myth. It’s just the state of the literature. ASD is linked to interactions between a number of genetic variants and environmental influences ranging from maternal infection to air pollutant exposure.

As for this new review, a key problem is that most of the studies don’t belong there in the first place.

For instance, the authors make much of a study showing that newborn rats who receive acetaminophen lose the ability to find food in a maze. In another study, prenatal exposure to acetaminophen caused rats to engage in more wrestling behavior.

Do I need to say it? Rats aren’t people. Difficulties in locating food aren’t characteristic of ASD. Nor is a love of wrestling.

In Footnote 3 I say more about the technical shortfalls of the review. Here I just want to close with an observation about its integrity.

At the end of the review, in the Conflicts of Interest section, the authors write:

“The authors have no conflicts of interest relevant to this article to disclose.”

This is false. The authors note elsewhere that work on the review was funded WPLab, Inc., a non-profit corporation based in Durham, North Carolina.

WPLab is an advocacy group. They make no attempt to hide their biases. For example, here’s a typical screen shot from their Who We Are page:

That “without a reasonable doubt” alone disqualifies the statement from serious consideration. Even the authors of the 2013 and 2015 studies acknowledged limitations, entertained doubts, and hold that judicious use of acetaminophen during pregnancy is acceptable.

The “many or most cases” phrase appears throughout the website, and in the preprint. Why? Because the review was not just funded by WPLab. The three WPLab leaders, including the CEO, William Parker, are among the co-authors. (The lead author, M. Vishnu Patel, is an emergency medicine physician.)

It would be hard to find a clearer example of conflict of interest. The folks who run a nonprofit corporation that advocates for a certain claim pay themselves to write a review that pushes that claim.

In short, it’s a terrible, meaningless piece of scholarship grounded in a lie about the impartiality of the authors. And the most powerful health official in the U.S., in his X post on Friday, referred to it as a “rigorous scientific framework.”

Why this is important

I’m not just sharing with you some crap studies I stumbled across this week. RFK Jr. cited these studies in a cabinet meeting with the President, and they are guiding the use of your tax dollars in the way HHS is implementing some of its public health policies.

Moving forward, how will we be affected by RFK Jr.’s Gish Gallop through bad data?

1. Declining trust in public health agencies.

Although this is happening now to a greater extent among Democrats and Independents as opposed to Republicans, there are signs of nonpartisan erosion of trust.

2. Increasing normalization of ideologically-driven junk science.

Reputable scientists, including those who work at HHS agencies, can see through RFK Jr’s “evidence”, but his ongoing reliance on low quality and/or unpublished data makes the practice incrementally more conceivable. Think of this as the “broken window” effect of bad science.

3. Dumbing down of public discourse.

RFK Jr. repeatedly describes ASD as preventable, claiming that “genes don’t cause epidemics...you need an environmental toxin.” This is a “bad guy” approach to ASD: There must be one culprit, like vaccines, or acetaminophen. The public needs to be working with more nuanced models – e.g., the notion that ASD arises from genetic vulnerabilities to certain kinds of environmental exposures. The genetic piece is critical, as it helps ensure that parents don’t blame themselves for having done something wrong.

4. Increasing distraction from credible data.

Somehow, RFK Jr. manages to keep ignoring a brilliant, widely-esteemed JAMA study from last year ruling out associations between acetaminophen use during pregnancy and ASD diagnoses in the children. Strengths of the study include its size (nearly 2.5 million children) and direct evidence that any apparent links between prenatal acetaminophen use and ASD are likely due to confounds – e.g., an infection the mother had during pregnancy, rather than the Tylenol she took, that preyed upon a genetic vulnerability.

5. Increasing distraction from time-sensitive concerns.

At the October 9 cabinet meeting, RFK Jr. spoke for about 7 minutes. He spent roughly half that time on discussing acetaminophen and autism. Nothing about the current, record-breaking measles outbreak, or about dozens of other pressing health issues on which you might want HHS agencies to take the lead. Instead, we just had one of RFK Jr’s pet fringe theories, temporarily sucking the oxygen out of the room.

The good news

Last month, the National Institutes of Health (NIH) began awarding grants from a new, $50 million dollar program, the Autism Data Science Initiative. ADSI money has actually begun reaching scientists, in spite of so many other cuts to research funding. Best of all, the projects themselves embody complex, high quality science, not just “does Tylenol cause autism?” research.

Helen Tager-Flusberg, director of the Center for Autism Research Excellence at BU, summed up the big picture quite nicely in a recent NPR interview:

“What we’re seeing coming out of this administration is extremely complex and in many ways quite contradictory...

“On the one hand....we had … from President Trump, from Secretary Kennedy, bogus simplistic ideas that Tylenol and vaccines are responsible for the rise in rates of autism...

“And in contrast to that, as I read every one of the abstracts for these newly funded grants, not a single investigator is taking such a simple-minded approach... They’re all taking a very sophisticated, forward-looking, exciting perspective on the very complex and multifaceted potential nature of factors that contribute to risk for autism.”

Here we see why self-contradictory policies aren’t all bad necessarily. If some of those policies are harmful, the ones that aren’t may turn out to be helpful. The ADSI grants are among the silver linings in an otherwise dark cloud currently hanging over American public health.

Thanks for reading!

---

Footnotes

1. The “Gish Gallop” is a term invented by Eugenie Scott, former Executive Director of the National Center for Science Education, to describe the debating style of the late creationist Duane Gish. I’m grateful to Dr. Scott for inventing the term (and for some helpful advice she shared with me early in my career).

2. Key data in the 2015 study are implausible. For instance, critics have noted that only 10.9% of the Muslim boys in the sample were found to be circumsized, but that among Muslim communities in Denmark (and elsewhere) the practice was vastly more common.

3. A technical issue. The authors of the 2025 preprint assume (without good evidence) that oxidative stress predisposes young people to ASD, and that acetaminophen exposure then triggers the disorder. In their view, the reason that mainstream scientists miss this is that they treat the contributors to oxidative stress as potential confounds, controlling for them in order to determine whether acetaminophen, all by itself, is linked to ASD. (Contributors to oxidative stress include a range of lifestyle factors such diet, sleep, smoking, etc., health conditions such as inflammation, and environmental exposures such as pollution and pesticides.) The authors are doubly wrong here. First, the best models of ASD causality routinely incorporate multiple predictors and complex pathways of influence. It’s only cranks like the preprint authors who would simply run a model controlling for all potential influences on ASD other than a single factor, such as acetaminophen. Second, there are literally no studies providing evidence for the interaction that the authors speak of (oxidative stress as a necessary predisposition to ASD, followed by acetaminophen exposure as a unique trigger).