Retraction of Controversial Keto Diet Study Highlights Research Integrity Issues
A study published in 2025, which argued that the ketogenic diet does not contribute to the formation of arterial plaques, has been officially retracted following extensive scrutiny regarding its methodology and validity. The journal stated that “the identified errors are too great to be corrected with a corrigendum,” as noted in the retraction notice.
Initially published in April 2025 in JACC: Advances, the study examined plaque development in 100 generally healthy individuals whose cholesterol levels had risen while following a keto diet. Researchers claimed that scans conducted a year apart by Cleerly indicated no correlation between the diet and arterial plaque formation.
This assertion contradicted findings from previous research and sparked what Wired described as “a new war in the nutrition world.”
Soon after its release, the paper faced criticism, including a May 2025 letter to the journal that cited concerns about “selective reporting,” questionable statistical methods, and a study duration that was deemed too short. Critics highlighted that one co-author, Dave Feldman, lacked formal medical training, being primarily a software engineer obsessed with keto and cholesterol.
Three co-authors of the paper—Nicholas Norwitz, Adrian Soto-Mota, and Feldman—communicated to Retraction Watch that they did not have access to the data before its publication and were unaware that Cleerly’s analysis was not double-blinded. When they finally examined the data post-publication, they alerted the journal “immediately,” advocating for a publication of an expression of concern, which was later attached to the paper in January. They were also uninformed that co-author James Earls was serving as the chief medical officer at Cleerly during data collection.
In a Substack post addressing the retraction, Norwitz explained that they intentionally did not access the data until after release to maintain the integrity of the process. “After publication, the raw, anonymized data was then provided to the Citizen Science Foundation,” which funded the investigation, “and thus to Dave Feldman, who discovered numerous anomalies,” Norwitz stated.
Reflecting on the experience, Norwitz expressed regret at not reviewing the data beforehand. “At the time, it seemed responsible to forgo access until publication. This way, no one could accuse me of biases in the analysis,” he commented.
Feldman added that quality-control measures should have been mandated for any external vendor supplying analytical data to the research, including provisions for blinded re-analysis if any serious issues arose. “In the future, I would not leave such quality control to mere assumption,” he stated.
A representative from JACC: Advances confirmed that Earls disclosed his connection to the company during manuscript submission and had informed the journal of his equity stake in Cleerly upon acceptance. Although the publication acknowledged Earls’ relationship with Cleerly, the equity stake was not disclosed.
In January, six team members released a preprint showcasing a reanalysis of the data through an independent, blinded confirmation by HeartFlow. They indicated that this preprint was under peer review with another journal, addressing previous criticisms in their new manuscript.
Earls is no longer associated with the group as a co-author and is absent from the Cleerly leadership page. Christy Sievert, the public relations manager for Cleerly, declined to comment on Earls’ departure or the lack of double-blinding in the data analysis.
Brad Stanfield, a New Zealand nutritionist and vocal critic of the study, stated that the new analysis “raises new concerns rather than resolving the old ones.”
He argued that the retraction was justified and that reframing the same data to support the original claims was more advocacy than legitimate science.
Internist Michael Mindrum, another critic of the now-retracted study, acknowledged the improvements in the new analysis but expressed his concern over the motivations driving the research. In his view, “this whole endeavor exists at the intersection of social media influence and medical science,” which is quite unfortunate. He offered that it seemed the authors “will try to fit any data into their narrative.”
Mindrum criticized two of the authors, Norwitz and Feldman, as “social media influencers with an outsized media presence.”
In his Substack post, Norwitz expressed his delight at retracting the initial paper, asserting that the fundamental findings remain “incredibly robust” and should now be published without the controversies that plagued the original release.
However, Stanfield countered Norwitz’s belief in the soundness of the original methodology. “If the original method was strong enough to be considered revolutionary, it would not require retraction. If retraction was necessary, then the promotional efforts that accompanied its release were, at best, premature,” he emphasized. The Lundquist Institute for Biomedical Innovation at Harbor-UCLA Medical Center, where three co-authors are employed, previously heralded the research as a potential shift in understanding.
“The public, particularly patients making decisions about their health, deserved greater humility from the beginning,” he concluded.
If you appreciate our reporting, you can make a tax-deductible contribution to support our work, follow us on X or Bluesky, like us on Facebook, connect with us on LinkedIn, add us to your RSS reader, or subscribe to our daily digest. If you encounter a retraction not listed in our database, you can let us know here. For feedback, you can email us at [email protected].
