The Tylenol-Autism Debate Isn’t Over. A New Review Also Challenges the ADHD Link

A 2019 study made the Tylenol question hard to ignore: children with the highest levels of acetaminophen detected in umbilical-cord blood had substantially higher odds of autism and ADHD.

Then, a Swedish study of nearly 2.5 million children found something very different when researchers compared siblings.

Now, a new review published in September 2026 has taken a closer look at why the findings have differed—and found that the apparent association weakens when researchers give greater weight to studies with stronger methodology.

For pregnant patients, that distinction matters.

The Study That Raised The Alarm

In 2019, researchers studying 996 mother-child pairs in the Boston Birth Cohort measured acetaminophen biomarkers in umbilical-cord blood.

Compared with children in the lowest exposure group, those in the highest group had greater odds of ADHD, with an odds ratio of 2.86, and autism, with an odds ratio of 3.62.

The numbers were striking, but they did not prove that acetaminophen caused either condition.

The study was observational, and acetaminophen was detected in every cord-blood sample. The researchers were therefore comparing higher versus lower biomarker levels, rather than children exposed to the drug versus children with no exposure at all.

Other observational research also reported associations. A 2014 Danish study of 64,322 mother-child pairs found prenatal acetaminophen use was associated with a 37% higher hazard of hospital-diagnosed hyperkinetic disorder, an ADHD-related diagnosis.

But observational associations leave an important question unanswered: could something about the families taking acetaminophen, rather than the medication itself, help explain the results?

The 2.48 Million-Child Study Changed The Picture

Researchers addressed that question in a nationwide Swedish study published in JAMA in 2024.

The study included 2,480,797 children and used sibling comparisons to examine prenatal acetaminophen exposure and autism, ADHD, and intellectual disability.

The conventional analysis showed small associations. But when researchers compared siblings, those associations disappeared.

The sibling analysis found a hazard ratio of 0.98 for autism, 0.98 for ADHD, and 1.01 for intellectual disability.

Brian Lee, PhD, a co-senior author of the study, explained why sibling comparisons matter: “Sibling comparisons allow us to control for familial characteristics that might explain an apparent relationship between acetaminophen use during pregnancy and risk of neurodevelopmental conditions.”

The design cannot eliminate every possible confounding factor, but it can account for some characteristics siblings share that ordinary observational comparisons may not fully capture.

Related: 7 Signs of Autism in Adults That Often Go Unnoticed

The New Review Looked At The Quality Of The Evidence

The September 2026 review, led by Hoda Tayebi-Hillali of the University of Santiago de Compostela, examined seven previous meta-analyses and additional analyses of their underlying studies.

The evidence involved more than 3 million participants, although some participants appeared in more than one analysis.

When researchers pooled the evidence without restricting it by methodological quality, the risk ratio was 1.17 for autism and 1.29 for ADHD.

But the estimates became smaller when the researchers focused on stronger studies.

Among primary studies considered to have a low risk of bias, the pooled risk ratio fell to 1.05 for autism and 1.17 for ADHD.

At the highest review-quality level, the autism risk ratio was 1.10, with a 95% confidence interval of 0.98 to 1.24. Because that interval included 1, the result did not establish a statistically significant association.

The researchers concluded that the apparent association weakened as methodological quality improved.

Tayebi-Hillali said the findings “challenge the interpretation of a causal link.”

Why Earlier Studies Reached Different Conclusions

The studies did not all measure acetaminophen exposure in the same way.

The 2019 Boston research used umbilical-cord biomarkers. The Danish research relied on reported medication use and medical records. The Swedish study used nationwide health and prescription records and added sibling comparisons.

The outcomes also varied. Some studies examined ADHD-related symptoms or diagnoses, while others looked at autism-spectrum symptoms or clinical diagnoses.

Those differences matter because observational research has to account for factors that may be associated with both medication use and a child’s later health.

A November 2025 umbrella review published in The BMJ reached a similarly cautious conclusion. It examined nine systematic reviews covering 40 primary studies and rated the confidence in two reviews as low and seven as critically low. The authors concluded that the existing evidence did not clearly link prenatal acetaminophen use with autism or ADHD.

At the same time, an August 2025 systematic review in Environmental Health reached a more precautionary interpretation, finding evidence supporting an association while acknowledging observational limitations and possible residual confounding.

The disagreement is therefore not simply about whether one side believes the drug is “safe” and the other does not. It is also about how much confidence researchers should place in different types of observational evidence.

What About The Warnings Pregnant Women Have Heard?

The issue became even more controversial in September 2025, when President Donald Trump urged pregnant women not to take Tylenol, saying, “Don’t take Tylenol. Don’t take it.”

But his statement did not establish that acetaminophen causes autism.

The FDA announcement accompanying the administration’s action acknowledged that although an association had been described in studies, “a causal relationship has not been established” and that contrary studies exist.

That distinction remains important.

Pain And Fever Are Part Of The Decision

Image Credit: milkos via 123RF

The question for a pregnant patient is also not simply whether a medication might carry a potential risk. It is why the medication is being considered.

Steven J. Fleischman, identified as president of the American College of Obstetricians and Gynecologists in its September 2025 statement, said acetaminophen is one of the few options available to pregnant patients for treating pain and fever, “which can be harmful to pregnant people when left untreated.”

He also emphasized the importance of considering “all potential risks along with any benefits.”

That does not amount to a recommendation for every pregnant patient to take acetaminophen. It underscores why medication decisions during pregnancy involve weighing more than one potential risk.

What The New Review Actually Shows

The September 2026 review does not prove that acetaminophen has zero risk during pregnancy.

It also does not establish that prenatal acetaminophen exposure causes autism or ADHD.

What it does show is that the apparent associations became weaker when researchers placed greater weight on methodological quality. That finding is consistent with the 2024 Swedish sibling study, where associations seen in conventional analyses disappeared when researchers compared siblings.

For pregnant patients, the most defensible takeaway is narrower than many of the headlines surrounding this debate:

Current evidence has not established that taking acetaminophen during pregnancy causes autism or ADHD.

Anyone who is pregnant and considering medication for pain or fever should discuss the decision with an obstetric clinician rather than making a sudden change based on a headline, social-media post, or political statement.

Have conflicting health headlines ever left you unsure about what to believe?