The debate over a possible link between acetaminophen and autism was a key topic at the 42nd annual conference of the International Society for Pharmacoepidemiology, held in Milan, Italy, from August 29 to September 2, 2026.

In 2025, statements by US government officials about the possible risk for autism in children associated with the use of acetaminophen during pregnancy drew substantial media attention and raised public concerns. Acetaminophen has traditionally been considered safe during pregnancy; therefore, reports of its possible association with neurodevelopmental problems understandably worried many expectant mothers. Several scientific societies and drug regulatory agencies, including the International Federation of Gynecology and Obstetrics, European Medicines Agency, and Italian Medicines Agency, promptly published statements and documents to clarify the evidence. However, these scientific reassurances did not quell the media uproar.

Experts participating in the 2026 session assessed the available evidence, discussed the most appropriate research methods for studying the association, and emphasized accurate communication with the public and patient involvement in research. The session was designated a “Hot Topic” at the meeting.

“The patient should be heard as an equal by all experts involved in research, from the earliest stages of the work, including study design, to understand their actual needs,” said Annalisa Gebbia, MBA, CSCP, patient expert from the EUPATI Training Programme, an educational program by the European Patients’ Academy on Therapeutic Innovation that teaches patients and advocates how medicines are researched, developed, and regulated. She is also a member of the Italian Association of Patients with Cardiomyopathies in Milan, Italy, and participated in the discussion.

Data Speak

An informed discussion must first establish what the available data actually show. Krista F. Huybrechts, PhD, professor of medicine at Harvard Medical School and professor of epidemiology at Harvard T.H. Chan School of Public Health, as well as an epidemiologist in the Division of Pharmacoepidemiology and Pharmacoeconomics at Brigham and Women’s Hospital in Boston, provided an overview of the main studies examining the topic.

Huybrechts noted that interest in a possible association between prenatal acetaminophen exposure and autism spectrum disorder increased following ecological studies published more than a decade ago, including a study published in Environmental Health. However, ecological studies have several limitations. Assessing this association is particularly challenging because of possible exposure misclassification and, more importantly, confounding by indication and maternal, genetic, and clinical factors.

Available data show that acetaminophen is widely used during pregnancy, but typically for short periods. In most cases, it is an over-the-counter medication used to treat headaches and migraines, whereas its use for fever appears to be considerably less common. Women who use acetaminophen during pregnancy also often have comorbidities, including obesity, mental health disorders, and hypertension, which may influence neurodevelopmental outcomes in their children.

Available studies can be broadly divided into two groups: prospective studies and more recent cohort studies based on administrative data. The first group has substantial methodological limitations and considerable heterogeneity across study populations and designs, resulting in important differences in the findings. The second group includes very large cohorts, with up to approximately 2.5 million pregnancies. These studies have primarily used prescription records to define exposure and International Classification of Diseases codes to identify outcomes using different study designs.

A comparison of these studies shows substantial variation in exposure prevalence, ranging from 2.1% to almost 48%. There is also no consistent evidence of a dose-response relationship or specific periods of susceptibility during pregnancy. Similar associations have been observed with other analgesics in addition to acetaminophen.

Overall, the available evidence supporting a link between prenatal acetaminophen exposure and an increased risk for autism spectrum disorder remains weak and does not establish a causal relationship. Huybrechts also noted that even a small increase in relative risk can have a substantial population-level impact when exposure is common. This is another reason to carefully evaluate the potential association between acetaminophen and autism, beginning with the study design.

Viktor H. Ahlqvist, PhD, researcher at Karolinska Institutet in Stockholm, Sweden, focused on studies that compare members of the same family. This approach can help control confounding factors such as genetic and environmental factors shared within a family. Furthermore, family members also tend to be more similar than unrelated individuals with respect to many unmeasured characteristics.

