At an last Monday, Donald Trump and his medical acolytes revealed an actionable cause of autism at long last: Tylenol. In what felt like a social media post come to life, Trump fumbled his way through the word “acetaminophen” and declared it “NOT GOOD.”
Trump’s deranged argumentation reified the disgraced Andrew Wakefield’s twenty-five-year-old anti-vaccine talking points—that they’re pumping “beautiful babies” with gallons of vaccine juices like they’re horses, that the measles-mumps-rubella (MMR) vaccine is too dangerous as a combination shot, that vaccinating a kid against hepatitis B makes no sense because you can only get it from sex—and it was followed by Robert F. Kennedy Jr, Jay Bhattacharya, Marty Makary, and Mehmet Oz acting as paternalistic doctors coming to save us all. In record time, they fingered acetaminophen (the active ingredient in Tylenol, also known as paracetamol) as a “very increased risk of autism” and they found a treatment for the condition, a prescription-based drug called leucovorin. The not-so-subtle message being communicated was that past administrations were lazy: look at what can be accomplished when competent men are put in charge!
While brand-name Tylenol was singled out in no uncertain terms by Trump, acetaminophen is an ingredient in , including the store-brand equivalents of Tylenol, many formulations of cold-and-flu medicine, Actamin, Tempra, as well as some combination drugs available with a prescription. To target Tylenol specifically, even if it was proven to be guilty, would be flagrantly insufficient.
The acetaminophen-autism connection is not a lie concocted by Robert F. Kennedy Jr, browsing the shelves at his local drugstore and looking for a random drug to blame for what was distressingly referred to during the press conference as an epidemic of autism. This potential association has been studied for two decades. To me, this feels closer to panicked employees being told they need a culprit by September, something in the environment to fit RFK Jr’s obsession with toxins, and they waded through the scientific literature and took meetings with fringe figures all-too-happy to share with them their pet theory.
So, what do we know about the possible link between taking acetaminophen during pregnancy and having an autistic child, and what sort of data compelled the FDA commissioner to approve leucovorin for children with autism?
An increase in autism diagnoses is not a crisis
First, this entire conversation needs to be framed around autism: what it is, what it isn’t, and what we already know about it. Trump called it a “crisis” and “one of the most alarming developments in public health history,” which is a hallmark of the anti-vaccine movement. In those circles, autism is depicted as a horrendous disease that steals a child’s personality, and childhood vaccines are inevitably blamed despite mountains of evidence clearly demonstrating .
Autism is a spectrum, and this spectrum was explicitly created in by bringing under the autism umbrella a number of older diagnoses like Asperger’s disorder. Autism is a condition characterized by differences in how people act and communicate in social situations, as well as focused interests, a preference for routine, and repetitive behaviours. These differences range in severity, from quirks to life-altering disabilities. At one end of this wide spectrum are perfectly happy people with friends, romantic partners, jobs, houses, and passions—yes, some even —while at the other end, we see people who don’t communicate, harm their body, and become physically violent toward their caregivers. I have friends who are autistic and I worked with the mother of an autistic child who was aggressive, and it opened my eyes as to the many variations of this condition.
What causes autism is not fully understood, but there is a very strong genetic component to it, meaning that variations in specific genes change the biology of the brain. ( of genes have been linked to the condition so far.) Could environmental exposures also be involved? It is possible but is .
Changes in diagnostic criteria and in our culture, such as television shows, public awareness campaigns, and advocacy groups, have absolutely contributed to the increase in the number of people who are diagnosed as autistic. You can’t count what you don’t see. Anything that is taboo within a culture—being gay, being trans, having a mental illness—will look more common when said taboo begins to be lifted.
The idea of identifying a “culprit” and a “treatment” for autism rubs a lot of autistic people the wrong way, and understandably so. But when a child does not communicate and is violent toward their parent, can we blame the latter for wanting some answers?
A deeply flawed body of evidence
Around the time that the MMR vaccine was wrongly accused of causing autism, some scientists worried that it was an infection during pregnancy or during infancy that led to autism… or it was the medication taken to suppress the fever that was to blame. It was hypothesized that , including acetaminophen, leading to autism. These hypotheses have over the years mingled with each other and become a real mess. When the MMR vaccine was exonerated, some still thought there was something there, because parents would often say their child changed after receiving the shot: it turns out that autistic traits often manifest around the time in life that an infant receives their childhood vaccines. It’s just a coincidence.
