The first clue came from fathers.
In a review of more than 25 million pregnancies, children were more likely to receive an autism diagnosis even when the antidepressant in question had been taken by their father during the pregnancy. His medication could not have directly reached the fetus. An apparent link between a mother's prescription and her child's diagnosis began to look less like a drug effect—and more like a family history.
The Unexpected Clue in Fathers’ Prescriptions
On May 14, 2026, researchers at the University of Hong Kong published a review of 37 studies in The Lancet Psychiatry, covering 648,626 pregnancies exposed to antidepressants and nearly 25 million unexposed pregnancies. Their conclusion challenged a persistent suspicion: after accounting for parental mental health and other shared factors, the evidence did not establish that commonly prescribed antidepressants caused autism or attention-deficit/hyperactivity disorder (ADHD).
At first glance, the numbers pointed in another direction. Children exposed to antidepressants before birth had a 69% higher relative risk of an autism diagnosis and a 35% higher relative risk of ADHD. But these were observational associations, not proof of causation. Once researchers considered underlying maternal psychiatric conditions, compared siblings, and examined other family characteristics, those associations weakened or lost statistical significance.
The researchers also examined mothers who had taken antidepressants before conception but not during pregnancy, as well as fathers who took them while their partners were pregnant. Similar patterns suggested that inherited vulnerabilities and parental mental health could explain much of the apparent risk. Higher doses did not produce higher rates of autism, further weakening a straightforward drug-exposure explanation. A large 2022 American study had found a comparable pattern after more rigorous adjustment.
The distinction carried unusual political weight. U.S. Health Secretary Robert F. Kennedy Jr. had raised unsupported concerns about prenatal antidepressant exposure. Days before the new study, Reuters reported that health officials had explored whether certain selective serotonin reuptake inhibitors (SSRIs) could be restricted. The Department of Health and Human Services denied discussing a ban, and Kennedy said his initiative to reduce prescribing was not an instruction for patients to stop taking medication.
The research was reassuring, not absolute. Its authors rated the certainty of evidence low to very low; only two studies reported participants' ethnicity. Two older antidepressants, amitriptyline and nortriptyline, retained possible associations in some analyses, although researchers cautioned that more severe underlying illness could be responsible.
But uncertainty cuts both ways. Catriona Waitt, a University of Liverpool pharmacologist, warned that studies failing to account for why women were prescribed medication can overstate its harms, potentially leaving maternal illness untreated. For patients with moderate or severe depression, the question is not whether medication carries zero risk. It is what happens when a treatment is stopped, sometimes abruptly, on the strength of a misleading comparison.
When a Correlation Becomes a Family Story
For a generation encountering medical claims in headlines, clips, and increasingly AI-generated summaries, the easily repeated number is the unadjusted 69%. The harder story is why that number changes. When the distinction disappears, a statistical variable—maternal medication—can harden into a moral accusation.
That transition was visible months earlier in a controversy over a different medicine, acetaminophen, sold as Tylenol. After President Donald Trump publicly suggested an unproven prenatal link to autism in September 2025, 21-year-old autistic Jack Cariello texted his mother: "Did you take Tylenol while you were pregnant with me?" His mother, Carrie, recounted the exchange to Yahoo. She also said that, after Kennedy spoke about autism destroying families, Jack had asked, "Did I destroy our family?"
His questions were not about antidepressants. Their relevance is what causal rhetoric can do once it leaves the laboratory: a young person starts interpreting his own existence through what his mother might have swallowed.
Another young autistic adult, Will Palmucci, 21, had already challenged this framing. Accepted to Wesleyan University to study film, he told CBS Boston in April 2025: "Just because you have autism doesn't necessarily indicate that you're incapable of doing anything." His account cannot represent the entire spectrum; families of autistic people requiring lifelong assistance have also called for greater research and support. Those needs, however, are different from assigning responsibility to a pregnant person's treatment decisions.
The Lancet review itself notes that no people with relevant lived experience participated in its research or writing. That does not invalidate its statistics. It exposes the distance between answering whether a medicine causes a diagnosis and understanding what the accusation does to those living with that diagnosis.
A medical association can be corrected by better evidence. The suspicion that someone's mother could have prevented who they are is not so easily retracted.
Explore more exclusive insights at nextfin.ai.
