One evening in 2019, when I was a pediatrics resident, I admitted a two-month-old to the hospital for observation after a minor surgery. I explained to the baby’s mother that I planned to order acetaminophen—commonly sold over the counter as Tylenol—every six hours, because the baby had an obvious source of pain. If pain still kept the baby from eating, sleeping, or calming down, the mother could ask for an opioid. I was just leaving the room when the mother stopped me to ask about the acetaminophen.
“Doesn’t it cause autism?” she said.
“I’m not familiar with any research linking Tylenol to autism,” I told her. “But I’ll look into it and get back to you.” In the meantime, we agreed to use both the acetaminophen and the opioid as needed, instead of administering them on a schedule.
When I sat down at a hospital computer, I was surprised to find that the mother’s question had a basis in mainstream research. Small studies had associated acetaminophen exposure in utero with a baby’s risk of developing autism. But this wasn’t the same as saying that Tylenol caused autism. Perhaps whatever the drug was treating—for example, fevers, infections, or painful chronic conditions—contributed to autism, and acetaminophen did not.
For babies under three months, as for pregnant women, Tylenol is considered the safest medicine for fever. (Ibuprofen and similar medications, NSAIDs, have been associated with kidney injury in babies.) I continued to order it for patients, if a little more cautiously. Then, in 2024, a more rigorously controlled study of more than two million children born in Sweden found no connection between acetaminophen and neurodevelopmental disorders. As the epidemiologist Katelyn Jetelina has written, “the evidence leans heavily towards correlation, not causation. (Tylenol is not the cause.)”
Last week, during a press conference, President Trump contradicted existing research by urging women to “tough it out” and avoid Tylenol during pregnancy. “Fight like hell not to take it,” he said. On Friday, he wrote on Truth Social, “DON’T GIVE TYLENOL TO YOUR YOUNG CHILD FOR VIRTUALLY ANY REASON.” Medical groups disagreed. The American College of Obstetricians and Gynecologists (ACOG) advised women to continue using Tylenol when needed. “Misleading claims that the medicine is not safe and is linked to increased rates of autism send a confusing, dangerous message to parents and expectant parents and does a disservice to autistic individuals,” the American Academy of Pediatrics said.
Now doctors are reporting that pregnant women are hesitating to use the medicine, even when professionals recommend it for pain or for fever. One reason that Trump’s claims are difficult to dispel is that he makes them sound certain. Doctors, in contrast, can say that the strongest research does not show a link between Tylenol and autism, and that medical professionals consider it the best option for pain and fever in pregnancy. But we know that there is uncertainty in medical research, and so we speak with care.
It ought to be possible for doctors to be more definitive. Pregnancy is not a rare condition—millions of people get pregnant each year—yet those who experience it are frequently told that there’s not enough research to guarantee that a medicine is safe. When a pregnant woman needs medicine, whether for lupus or for high blood pressure, she may feel that she faces an impossible choice: suffer through a condition that may itself harm her or her baby, or else allow an uncontrolled experiment inside one’s own body. Isn’t there a better way?
Many of the rules that govern research on human beings date back to the Nuremberg Code, a response to Nazi doctors who conducted brutal experiments in concentration camps. To this day, research participants must consent; trials must be stopped if there is evidence of substantial harm. In the nineteen-sixties, the public became aware that thalidomide, a widely used nausea drug, had caused birth defects in an estimated eight thousand children outside of the United States. The drug became a case study for the growing field of bioethics. In 1977, the F.D.A. excluded not only pregnant women but also women “of childbearing potential” from early-stage clinical trials, which focus on safety and toxicity. Drugs are instead tested for toxicity in pregnant animals—usually, rats and rabbits—which often respond very differently than humans.
According to Ruth Faden, the founder of the Johns Hopkins Berman Institute of Bioethics, these efforts to protect women and babies had unintended consequences. In fact, she said, public responses to thalidomide were a “misreading of history.” Thalidomide wasn’t systematically tested on pregnant women. If it had been, then its risks likely would have been discovered sooner, and fewer children would’ve been affected.
In the nineteen-nineties, the F.D.A.’s exclusion of women “of childbearing potential” ended, and the National Institutes of Health established guidelines to encourage women in general to participate in research. In 2018, the Common Rule, which establishes scientific practices for twenty federal agencies, was revised to make it easier for researchers to include pregnant women. But, in 2024, a report from the National Academies of Sciences, Engineering, and Medicine asserted that that “very little progress has been made on research involving pregnant and lactating women.” Too often, Faden told me, pregnant women are still viewed as fragile vessels who need to be sequestered from the world. “We need to protect women through research, not from research,” she said.
Thoughtful safety protocols would be necessary to ethically include pregnant women in research studies. Medicines whose mechanism could plausibly harm a fetus—those that inhibit crucial nutrients such as folate, for example, or that stop cells from dividing—would be excluded. So would drugs that have worrisome effects in pregnant animals. As in any research study, participation would need to be voluntary, and the trial would need to be more likely to benefit women than to harm them. If scientists started to suspect that women or their babies were being harmed, the trial would need to be paused or stopped altogether.