As someone trained as a scientist and who teaches science at Michigan State University, watching the Sept. 22 press conference in which President Trump and DHS Secretary Robert Kennedy Jr. alleged a link between autism and Tylenol use during pregnancy was as cringe-inducing as seeing students submit low-quality assignments written with ChatGPT.
The claim that acetaminophen causes or increases the risk of autism is based on inconclusive science, at best. Using the platform of the presidency to spread dubious information will harm mothers, harm children, and further erode the public’s trust in science.
Responsibly using science to guide public policy, especially when it comes to medicine, is more than just holding up whatever study confirms one’s pre-existing beliefs. Knowledge of a field is often required to distinguish well-constructed, rigorous research from sloppy studies. President Trump and Secretary Kennedy cited a 2025 paper that reviews the results of seven previous studies investigating connections between acetaminophen use and autism. Of those studies, three reported no significant association, three reported slight correlations, and one reported a more significant correlation but had a low sample size compared to the other studies— limiting its ability to account for other variables that could have affected their results. It’s worth noting that the senior author on that review, Andrea Baccarelli, has previously served as a paid witness for plaintiffs in (unsuccessful) civil suits against Tylenol’s manufacturer.
The paper cited by Trump and RFK also downplays the results of the single largest study of acetaminophen use and autism risk, published in 2024 by researchers from Sweden and Drexel University in Pennsylvania. That study analyzed almost 2.5 million children born in Sweden between 1995 and 2019. The researchers also accounted for cases in which a woman took acetaminophen while pregnant with one child but didn’t take it while pregnant with another. If acetaminophen increased risk of autism, you would expect to see more autism diagnoses in the “exposed” siblings than in the unexposed ones. However, the study’s authors found no such correlation. Research conducted in Norway has also indicated that certain genetic characteristics that make women more likely to have children with autism or ADHD also lead to increased pain and more frequent migraines during pregnancy, which would make those women more likely to need pain-relieving medication while pregnant. In other words, factors that lead to autism may be prompting more Tylenol use among pregnant women. That could show association, while pointing against causation.
Even if you only accept the studies finding an association between Tylenol use and autism, they only showed a 19 percent increased risk of autism in children exposed to acetaminophen during pregnancy. To be clear, increasing the risk of a condition by 19 percent doesn’t increase the total probability of that condition occurring in a child from, say, two percent to 21 percent. Since the percentage rise is based on the original probability, it would, in this instance, increase the risk from two percent to 2.38 percent. A 19 percent increased risk of autism would not come close to explaining the rise in autism diagnosis rates that Trump and RFK Jr. have pointed to in recent months.
It’s true that these rates have dramatically increased, from 1 in 2,500 children in 1966 to 1 in 31 in 2022, according to data from the Centers for Disease Control and Prevention.
A sizable percentage of that rise can be attributed to changes in educational policy, recommended medical practice, and diagnostic criteria that took place during the 1990s through the 2010s. More children now get screened and the definition of autism is now broader than it was in the past, and so more cases get caught and diagnosed. Sure enough, this increase is primarily driven by increased diagnoses of mild or “non-profound” autism. These are people who live normal lives but for a bit of social awkwardness and who would have never been diagnosed in the past. The Centers for Disease Control and Prevention tracked the rise in both mild and “profound” autism, which is autism combined with an intellectual disability or severe speech disability. They found that mild autism diagnoses increased by 3.7 times between 2000 and 2016, from one in 256 children to one in 70. In contrast, profound autism diagnoses increased only by 1.7 times over that same period, from one in 370 to one in 217.
So, what does cause autism? Scientists have known for a long time that autism is primarily a genetic condition, because identical twins (who share 100 percent of their genes) are much more likely to both have autism compared to fraternal twins (who only share 50 percent of their genes on average, being no more alike than any other siblings). That said, there’s no one “autism gene” responsible for it. More than 200 genes have been identified that can contribute to it, and parental exposure to certain chemicals in the environment or workplace can interact with some of those genes to further increase a child’s risk.
Notably, high fevers during pregnancy have long been known to increase risk of autism in addition to being very dangerous in other ways for a developing fetus. Acetaminophen is the only pain and fever reducing drug that doctors routinely prescribe as safe for pregnant women. Others, like aspirin, naproxen, and ibuprofen come with well-documented risks.
For a president and a cabinet secretary, neither of whom has any scientific background, to blame mothers for using the one form of pain relief most prescribed by doctors as the cause of their child’s autism is irresponsible. If this announcement leads to pregnant women going without fever relief, it’s likely to lead to more tragic outcomes for their children.