Why we need to experiment on pregnant women

The research on Tylenol is only OK. We can settle the debate.

President John F Kennedy speaks about the emerging health crisis over thalidomide (Photo by Abbie Rowe/PhotoQuest/Getty Images)

The Trump administration is using shoddy evidence to discourage pregnant women from using Tylenol. On Sept. 22, 2025, the FDA moved to initiate a label change for acetaminophen due to data showing a correlation between prenatal exposure and autism. These actions have led to uncertainty and angst among pregnant women, emphatic reassurances from professional organizations, and, most recently, a questionable lawsuit against the makers of Tylenol by Texas’ attorney general.

We have ample observational evidence — including newer and better studies than those cited in the announcement — demonstrating Tylenol’s safety and medical necessity. But even these studies fall short of the gold standard that could put concerns to rest because of one nagging problem: It’s unreasonably difficult to include pregnant women in clinical trials. As a result, pregnant women, like me, are often left agonizing over decisions about whether to take medications we need.

Some critics of this policy shift simultaneously lamented that there was no ethical way to collect better data. For instance, New York Times science journalist Azeen Ghorayshi reported that “Most experts believe that it would be unethical to perform pharmaceutical research on pregnant women.” On The Daily, Ghorayshi argued that “We can’t conduct randomized controlled trials on pregnant women. It is widely held to be unethical to do something like that.”

As a physician, bioethicist, and expectant mother, I disagree: We can and should conduct experiments with pregnant women. In fact, we already are. The only question is whether they should be randomized controlled trials (RCTs) or ad hoc studies with little generalizability. Only 5% of drugs approved between 2003 and 2012 had human pregnancy data, despite the fact that 90% of women take medications during pregnancy. That means pregnant patients are making decisions about whether to continue or start medications based on limited data.

There are risks in both directions. While taking under-researched medications can impose risks on a pregnant woman and her unborn child, stopping treatment for a medical condition, like diabetes or depression, also comes with risks to both. Discontinuing medication puts pregnant women at increased risk of having an undertreated illness during a nine-month period when many face additional physiological, financial, social, and emotional stressors.

What do we know about the safety of Tylenol?

The Trump administration’s Tylenol skepticism is officially based on two observational studies; in other words, neither study randomly assigned pregnant patients to take acetaminophen or a matching placebo.

One of the studies found that patients with higher levels of acetaminophen metabolites in their umbilical cord plasma at the time of birth were more likely to have children subsequently diagnosed with autism.