Ahlqvist specifically discussed the advantages and limitations of the so-called sibling design, which compares siblings. As one of his key take-home messages, he emphasized that “comparing siblings is not automatically more robust than a conventional analysis: It may offer greater protection against certain bias structures but prove less effective against others, and its value therefore depends on the specific research question and the structure of the expected biases.”

Sonia Hernández-Díaz, MD, DrPH, professor of epidemiology in the Department of Epidemiology at Harvard T.H. Chan School of Public Health, described different study designs and focused on potential sources of systematic errors that can arise at different stages of research.

Several questions remain unanswered in studies of acetaminophen and autism. For example, why have some studies identified an association despite problems with exposure misclassification? If acetaminophen itself is not responsible for the observed association, what may explain these findings?

Hernández-Díaz also raised an important possibility: “If migraine or pain are significant confounding factors, their role could provide clues for identifying potential preventable causes of autism spectrum disorders.”

In addition to epidemiologic studies, decision support tools such as Teratogenic Risk Impact and Mitigation (TRIM) are available. Almut G. Winterstein, PhD, director of the Center for Drug Evaluation and Safety at the University of Florida in Gainesville, Florida, presented the tool, which uses explicit and quantifiable criteria to prioritize teratogenic drugs and identify potential risk mitigation strategies.

The TRIM assessment indicated that the evidence for a possible risk associated with acetaminophen is weak and that the risk attributable to the drug is low. Consequently, acetaminophen is not considered a priority for risk management despite its widespread use.

Winterstein clarified that this assessment does not address individual risks or decisions that individual patients need to make. She also emphasized that the TRIM criteria, metrics, and weightings need to be replicated or calibrated in countries outside the US.

Having the data is not sufficient. Communicating the findings accurately is equally important when addressing health policy decision-makers, physicians, or the public.

Science journalist Roberta Villa, regular contributor to Univadis Italy, part of the Medscape Professional Network, offered insights into communicating health information accurately and effectively outside research settings and expert meetings.

Communication is not simply a matter of reporting data. Communication is “framed” by the source through the context and language used and is also filtered through the trust that individuals place in that source. Trust is built through transparency, dialogue, and respect. Simultaneously, individuals interpret messages through their own beliefs, identities, values, and experiences.

Risk communication has another important element: fear. Villa explained that fear “drives us toward a preference for ‘greater safety,’ even when the seemingly safer choice is bound to prove worse in the long run.”

Effective risk communication, therefore, begins with empathy and respect, not with detailed data. Her final message to those communicating sensitive topics such as acetaminophen and autism was “KISS — Keep It Short and Simple.”

Speaking with Univadis Italy, Villa said, “If the general public and journalists had attended this session and seen the seriousness and attention the scientific community gives to even the slightest indication of a potential risk from a drug considered safe, such as acetaminophen, I think they would feel very reassured.” She urged researchers not to be afraid to communicate with the public and journalists.

“What matters is explaining the context clearly and with extreme transparency; this also helps prevent information from being misused,” she added.

Huybrechts reported serving as principal investigator for research grants awarded to her institution by UCB, Takeda, and GSK for studies required by regulatory authorities and unrelated to the work presented.

Ahlqvist reported receiving speaker honoraria from Angelini Pharma S.p.a., professional fees from the Organization of Teratology Information Specialists, funding from the Swedish Society for Medical Research, a nonprofit foundation, and funding from Karolinska Institutet, which maintains university-wide collaborations with industry.

Hernández-Díaz reported consulting for Vertex Pharmaceuticals and Roche during the past 12 months on unrelated topics; participating as an investigator or researcher in projects funded by Takeda, GSK, and UCB; and serving as a consultant for the North American Antiepileptic Drug Pregnancy Registry and the Antipsychotics Pregnancy Registry, both of which are funded by various companies.

Winterstein reported having no actual or potential conflicts of interest or financial relationships to disclose in relation to the content of the presentation. Development of the TRIM tool was funded by the FDA.

Villa and Gebbia reported having no conflicts of interest in relation to the presentation.

This story was translated from Univadis Italy.