The paracetamol-autism association and the problem of politicizing the evidence

On the same topic:

The move to link paracetamol to autism— a move pioneered by the Trump-Kennedy-Oz political-media trio —works because it frames primal fears within a simple narrative: "They lied to us about a widely used drug, and now we will tell the truth and defend children." It's a moral battle narrative that doesn't require solid evidence: a methodologically flawed observational association presented as "proof" is enough to shift the burden of proof and put on the defensive those who urge us to look at the enormous amount of data that proves the opposite. In recent hours, in fact, newspapers and scientific societies have pointed out that there is no new data capable of transforming correlations into causation, while the American College of Obstetricians and Gynecologists and the Society for Maternal-Fetal Medicine reiterate that acetaminophen, when used correctly, remains the recommended option during pregnancy; even the EMA has confirmed its safety according to current guidelines. The “hard” news is political, not scientific, and exploits the dissonance between government statements and clinical consensus to erode trust in scientific institutions.

From the perspective of the studies cited by charlatans as "hard evidence," the crux of the problem lies in confounding factors. Many studies reporting associations between use during pregnancy and neurodevelopmental disorders are observational and susceptible to "confounding by indication": acetaminophen is taken because of fever, pain, or infection—conditions that can in themselves alter the risk of autism or ADHD. When more robust designs are adopted, such as sibling analyses within the same family, the association disappears, indicating that shared genetic and environmental factors explain much of the signal. A Swedish cohort of nearly 2.5 million individuals published in JAMA in 2024 showed precisely this, while prospective studies on the role of fever during pregnancy indicate an increased risk regardless of the drug, strengthening the hypothesis that maternal illness, rather than treatment, is the driver of the signal. This is why clinical societies continue to recommend treating fever and pain during pregnancy: the risks of an untreated fever syndrome are real, but the causal risks of acetaminophen on autism are not.

Why then does declaring "we have the evidence" pay and continue to pay? Because it activates a powerful set of cognitive biases. The intentionality bias always seeks a culprit for a perceived harm: a daily medication is a more psychologically satisfying target than a patchwork of genetic and environmental factors that are difficult to control. The availability bias makes vivid anecdotes ("I took Tylenol, then I noticed...") more credible than abstract numbers; the proportionality bias demands "strong" causes for "large" effects; the confirmation bias offers a public already wary of vaccines and public healthcare a new piece of the alleged conspiracy. Above these, the "zero-risk bias" and the aversion to ambiguity operate: during pregnancy, the seemingly "purest" choice is preferred, even when the real risk lies in not treating fever and pain. And when a political authority confirms those fears, the identity-based "authority bias" kicks in: the strength of the evidence doesn't matter, what matters is loyalty to the group. The dynamic is amplified by the “ illusory truth effect ”: repeating a simple message (“Tylenol = autism”) increases its perceived plausibility, especially if relaunched by government channels and friendly media.

Then there's the political strategy. First, mobilization: "Protect the children" is the most powerful call to action, capable of transforming science into self-interested technocratic cynicism. Second, delegitimization: if the FDA, ACOG, or researchers are a brake and an enemy because they are capable of well-placed criticism of political power, science is portrayed as a conniving bureaucracy that "denies the evidence" to protect industrial interests. Any call for methodological rigor thus becomes merely a stall or an attempt to deny the "facts" everyone is convinced of. Third, shifting the agenda: instead of investing in the network of services for autism spectrum disorders, a culprit and a quick fix are promised, even relaunching "life-saving" drugs with still weak foundations for complex neuropsychiatric uses. This is exactly what we see: in the face of frantic political attention to a supposed causal link refuted by the most robust experimental designs, while the clinical community reiterates, with documents and public pages, that paracetamol, when indicated and in the correct dose, remains the safest choice, the miracle of leucovorin is announced , touted as a breakthrough. Leucovorin is actually a targeted remedy for specific phenotypes of folate deficiency, with still limited evidence, and certainly does not "cure autism." However, coincidentally, one of the trio—Dr. Oz—sells multiple remedies based on this substance through his companies.

This is how it works, then: you take a topic where the observational literature is heterogeneous and controversial, you ignore the fact that family-controlled studies and data on maternal fever disprove direct causality, you ignore the converging statements from ACOG, SMFM, and EMA, and you announce a definitive truth that justifies immediate interventions. This way, you garner consensus, divide public opinion, and undermine the authority of scientific institutions; the cost is paid by patients, exposed to messages that pressure them not to treat conditions that, during pregnancy, should be treated. Science here doesn't have a problem with a "lack of evidence," it has a problem with the politicization of evidence. The solution to this lies not in science itself, but in good politics and in defending the bastions that allow a liberal democracy to remain standing. Provided that there are sufficient interested citizens, which is far from certain.

More on these topics: