I Found This Hanging in My Chiropractor’s Office
This week, I went to my chiropractor because my foot has been hurting, which is an entirely different story involving plantar fasciitis, an X-ray, and the unfortunate realization that apparently I am now a person who has strong opinions about supportive footwear.
While I was there, I saw something hanging on the wall. It was a printed article with the headline: “The Connection Between Acetaminophen and Autism: A Bizarre Response From the Mainstream Scientific Community, and a Brand New (For the First Time) Line of Evidence.”
Naturally, I took a picture. Because of course I did. Then I went home, found the article, and read the whole thing.
I have regrets.
Haven’t We Done This Already?
Not that long ago, I wrote a post called You Did Not Cause Your Child’s Autism, and Other Things You Needed to Hear Today. I wrote it because parents—mothers in particular—have been handed an astonishing number of things to feel guilty about over the years. Was it something you ate? Something you took while pregnant? Tylenol? Vaccines? Something you did when your baby was an infant?
We have been doing this to mothers for decades. Once upon a time, we literally blamed autism on mothers not loving their children enough. So when I saw an article hanging in a healthcare office stating that acetaminophen exposure is responsible for “many if not most cases of autism,” I didn’t just see another questionable article about Tylenol.
I saw the same old blame game wearing a lab coat.
But this particular article is worth talking about because, at first glance, it looks impressive. It has references. Lots of them. It talks about statistical models, epidemiology, biological mechanisms, computer simulations, and what the authors describe as 33 separate lines of evidence.
If you aren’t someone who spends an unreasonable amount of time reading scientific papers, you could reasonably look at it and think, Well, damn. This certainly looks scientific.
And that’s exactly the problem.
Six Years of Learning to Be Skeptical
I went to graduate school for a total of six years. Yes, six. Because apparently a pandemic, having kids, and running a company at the same time means it takes me a little longer than most people to get things done.
But during those six years, I was taught not only how to conduct scientific research, but how to evaluate research that other people have published—and to do it with a very skeptical eye.
That distinction matters. Being scientifically literate isn’t just knowing how to run a study or understanding what a p-value is. It’s learning to ask uncomfortable questions about research, including research that supports something you already believe. How were the variables measured? Who was included? Who wasn’t? What other explanations could account for the result? Does the study design actually allow the authors to make the conclusion they’re making? Does the citation being used actually support the sentence sitting next to it?
And perhaps most importantly: What would change my mind?
Good science requires skepticism. Not the “I did my own research on Facebook” kind of skepticism, but the kind where you are willing to scrutinize evidence you want to be true just as carefully as evidence you don’t.
Which brings me back to the article on the wall.
A Citation Is Not a Magic Wand
One of the most important things I learned in those six years is that putting a little number at the end of a sentence does not magically make the sentence true. You have to look at what the referenced study actually found.
The article hanging in my chiropractor’s office argues that acetaminophen—Tylenol—is responsible for many or even most cases of autism in children who are biologically susceptible.
That is an enormous causal claim.
So let’s look at some of the evidence.
Enter: 2.5 Million Swedish Children
In 2024, researchers published a study involving nearly 2.5 million children in Sweden examining acetaminophen use during pregnancy and later diagnoses of autism, attention-deficit/hyperactivity disorder (ADHD), and intellectual disability.
The study was published in JAMA, the Journal of the American Medical Association, one of the major peer-reviewed medical journals in the world. “Peer reviewed” means other qualified experts evaluate scientific work before publication. It doesn’t mean every published study is automatically correct—science doesn’t work that way—but a massive study published in a rigorous medical journal carries considerably more evidentiary weight than an article interpreting selected studies.
Initially, researchers saw a very small association between acetaminophen exposure and neurodevelopmental diagnoses. That’s the part that makes a great headline: TYLENOL LINKED TO AUTISM. Everybody panic.
Except the researchers didn’t stop there.
They compared siblings from the same families. Why? Because pregnant people don’t randomly take Tylenol. They take it because they have fevers, infections, migraines, pain, or other medical issues. Two unrelated pregnant women may also differ genetically, environmentally, socioeconomically, and in countless other ways. Scientists call these alternative explanations confounding variables.
Comparing siblings helps control for many of those shared genetic and environmental factors. And when researchers did that, the association disappeared. They found no evidence of increased risk of autism, ADHD, or intellectual disability associated with prenatal acetaminophen exposure in the sibling analysis, and no dose-response relationship.
And in 2026, another large sibling-matched study published in JAMA Internal Medicine reached a similar conclusion: after accounting for familial factors, prenatal acetaminophen exposure was not associated with increased autism or ADHD risk.
Does that prove acetaminophen could never have any effect on neurodevelopment under any circumstances? No. Science rarely gives us absolutes.
But those findings certainly do not support the statement that acetaminophen causes “many if not most cases” of autism.
“Actually, That Supports Our Theory Too.”
Here’s where the article gets interesting. The authors argue that scientists are interpreting these results incorrectly. Their theory is that certain children are biologically susceptible to neurological injury from acetaminophen. They created a computer simulation based on that proposed relationship and demonstrated that, under their assumptions, adjusting for susceptibility could make an acetaminophen association disappear.
Therefore, they argue, the disappearing association in the Swedish study may actually be consistent with their theory.
Computer modeling can be incredibly useful, but there’s an important distinction here: creating a model in which your hypothesis produces a particular result demonstrates that your hypothesis could produce that result. It does not demonstrate that your hypothesis actually caused the real-world result.
