The 9/22 White House press conference regarding acetaminophen raises two serious concerns. First, the scientific accuracy of the purported link between acetaminophen and Autism Spectrum Disorder (ASD) is dubious. Additionally, the norms of public communication concerning whether a President should offer medical advice existed for good reason. The heads of CMS, NIH, and FDA were present, but did not dispute in real time the essential claim that Tylenol causes autism.
Second, only someone who has stood at the bedside of a six-year-old in coma with off-the-charts liver enzymes, or a family with an empty chair at the table and a wedding not attended, can fully appreciate the horror that is Reye syndrome. The President did not limit his assertions to use of the drug by pregnant women. He implied it could cause autism when given to babies and children, saying, “[t]aking Tylenol is not good.” It sounded like an unscripted remark, perhaps not so dissimilar to comments about disinfectants and sunlight as treatments for COVID which were likewise not fact-checked by the physicians in the room in real-time.
After the press conference ACOG reaffirmed its longstanding guidance: Acetaminophen remains first-line in pregnancy, used judiciously at the lowest effective dose for the shortest time in consultation with an obstetrician. The AAP also reaffirmed its guidance that acetaminophen is safe for children when taken as directed.
Association is not causation.
The core scientific claim cited in the press conference, based on the Prada et al., meta-analysis of observational studies, conflates association with causation. Observational studies can identify correlations but cannot establish causality. For instance, some believe wearing baseball caps causes baldness, but bald men wear caps to cover their heads. Correlation not causation. Plus, both baseball caps and baldness are common, the fallacy of ubiquity. It is easy to establish a correlation between common things. All cancer victims wear shoes but wearing shoes does not cause cancer.
President Trump’s examples, Cubans and the Amish allegedly do not take acetaminophen and do not have ASD, suffer from the same correlation/causation flaw. Both groups differ from the U.S. population in countless ways unrelated to acetaminophen use. Further, the claim that Cubans do not take acetaminophen is inaccurate; they take paracetamol, the same compound under a different name.
And Cubans suffer from autism at the same rate as other low- and middle-income countries, whose lower reported prevalence probably reflects underdiagnosis and underreporting. Acetaminophen/paracetamol use also varies by country, but that variation does not correlate with autism prevalence and high-quality studies do not support a causal link.
Scientific evidence remains weak.
Gold-standard science does not infer causal relationships from circumstantial evidence. The Prada meta-analysis, published in a less widely cited journal, represents observational data that may suggest associations but cannot justify policy-level claims of causation. Furthermore, Prada et al’s recommendation basically reflected the ACOG guidance—not complete avoidance. The use of this study to support blanket warnings to avoid acetaminophen ignores the established benefits of fever and pain management during pregnancy in addition to the key findings of the research, i.e., to take the lowest possible dose for the shortest possible time.
Historical and epidemiological context
Acetaminophen has been recommended for pain and fever in pregnancy for decades, while the sharp rise in autism diagnoses occurred in the 1990s and 2000s. Before the first use of the term autism in 1943, children exhibiting autistic traits were often diagnosed with childhood schizophrenia, intellectual disability or other developmental disorders. Autism likely existed for generations before acetaminophen came into widespread OTC use in the 1950s.
The Bradford Hill criteria are the epidemiologic equivalent of Koch’s postulates: heuristics for determining causation when controlled experiments aren’t possible. The one indispensable criterion is temporality: The cause must precede the effect. The timeline alone undermines any claim that acetaminophen causes autism and violates the one indispensable Bradford Hill criterion for establishing causation. Increased awareness and broadened diagnostic criteria, rather than biologic factors, almost certainly explain the so-called autism “epidemic.”
Reye syndrome
The claim that acetaminophen use by children causes autism sets the stage for the rebirth of Reye syndrome, a mysterious cause of coma, liver failure and death in children that plagued the world for generations until research proved it was linked to the use of aspirin for febrile illnesses, particularly influenza and varicella. Parents stopped using aspirin and the incidence of Reye syndrome dropped by 90 percent between 1980 and 1987. It essentially disappeared from the face of the Earth. The average U.S. physician was ten years old when the disease became a factoid and knows it only as a board exam question and the reason we don’t give aspirin to kids with viral illnesses.
The rhetoric about the use of acetaminophen in pregnancy risks a spillover into a general aversion to the use of acetaminophen in children. Compounding the problem, the leader of the free world stood at his bully pulpit and apparently just winging it said, “Don’t give Tylenol to the baby after the baby’s born.” He has since ranted on Truth Social, “DON’T GIVE TYLENOL TO YOUR YOUNG CHILD FOR VIRTUALLY ANY REASON.”
If parents return to aspirin for childhood fevers, we flirt with re-creating the exact conditions that once caused Reye syndrome to devastate families. Public health authorities would have to emphasize using the safer, but more expensive, ibuprofen instead. But who trusts the public health authorities anymore since they now have a deserved reputation for propagating misinformation. We risk seeing a return of this devastating condition.
Conclusion
The association between Reye syndrome and the use of aspirin is real; the purported link between acetaminophen and autism is not supported by any credible evidence. A simple test for evaluating public medical advice: The speaker should at least have done enough homework to be able to pronounce the medication in question. In this case, that basic standard was not met. The scientific evidence does not support the claims made in the press conference. Advising pregnant women, and children, to avoid acetaminophen entirely is both scientifically unfounded and potentially harmful. Clear, evidence-based communication is essential to maintain public trust and safeguard maternal and child health.
The authors are anonymous physicians and a health care outcomes researcher.