But this coincidence was hard for some parents and scientists to ignore. So maybe it wasn’t the vaccine… maybe it was the Tylenol given to the child after they spiked a fever following the MMR shot!
I have seen people dismiss this out of hand because autism predates the use of Tylenol, which began to be sold in the 1950s, but its active ingredient could still, in theory, be a risk factor for autism without being its sole cause.
Many studies have been published on this, and most do show a positive link between the two (for example, see this or ). However, the list of caveats is long and paramount. The studies are observational in nature, which means that they can’t prove that acetaminophen causes autism. In most studies, mothers are asked to remember if they took Tylenol in the last few months of pregnancy and also a few months after birth, and if you have ever had to fill out a food questionnaire asking you how many avocados you have eaten in the last year, you know that our memory is not reliable. This kind of questioning can foster a recall bias. Mothers of a child recently diagnosed with autism may be looking for a culprit, and they are asked in the context of a study on autism if they took Tylenol while pregnant. They are thus more likely to say yes, assuming this must be the cause the scientists are closing in on.
Some scientists tried to get around that by objectively measuring the presence of acetaminophen . The critical flaw? The half-life of acetaminophen in the human body is less than 3 hours, so these cord blood measurements were essentially useless to figure out if the fetus had been exposed to the drug over the course of nine months. (Funnily enough, this is the only study linked to in on the putative acetaminophen-autism connection. One of the worst papers they could have chosen to make their point.)
These studies also rely on a range of tools to identify which child does indeed have autism, not always limiting themselves to a proper diagnosis made by a team of qualified healthcare professionals. Moreover, the risk increases reported in these studies tend to be small and are not always consistent. In , the use of acetaminophen during pregnancy was not associated with having a child with autism spectrum disorder, but it was associated with slightly higher chances of having a child with autism accompanied by hyperkinetic symptoms, a subset of autism where the person also has a severe form of ADHD. If you cut the pie into smaller and smaller pieces, you’re bound to find an association with one of them by chance alone.
All of these problems were seemingly resolved with the publication last year of a massive study coming out of Sweden, which reported no link between acetaminophen in pregnancy and autism. But this study too has a significant weakness.
The challenge of counting pills
The looked at nearly 2.5 million children born in Sweden between 1995 and 2019, with 68,584 of them (2.8%) subsequently receiving a diagnosis of autism. While its authors initially found a slightly higher proportion of children with autism when acetaminophen had been taken during pregnancy, that increase vanished when they did an interesting calculation. They compared autistic children who had been exposed to acetaminophen in utero to their brother or sister who was not exposed to acetaminophen in utero. This is because exposure to acetaminophen during pregnancy may not be the only difference between two groups of people: other variables may come into play. Researchers try to document these and ensure that the two groups are equivalent in every other way, but so-called residual confounders remain because you can’t document everything. Using siblings as a control group means removing as potential culprits many of the factors that play a role in the early upbringing of a child, as well as some genetic factors, since siblings share half of their DNA. (Twins were not used for this analysis, only single-child pregnancies.)
Based on this study alone, it would appear as if the positive signal from previous research was not due to acetaminophen. However, there is an important discrepancy in this much-heralded Swedish study.
The percentage of individuals in this study that were recorded as taking acetaminophen during the pregnancy was 7.5%. In many other studies, the percentage was between 50 and 60%, although a few other studies also reported low percentages. Why such a staggering difference? It all boils down to how we know that someone is taking acetaminophen. The Swedish study looked at birth cohorts from 1995 to 2019, and , the only information they had about acetaminophen was that the person was asked at their very first visit, usually eight to ten weeks into gestation. A midwife would ask them what over-the-counter and prescription drugs they were taking regularly, not sporadically—which is a problem if you’re trying to document the taking of this drug to treat a temporary bit of pain or a fever during pregnancy.
For the other 17 years, additional information came from subsequent interviews done throughout the pregnancy, though how many of these interviews were done is not stated. Starting in 2005, they also added information from a prescription drug register, although given that acetaminophen is commonly purchased over the counter, I suspect this additional information is unlikely to make a big difference.
If you feel a headache coming on, you’re not alone. Documenting who took acetaminophen during pregnancy, and how many pills, and for how long is not easy. Based on the imperfect data we have, professional bodies state that acetaminophen is still the safest drug to relieve pain and fever during pregnancy, though better studies would be welcomed. Acetaminophen is unlikely to be a risk factor for autism: if it is, it’s a small one. It is very likely that the associations seen in positive studies are simply mirages.
