The numbers in Jeremy Faust’s spreadsheet shouldn’t have moved. Not like that. Not that fast. Faust, an emergency physician at Brigham and Women’s Hospital in Boston, was scrolling through prescribing data from more than 1,600 American hospitals when he saw two lines on a graph do something that ordinarily takes years of clinical trials, guideline revisions, professional education campaigns, and persistent nudging from medical societies to achieve. They moved the week after a White House briefing.
Paracetamol prescriptions for pregnant women in emergency departments dropped sharply. Leucovorin prescriptions for children with autism diagnoses more than doubled.
The briefing in question took place on September 22, 2025, and included comments from the US President and the FDA Commissioner. Officials discouraged the use of paracetamol during pregnancy, citing research on links to neurodevelopmental disorders. They also promoted leucovorin, a folate-based drug approved for treating certain cancers and metabolic conditions, as what speakers called an exciting therapy that may benefit large numbers of children who have suffered from autism. There was, at the time, no new clinical trial data. No formal guideline revision. The drugs mentioned at the briefing were the drugs that moved; the drugs not mentioned did not. “The results were astounding to me,” says Faust.
The scale and specificity of the change is what makes it scientifically striking.
Faust and his colleague Michael Barnett, a professor of health policy at Brown University’s School of Public Health, reached into Cosmos, a database built from Epic’s electronic health records system covering roughly 294 million patient records across the US. They ran an interrupted time series analysis, building a statistical model of expected prescribing patterns from the three months before the briefing, then comparing those projections against what actually happened after September 22. The results are published today in the Lancet.
Emergency paracetamol orders for pregnant women fell by 10 percent overall in the study period, with the sharpest drop (20% below expected levels) in the third week after the briefing. The same analysis run on non-pregnant women of the same age showed no significant change, which matters enormously: it means the effect was specific to the population the briefing discussed, not just a general drift in paracetamol usage.
Leucovorin prescriptions for children aged 5 to 17 climbed 71 percent above expected levels across the full study period, surging to 93 percent above expected in the first month and more than doubling in the second week. About 72 percent of those new prescriptions were written for children with autism diagnoses, a group that represents only around 4 percent of the paediatric population in the dataset. Folic acid prescriptions, structurally similar to leucovorin and theoretically a plausible alternative if parents were simply seeking folate supplementation, showed no significant change. Neither did opioids nor IV fluids in emergency departments. Something specific was happening, and it was following the briefing’s exact contents.
Whether patients drove those changes or physicians did is harder to disentangle. Barnett reckons both were probably at work. “An important implication of these results is also that it’s not just patients who were influenced by the unconventional press conference,” he says. “Their doctors were either influenced themselves or pushed by patients to adopt a new practice.” That second pathway is perhaps the more disquieting one. It suggests that at least some clinicians adjusted their prescribing not because of new evidence but because their patients came in having watched a press conference. The clinical encounter, in those cases, was downstream of political communication.
The evidence base for both claims made at the briefing is, to put it charitably, unsettled. On paracetamol: a major systematic review published in the Lancet in January 2026 found no increased rates of autism, ADHD, or intellectual disability among the offspring of people who used paracetamol during pregnancy. The drug has a decades-long safety record in pregnant patients and is generally considered safer than the available alternatives. Fever in pregnancy, on the other hand, is a documented risk factor for neurological disorders in the child. Undertreating a fever, in other words, carries its own risks. On leucovorin: small trials for autism have produced mixed, preliminary results, and it appears in no standard autism treatment guidelines. The drug is approved for cerebral folate deficiency, a rare condition.
None of that evidence changed after September 22. “The White House briefing was an extremely unusual mechanism to communicate medical information and bypassed many standard checks on ensuring accurate messaging,” says Barnett. The authors are careful about the limits of their analysis: it does not, and cannot, prove causation, nor does it assess whether patients experienced better or worse outcomes. A leucovorin shortage appeared in the US in November 2025 (possibly a consequence of the demand surge), which may partly explain why aripiprazole and risperidone prescriptions ticked up later in the study period, possibly as clinicians sought alternatives. The statistical models didn’t account for seasonal patterns in paracetamol use, which typically rises in late autumn with cold and flu season.
Still, the paracetamol finding faded over the study period. The leucovorin finding did not, which is roughly what you’d expect: avoiding a drug for nine months is a discrete decision that gets harder to sustain as pregnancy progresses and fevers happen; adding a new treatment is a standing prescription that continues.
This is not the first time researchers have documented political speech reshaping medical behaviour at speed. During the COVID-19 pandemic, statements by then-President Trump about unproven treatments appeared adequate to drive shifts in prescribing and internet search behaviour, despite the evidence base being, at best, thin. What the Faust and Barnett study adds is something more precise: a controlled natural experiment, with comparator drugs, comparator populations, and a dataset large enough to distinguish signal from noise. “It can take years, even decades, for high-quality research to finally reach clinicians,” says Faust. “Here, by using the White House, it was done overnight. Unfortunately, they’re claiming breakthroughs that simply haven’t occurred.”
The broader implication is harder to shake. “The results show just how much political leaders can steer health behavior even when there has been no change in the evidence for these therapies,” says Barnett. What moves a prescribing graph, it turns out, is not always a clinical trial. Sometimes it’s a press conference.
Study link: https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(26)00243-6/fulltext
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.