I can create a perfectly coherent model explaining why my children leaving their shoes in the middle of the hallway causes me to drink wine. That does not make abandoned Crocs an established risk factor for Cabernet consumption.
Plausibility and causation are not the same thing.
Somehow, We Also Get to Circumcision. And Cannabis.
Stay with me.
The authors discuss earlier research reporting an association between circumcision and autism. They suggest the association could actually be explained by acetaminophen given around the procedure. Then another study failed to find the same circumcision-autism association.
Problem for the hypothesis? Apparently not.
The authors propose that American babies may have so many other opportunities for acetaminophen exposure that an effect associated with circumcision could become difficult to detect. The article also ventures into vaccines and inflammation and eventually proposes another possible line of evidence involving cannabis exposure, acetaminophen, and differences in autism prevalence.
And this is where we arrive at one of the most important questions you can ask about any scientific hypothesis: What evidence would convince the person making the claim that their hypothesis is wrong?
Good scientific hypotheses need to be falsifiable, meaning there must be some conceivable evidence that could show the hypothesis isn’t correct. If finding an association supports your theory, and not finding an association also supports your theory, we have a problem.
But There Are 33 Lines of Evidence!
Thirty-three sounds impressive. But evidence doesn’t become stronger simply because we divide an argument into more pieces. Thirty-three weak or indirect observations don’t automatically equal one strong causal conclusion.
Scientists care about the quality of evidence. Was the research conducted in humans? How many people were studied? Was exposure measured accurately? Were researchers able to account for other explanations? Has the finding been replicated? Is there a dose-response relationship? Do multiple high-quality studies point toward the same conclusion?
And perhaps most importantly: Does the evidence justify the size of the claim being made?
There is a canyon between “Researchers should continue studying whether acetaminophen exposure during development has neurodevelopmental effects” and “Acetaminophen is responsible for many if not most cases of autism.”
You cannot leap across that canyon by adding more citations.
Why the Healthcare Office Part Matters
This is actually why I’m writing this. I didn’t find this article in a Facebook group. My aunt didn’t send it to me with twelve exclamation points. I wasn’t three pages deep into Google at midnight.
It was printed and displayed on the wall of a healthcare provider’s office.
When a healthcare professional chooses to display something in their office, patients reasonably assume it carries some degree of professional endorsement. Most parents aren’t going to photograph the article, go home, find the original, pull up the studies it references, compare study designs, look up the journals involved, and spend their Sunday morning reading epidemiology.
I did. Because apparently this is how I relax now.
But parents shouldn’t need a PhD to determine whether something hanging in a healthcare office is giving them reliable medical information. Healthcare professionals have tremendous authority, and that comes with tremendous responsibility—especially when we’re talking about autism.
There Is a Real Family on the Other Side of This Claim
This is the part I can’t separate from the science. I’m not only a behavior analyst reading this article. I’m a mother.
I know what it feels like to receive an autism diagnosis for your child, and I know how quickly your brain starts searching backward. What did we miss? Was it something during pregnancy? Something during infancy? Something I should have known? Something I should have done differently?
Maybe you remember taking Tylenol when you had a fever while pregnant. Maybe you remember giving it after vaccinations. Maybe you remember standing in your baby’s room at 2 a.m., exhausted and scared, carefully measuring the dose because your baby was miserable.
And suddenly an ordinary parenting decision becomes evidence in a case you’re building against yourself.
Maybe I did this.
That thought has consequences. That’s why words like “many if not most cases of autism” require an extraordinarily high evidentiary bar. Because you’re not discussing numbers in a spreadsheet anymore.
You’re handing parents guilt.
So How Do You Know What to Believe?
You don’t need to become an epidemiologist. But when you encounter a dramatic medical claim, ask some questions. Who is making the claim? Where was it published? What does the actual research say—not just someone’s interpretation of it? Are large, well-designed studies finding the same thing? Is the author distinguishing correlation from causation? What evidence would prove the hypothesis wrong?
And perhaps the easiest question of all: Is the certainty of the language proportional to the certainty of the evidence?
“More research is needed” and “this causes most autism” are worlds apart.
Science Is Allowed to Change
If high-quality evidence someday demonstrates that a medication, environmental exposure, genetic interaction, infection, or anything else contributes to autism risk, scientists should follow that evidence. That’s the deal.
I don’t need acetaminophen to be innocent. I don’t have an emotional attachment to Tylenol.
Science isn’t about choosing your preferred answer and defending it forever. It’s about asking questions and being willing to discover that you were wrong. Research into autism should continue. Research into prenatal medication exposure should continue.
But we should be extraordinarily cautious when turning observational, indirect, or theoretical evidence into declarations of causation—especially declarations frightened parents are going to read.
And Once Again: You Did Not Cause Your Child’s Autism
I apparently need to say this again, so I will.
You did not cause your child’s autism because you treated a fever. You did not cause it because you vaccinated your child. You did not cause it because you made the best medical decisions you could with the information you had at the time.
Your child’s autism is not a crime scene.
We do not need to spend their childhood collecting evidence against you.
We can continue asking scientific questions while simultaneously refusing to turn every new hypothesis into another reason for mothers to blame themselves.
And healthcare professionals? Maybe let’s not print an article claiming we’ve figured out the cause of “many if not most” cases of autism and hang it on the wall unless the evidence can actually carry the weight of that statement.
Because some of us will take a picture. Some of us will go home and read all the references.
And some of us have blogs.
If you are like me and like to actually read the studies, here you go:
https://jamanetwork.com/journals/jama/fullarticle/2817406
https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2850975