And while we are fretting over a risk that probably does not exist, we ignore of fevers during pregnancy. They increase the risk of miscarriage when untreated, especially during the first trimester, as well as the risk of birth defects and premature birth. Meanwhile, untreated pain can lead to maternal high blood pressure, depression, and anxiety. Risks and benefits need to be weighed not just for treating a condition, but also for not treating it. Warning against the use of acetaminophen in the absence of solid evidence is maddening because it accuses women of being bad mothers and asks them to suffer unaided during pregnancy.
Demonizing acetaminophen right now is not warranted, but deifying leucovorin as a treatment for autism? That’s an even bigger stretch.
The folate fix
The Washington Post that, last August, a meeting took place between the head of the National Institutes of Health, Dr. Jay Bhattacharya, and a scientist named Richard Frye. Frye and colleagues met Bhattacharya to discuss “autism interventions that they felt were not getting the attention they deserved.” I wonder if Frye dropped a copy of his book, , on his way out.
There are many pet theories about what really causes autism, with some involving the immune system, the consumption of dairy, the microbiome, or exposure to pesticides. Frye’s hinges around a vitamin.
Folate (also known as vitamin B9) is important in the body, especially in the nascent brain, and Frye believes that in some people with autism, the immune system produces antibodies that block an important receptor for folate. It’s like trying to unlock a door when a piece of gum is jamming the lock: folate can’t reach the brain in sufficient quantities. He and his colleagues have tested giving a form of folate called leucovorin (also known as folinic acid) to children with autism to see if their symptoms would improve. (Many of these studies were funded in part by Autism Speaks, an organization accused for many years of .)
I looked at a dozen papers on this topic and the evidence is not convincing at all. What I believe to be on this had a measly two dozen children in each group (leucovorin versus placebo), and the score improvements on communication skills between the two groups overlapped. Trials run by other groups were similarly small and either showed or , finding a difference for only two of them when using leucovorin. This is not improvement; it’s noise.
What is revealing is that Trump’s lackeys, so quick to accuse previous iterations of the FDA of being hopelessly corrupt, are now using these meager crumbs to force the FDA into approving leucovorin for children with autism, with the FDA commissioner himself calling the link between the two “fairly established.” This is not gold-standard research; this is incompetence dressed up as accomplishment. (While Dr. Oz mentioned that leucovorin would be available with a prescription, it is also being sold as a dietary supplement at much lower doses. Dr. Oz is from a company that sells the supplement.)
On our side, though, we should be careful not to commit the genetic fallacy: rejecting what Trump and RFK Jr say without looking into it simply because we know these people to be grossly inept. Cherry-picking the studies we like without criticizing them too much is a problem across the political aisle. At the end of the day, it all boils down to the primary sources we have and how good the evidence is.
The American College of Obstetricians and Gynecologists’ academic journal published last February on the topic of Tylenol in pregnancy. Its authors acknowledged the issue with the Swedish study but concluded that acetaminophen taken during a pregnancy is “unlikely” to increase the risk of autism in a significant way, and they endorsed “the principle of lowest effective dose for the shortest duration of time for treatment with acetaminophen in pregnancy.”
That is much more sound and evidence-based than Donald Trump the following from the Roosevelt Room of the White House: “But with Tylenol, don’t take it. DON’T TAKE IT! And if you can’t live, if your fever is so bad you have to take one, because there is no alternative to that, sadly. First question, what can you take instead? It’s actually, there is not an alternative to that. And as you know, other of the medicines are absolutely proven bad, I mean, they’ve been proven bad with the Aspirins and the Advils and others, right?”
Don’t take medical advice from the orange man in the White House.
P.S.: Dr. Oz in an appearance on broke ranks with Trump and now recommends acetaminophen during pregnancy for high-grade fever and “when it’s appropriate.”
Take-home message:
- Acetaminophen (also known as paracetamol or by the brand name Tylenol) is still the safest pain and fever medication to use during pregnancy
- While some studies show a small increase in the odds of having a child with autism when acetaminophen is taken during pregnancy, these studies are unreliable because of a long list of shortcomings
- A large amount of scientific evidence points out to autism being mainly caused by variations in certain genes, not by environmental factors
- Leucovorin (folinic acid) is being approved as a prescription drug to treat certain symptoms of autism in the United States, but the scientific evidence behind this is of terrible quality
