The Health Report That Cited Papers Nobody Wrote

Health Secretary Robert F. Kennedy Jr. called it "gold-standard" science. More than 500 studies and other sources, cited to back a federal report on the health of American children. On X, he went further: never in American history had the federal government taken a position on public health like this. A milestone, he wrote.

It was a milestone. Seven of the citations in that report point to studies that do not exist.

The Ghosts in the Footnotes

The "Make America Healthy Again" report is 73 pages. Reporters at NOTUS went through the footnotes and found dozens of citation problems: broken links, wrong authors and wrong issue numbers. Ordinary sloppiness, the kind every researcher has committed at 2am with a deadline looming. Seven citations were something else. They pointed to studies nobody had written.

Take the citation on adolescent anxiety. The report lists epidemiologist Katherine Keyes as first author of a study on mental health and substance use in adolescents during the pandemic, supposedly published in JAMA Pediatrics, volume 176, issue 12. Keyes told NOTUS by email that the paper "is not a real paper that I or my colleagues were involved with." That issue does not contain the study.

The report's section on "corporate capture of media" cites two studies on direct-to-consumer drug advertising, attributed to Shah, M. B., et al. 2008 and Findling, R. L., et al. 2009. NOTUS looked for both. Both were nowhere. A spokesperson for Virginia Commonwealth University, where psychiatric researcher Robert L. Findling currently teaches, confirmed he never authored the article credited to him. He never wrote it, full stop.

Then there's the asthma citation. The report claims 25% to 40% of mild asthma cases are overprescribed corticosteroids, sourced to a paper titled "Overprescribing of oral corticosteroids for children with asthma." Search for that title and the only result is the MAHA report itself, citing itself into existence like a citation ouroboros. Pediatric pulmonologist Harold J. Farber, listed as the study's first author, told NOTUS he never wrote it and doesn't know the other listed authors. He added, for good measure, that even a real version of that finding would be a stretch: generalizing from one Medicaid managed-care program in Texas using 2011–2015 data to national care patterns in 2025 would be, in his words, "a tremendous leap of faith." The study was fiction. Even its imaginary self would have been misapplied.

The Studies That Existed and Still Got It Wrong

Fabricated citations are the loud problem. The report has a quiet one too: real studies, cited to say things they don't say.

Biostatistics professor Joanne McKenzie authored a review the report cites as proof that therapy is as effective as or more effective than psychiatric medication. McKenzie told NOTUS her review never touched psychotherapy at all. "We did not include psychotherapy in our review," she wrote. "We only compared the effectiveness of (new generation) antidepressants against each other, and against placebo." The report built a claim about therapy out of a study that measured antidepressants.

Elsewhere, the report says a cited paper found antipsychotic prescriptions for children rose 800% between 1993 and 2009. The actual study found an eight-fold increase, between 1995 and 2005. Different number, different decade — a citation wearing someone else's data like a coat that doesn't fit.

Researcher Mariana G. Figueiro's screen-time study got a similar treatment. The report attributes it to Pediatrics, and treats the subjects as kids. Figueiro told NOTUS none of that holds. "The conclusions in the report are not accurate and the journal reference is incorrect," she wrote. "It was not published in Pediatrics." The subjects were college students. The report cited a study of college students as proof of something it was never built to measure.

The Response

Asked about all this, the White House and HHS declined to dispute the specifics. They downgraded them instead. HHS spokesperson Emily Hilliard called them "minor citation and formatting errors" that had been corrected, insisting the substance of the report remained "a historic and transformative assessment." The administration updated the report and removed the seven references to studies that don't exist. They filed it as an edit.

The seven ghost citations are gone from the document, quietly, the way a magician's assistant disappears behind a curtain nobody watches too closely. The commission has another report due in August, this one called "Make Our Children Healthy Again Strategy." Presumably it will also cite sources. One hopes somebody reads them first.

The Mechanism

No chatbot needed to hallucinate here, at least not one anyone has confirmed. The cause was simpler and older. Someone wrote citations that sounded right and looked like real papers. Nobody downstream checked whether the things existed before they went into a federal document with a Cabinet secretary's name on it. A citation that looks right and a citation that is right are different objects, and the gap between them is exactly where these seven studies were hiding. Fluent enough to pass a skim. Empty enough to not exist.

The report is 73 pages with hundreds of citations. Most readers are not going to pull every DOI and confirm the journal issue matches the claimed study. That's the whole reason a report like this can walk out the door with seven invented studies attached and get called "gold-standard" on the way out. Volume defeats scrutiny. Nobody has time to check every footnote by hand, so nobody does, until a newsroom decides to.

The 2am Version

This is a citation problem, and it belongs to anyone who has ever built an argument on a stack of sources they trusted without opening. A paper title that sounds exactly right, attached to a real author at a real institution, with a journal and issue number plausible enough to skim past. The MAHA report had all of it, seven times over. It took a newsroom cross-checking author emails against DOIs to catch it, after the report had already been called a milestone.

That's the expensive way to find out your citation is fiction: after publication, in the press, with your name on the document. The cheap way is to check before you hit send. A citation that gets checked against the live record — does the paper exist, does it say what you're claiming — catches Keyes and Findling and Farber's names attached to nothing, before it's Kennedy's problem instead of yours.

Boring is the goal. Boring is nobody at NOTUS ever needing to email your listed co-authors to ask if the paper is real.