Ice cream sales and drowning incidents both rise during the summer months. Should we assume that ice cream causes drowning? Probably not. Both are influenced by a third factor – hot weather. When the temperature climbs, people buy more ice cream and also spend more time swimming or boating, which unfortunately leads to more drownings.
This is a classic example of correlation without causation. Statisticians call “hot weather” a confounding variable, which influences both sides of the equation and creates the appearance of a relationship where none actually exists.
This same principle is relevant to the ongoing public confusion about whether taking acetaminophen (Tylenol) during pregnancy causes autism in children. Most of us in public health and medicine believe there is a correlation, but not a causal relationship.
Pregnant women commonly take acetaminophen to relieve fever or pain. Other pain relievers such as ibuprofen (NSAIDs-Non-Steroidal Anti-Inflammatory Drugs) are generally avoided during pregnancy, especially after 20 weeks, because they can increase the risk of congenital anomalies, fetal kidney injury, and premature closure of blood vessels that are important for fetal heart and lung development. In addition, untreated high fever itself is known to increase the risk of fetal harm, including associations with cleft lip and palate as well as neurodevelopmental disorders¹. For these reasons, acetaminophen remains the safest and most widely recommended medication for pain and fever during pregnancy.
Autism is not a single, uniform condition. Because symptoms vary widely in type and severity, scientists now refer to it as autism spectrum disorder (ASD) — a developmental condition that affects how people communicate and interact with others. One aspect that researchers broadly agree on is that autism has a strong genetic component. Hundreds of genes — and the timing of their activation during brain development — appear to influence whether and how autism manifests2,3. Because of this, any study that attempts to link an external factor, such as medication use during pregnancy to autism must carefully control for genetics. In this case, heredity is the potential confounding factor.
The gold standard in medical research is the randomized controlled trial (RCT). In an RCT, participants are randomly assigned to receive either the treatment or a placebo. If outcomes differ significantly between the two groups, the treatment can be said to have caused the effect. However, for obvious ethical reasons, RCTs that expose pregnant people or fetuses to potentially harmful substances are not permitted. Researchers must therefore rely on large, carefully designed observational studies to tease out patterns and control for confounders like genetics.
Two recent large-scale studies — one from Sweden and one from Japan — used an innovative “sibling control” design to account for genetic factors. The Swedish study examined 2.5 million children born between 1995 and 20194. The Japanese study analyzed data from over 217,000 children born between 2005 and 20225. In both, researchers compared pairs of siblings — one exposed to acetaminophen during pregnancy and one who was not. Because siblings share much of their genetic makeup and home environment, this method controls for many confounding variables.
The results of both studies showed no meaningful difference in neurodevelopmental or autism diagnoses between the acetaminophen-exposed children and their unexposed siblings. While some earlier and smaller studies had suggested a possible link, these larger and more rigorous analyses that remove the confounding variable of genetics indicate that acetaminophen exposure does not cause autism. So, just as ice cream doesn’t cause drowning, acetaminophen doesn’t cause autism — it’s a case of correlation mistaken for causation.
Does that mean these studies are perfect or that science will never refine its understanding? Of course not. These are the best data we have today, and future research may add nuance. As scientists and public-health professionals, we must stay open-minded and evidence-driven, following where new data lead.
At the same time, it’s important to trust the scientific process. Rigorous, well-designed research — replicated across markedly different populations — provides the most reliable foundation for medical practice.
As physicians and public-health advocates, we continue to trust the science and the people who conduct it when deciding what goes into our bodies or the bodies of those we advise. That doesn’t mean we ignore legitimate concerns about high drug prices, over-marketing to consumers, or other similar public concerns — those are real issues, but they are separate from the scientific question of safety.
In the end, we hope this explanation helps the public better understand how we interpret medical data — and why, for now, acetaminophen remains the safest and most appropriate treatment for pain and fever during pregnancy.
Thank you.
The authors, Thomas Massaro, MD, PhD and Amanda Ritvo, MD, both serve on the Marblehead Board of Health. This column is based on material prepared for discussion at the open board meeting of October 14, 2025
¹ Powers EA, Tewell R, Bayard M. “Over-The-Counter Medications in Pregnancy.” American Family Physician, 2023.
² Zhang X, Grove J, et al. “Polygenic and Developmental Profiles of Autism Differ by Age at Diagnosis.” Nature, October 2025.
³ Litman A, Sauerwald N, et al. “Decomposition of Phenotypic Heterogeneity in Autism Reveals Underlying Genetic Programs.” Nature Genetics, July 2025.
⁴ Ahlqvist V, Sjöqvist H, et al. “Acetaminophen Use During Pregnancy and Children’s Risk of Autism and ADHD.” JAMA, April 2024.
⁵ Okubo Y, Hayakawa I, et al. “Maternal Acetaminophen Use and Offspring’s Neurodevelopmental Outcome.” Paediatric and Perinatal Epidemiology, September 2025.
