Large Study of 700,000 Pregnancies Finds No Link Between Acetaminophen and Autism or ADHD

What the study found

Researchers analyzed electronic health records spanning 2001 to 2023 for more than 700,000 mother-child pairs in Hong Kong. About 43 percent of those children had been exposed to acetaminophen — sold in the U.S. as Tylenol and also known as paracetamol — while in the womb.

The key methodological move was a sibling-matched analysis. Researchers compared autism and ADHD rates between siblings where one was exposed to acetaminophen in utero and one was not. That design strips out shared genetic and environmental factors that could otherwise muddy the results. The autism cohort covered more than 124,000 sibling-matched children; the ADHD cohort covered more than 97,000.

The result: no link between prenatal acetaminophen use and either condition. Dosage didn't matter. Trimester didn't matter. Frequency didn't matter. Mother's age didn't matter.

Why earlier studies looked different

Some earlier research had found a correlation — and this study actually reproduced that apparent signal when researchers dropped the sibling-matching design and simply compared exposed versus unexposed children from different families. That's a weaker study design because it can't control for genetics or shared household conditions.

The Hong Kong team stress-tested this by running a "negative control" analysis: they looked at mothers who had taken acetaminophen before pregnancy or after giving birth and compared them to mothers who never used it. That group also showed an association with autism, which makes no biological sense since acetaminophen taken before conception or after delivery cannot affect fetal development.

The researchers concluded in JAMA Internal Medicine that the apparent signal in conventional analyses "reflect residual familial confounding, rather than a true pharmacologic effect." Families that use more acetaminophen may have other characteristics — genetic, behavioral, environmental — that independently elevate autism risk. The drug itself is not the cause.

The Trump-Kennedy press conference and its fallout

In September 2025, President Trump and HHS Secretary Robert F. Kennedy Jr. held a press conference asserting, without clear supporting evidence, that acetaminophen taken during pregnancy causes autism. Trump told pregnant women to "tough it out" rather than take Tylenol for fever or pain.

Medical organizations pushed back immediately, pointing out that untreated fever during pregnancy is itself a documented risk factor for autism, miscarriage, birth defects, and premature birth. Telling a pregnant woman to endure a high fever rather than treat it with a well-studied, widely used medication contradicts established medical guidance.

The warning had measurable effects. A study published in The Lancet in March 2026 found that acetaminophen use among pregnant patients in emergency departments fell 10 percent following the press conference. Texas also filed a lawsuit against the maker of Tylenol over the alleged autism connection.

The case for caution

Some researchers and policymakers have raised questions about acetaminophen and fetal neurodevelopment. A number of observational studies over the past decade did find correlations. Acetaminophen does cross the placental barrier. And the mechanism by which it could theoretically affect fetal brain development — its mild hormonal effects — is biologically plausible enough that some scientists argued for more investigation.

The issue is that Trump and Kennedy didn't call for more research. They stated a causal conclusion without the evidence to back it up, in a public forum, in a way that produced measurable changes in medical behavior.

What the science actually supports

This Hong Kong study is large, methodologically rigorous, and consistent with the broader scientific consensus. The sibling-matched design is considered one of the more reliable tools for ruling out familial confounding in observational research. The negative control finding — that pre-pregnancy and postpartum acetaminophen use showed the same spurious "association" — is a strong signal that earlier studies were measuring something about certain families, not about the drug.

No study is definitive on its own. This one adds substantial weight to a body of evidence that has repeatedly failed to find a causal link between acetaminophen and autism.

The unresolved question is a practical one: the Lancet data shows that the September 2025 press conference already changed behavior among pregnant patients in emergency departments. Researchers have not yet measured whether that behavioral shift translated into any change in maternal or fetal health outcomes — including rates of untreated fever-related complications — and that follow-up study has not yet been published.