Making Sense of the Acetaminophen and Autism Controversy

The news has recently been awash with stories about the link between acetaminophen (paracetamol) and autism. The White House issued new guidelines to pregnant women to avoid taking the drug and use “the lowest effective dose for the shortest duration when treatment is required.”

The key evidence presented by the Trump administration is a new analysis, carried out by Harvard scientists, of multiple studies investigating neurodevelopmental disorders in children, which reported a higher incidence when mothers had taken paracetamol during pregnancy (Prada et al. 2025).

This paper is the latest in a long line of studies into this contentious subject. For a couple of decades, the medical literature has hosted a robust debate about the effects of paracetamol on neurodevelopment.

Some studies find a correlation with various aspects of health at birth, others find no significant effect. Some mechanistic studies provide plausible ways that the drug might disrupt normal development; others point out the multiple confounding genetic and environmental factors that are hard to control for. The risks to the foetus of taking paracetamol are weighed against the risks of leaving fever or pain untreated.

It’s the classic, inevitable outcome of trying to understand the role of a single factor in a hugely complicated scenario. The results are noisy, confusing, and disputed.

What does that mean when it comes to advising concerned patients?

Well, most medical authorities settled on the conclusion that, given the ambiguous dataset and the lack of alternative safe painkillers for pregnant women, the old advice to take paracetamol when needed should stand.

The new guidelines instead take a more precautionary approach: Limit paracetamol use as far as possible.

To an apolitical observer, this might seem to be a fairly modest change in advice, but the reaction to the news has been far more psychologically interesting than that.

Collapse of epistemic trust

The concept of “epistemic trust” (Li et al. 2023) captures the idea that we all need to accept some non-intuitive ideas or opaque information to thrive in a complex world. We learn to trust authorities or institutions as reliable sources of information on topics for which we cannot reasonably develop our own expertise.

Medical information is a classic case. Even with sufficient time and access to information, most people could not assess the likelihood that paracetamol increases the risks of developing autism. We need to trust the institutions that act as intermediaries between the public and the experts to give an honest and accurate summary of the medical consensus.

It’s fair to say that there has been a collapse of epistemic trust in recent years, on both sides of the political aisle.

How else can we explain the TikTok trend of pregnant women theatrically popping Tylenol pills to prove they “trust the science, not Trump” or the YouTube conspiracy theorists claiming SARS-CoV-2 was developed as a bioweapon in covert laboratories in Ukraine?

We seem to have entered an era where people are using alignment with political tribe as a new proxy for epistemic trustworthiness.

How to regain trust

Putting aside the politics (if that’s possible), is there a way that academics and clinicians can regain some of the lost epistemic trust?

One thing within our control is how we communicate directly to the public.

To judge from the response to the White House announcement, some medical authorities seem to have made the calculation that a clear, robust counter-message needs to be delivered.

The president of the Autism Science Foundation stated, “No new data or scientific studies were presented or shared. No new studies have been published in the literature. No new presentations on this topic were made at scientific or medical conferences. Instead, President Trump talked about what he thinks and feels without offering scientific evidence.”

The chief safety officer at the UK Medicines and Healthcare Products Regulatory Authority said, “Patient safety is our top priority. There is no evidence that taking paracetamol during pregnancy causes autism in children.”

These statements are unambiguous, for sure, but they damage epistemic trust. There plainly is evidence that paracetamol could cause harm. Researchers are trying to sort out how robust the evidence is, how serious the risk is, how severe the effect could be, and how it ranks against other factors (including not managing fever effectively).

Politics Essential Reads

Finding Common Ground in a Divided Nation

To Boldly Go: A History of Human Migration

The general consensus seems to be that the risks are small and the evidence mixed, and so, the balance of probability suggests treating fever and pain during pregnancy with paracetamol is prudent. The new administration is an outlier in arguing for more cautiousness.

By using a binary statement like “There is no evidence,” neutral institutions diminish their trustworthiness with the wider public, who can read for themselves the many newspaper articles and scientific papers discussing the contested evidence over the last few years.

Nuanced messaging might have less clarity and impact, but it is perceived as more trustworthy (Hendriks et al., 2023).

By seeking to persuade rather than inform, institutions risk worsening the political polarisation that plagues healthcare debates.