In this video, we step back from discussing the multitude of statistical errors and reporting problems of the KETO-CTA studies and look at the whole saga. And we discuss a recent comparison of the KETO-CTA plaque data to the PARADIGM study. The whole story was not presented in that comparison.But, taking a step back, what are my thoughts from following this whole saga? I cannot but come to the conclusion that these researchers are extremely biased and are willing to spin the result of their studies in ways that make it appear more favorable, "reassuring", and fit the story they already believe. We clearly saw this in their promotion of the 2025 JACC paper, Plaque Begets Plaque, ApoB Does Not. In the words of Norwitz, "This reflects a level of progression that is expected with aging", and that “LDL and ApoB were pretty much irrelevant in determining or predicting whether or not plaque would progress”. And Feldman: “Many are focusing specifically on the change in % NCPV only, particularly given it is our preregistered endpoint.” Later in the same thread, Feldman added: “There was no correlation with LDL-C or ApoB — arguably the most relevant of all.”And Soto-Mota: “Are LMHRs progressing faster than others? No! They behave like other ↓CVDrisk groups” and “these first prospective data should be reassuring for LMHR patients and their physicians”. Reassuring?? This is exactly the framing or spin I am talking about.They promoted their questionable, preliminary and exploratory finding of no relationship between LDL/ApoB and plaque progression and downplayed or ignored the actual amount of plaque and plaque progression. It's as if they were implicitly saying: don't focus on the primary outcome of our study, because it's very concerning, focus instead on our fragile and weak exploratory statistical analysis.Now, everyone has some bias of course. But this, to me, was misleading and irresponsible. And whatever you believe about the Cleerly situation, that happened later, and has no bearing on how they promoted this study at that time. It is my view that they so fully believe their proposed theory that every study, data point, or result they get is analyzed within that framework. This allows them to ignore, downplay and explain away anything that isn’t “reassuring” to them. They don’t even appear to consider the alternative explanation, except perhaps in a few passing ‘strategic disclaimers’. The same pattern extends beyond their own studies to how they interpret and present other evidence, especially in The Cholesterol Code documentary and in comparisons such as PARADIGM versus KETO-CTA.I didn’t start from this position. I just wanted to analyze the study’s methods and data. But from following this whole saga, this is where I’ve ended up. So I caution anyone evaluating analyses from these authors that there may be much more to the story than the conclusions they present. Important caveats may be downplayed, relevant information may be left out or reframed to sound more reassuring, consequential statistical issues may be ignored or misunderstood, and any uncertainty may be interpreted asymmetrically through a pre-existing framework of belief.For these reasons, studies and analyses from these authors warrant especially strong scrutiny of the underlying data, methods, analytical choices, and what is and is not being presented.

In this episode of The Keto Files, we break down the new KETO-CTA preprint on keto, LDL-C, ApoB, and coronary plaque progression.The short version: the statistical analysis is misleading. The paper uses a misdefined central “LDL-C exposure” variable, presents false IQR bands that downplay plaque progression, uses an incoherent and ambiguously reported TOST equivalence procedure, presents one-sided "reassuring" power arguments, and doesn't report the ApoB model results anywhere.Ignoring the statistical modeling, we look at whether we can learn anything from just looking at the plaque progression data. At the end, we compare KETO-CTA to the Nature-CT cohort to put the plaque progression results in context.Paper discussed:KETO-CTA preprint: The Impact of Sustained LDL-C Elevation on Plaque Changes: Primary Coronary plaque progression results from the Keto CTA Study

Does reducing saturated fat actually prevent heart attacks?For decades, saturated fat from foods like butter, red meat, cheese, coconut oil, and full-fat dairy has been treated as something to reduce. But the new USDA/HHS dietary guidelines and the inverted food pyramid have made the message a lot less clear.In part two of this series, we look into three of the major randomized controlled trials behind the meta-anlayses: the Minnesota Coronary Experiment, the Sydney Diet Heart Study, and the LA Veterans Trial. Along the way, we also discuss relative risk vs absolute risk, what an ideal saturated-fat trial would look like, why these old trials are so hard to interpret, and how issues like trans fats, adherence, short duration, and missing data complicate the conclusions.Hopefully, there’s also something useful here about how to think through studies, evidence, and nutrition claims when the answer is not as clean as it first seems.This is the second in a series of videos on the subject. Slides:https://www.jsdatascience.com/saturated_fat_meta-analyses_RCTs/Papers:Steen et al. 2025 https://pubmed.ncbi.nlm.nih.gov/41397264/Yamada et al. 2025 https://pubmed.ncbi.nlm.nih.gov/40416032/Frantz et al. 1989 https://pubmed.ncbi.nlm.nih.gov/2643423/Ramsden et al. 2016 https://pubmed.ncbi.nlm.nih.gov/27071971/Woodhill et al. 1978 https://pubmed.ncbi.nlm.nih.gov/727035/Ramsden et al. 2013 https://pubmed.ncbi.nlm.nih.gov/23386268/Dayton et al. 1968 https://pubmed.ncbi.nlm.nih.gov/4176868/Dayton et al. 1969 https://www.ahajournals.org/doi/10.1161/01.cir.40.1s2.ii-1Wilkins et al. 2012 https://pubmed.ncbi.nlm.nih.gov/23117780/

Does cutting saturated fat actually prevent heart attacks?For decades, saturated fat from foods like butter, red meat, cheese, coconut oil, and full-fat dairy has been treated as something to reduce. But the new USDA/HHS dietary guidelines and the inverted food pyramid have made the message a lot less clear.In this episode, Calvin and John look closely at two 2025 meta analyses of randomized controlled trials: Yamada 2025 and Steen 2025. One says there is no clear benefit to reducing saturated fat. The other says there may be a benefit, but mostly depending on your cardiovascular risk and what you replace saturated fat with.We don’t just read the abstracts. We go deep into the studies, recreate some of the charts, check the reported metrics, look at the forest plots, compare odds ratios and risk ratios, and dig into where the conclusions seem strong, weak, or a little too confident.Hopefully, there’s also something useful here about how to think through studies, evidence, and nutrition claims when the answer is not as clean as it first seems.

In this episode, we look at how Americans die at different ages using U.S. mortality data from the CDC. John walks through an interactive mortality data tool he built with multiple interactive charts showing broad causes of death by overall rank, age at death, and over time, along with detailed subcauses.The charts have filters for sex, age range, time period, and metric. We discuss the patterns that stand out and why the tool is useful.

A recent government announcement claimed Tylenol use during pregnancy causes autism, sparking widespread concern. We dissect the controversial study behind this claim, uncovering major flaws and conflicts of interest, and reveal what higher-quality evidence actually says about the risk.

Critical analysis of the popular anti-aging supplement fatty15, dissecting its "science-backed" claims and deceptive marketing tactics.Is fatty15 a “new essential fatty acid” that’s “3× better than omega-3”?Is fatty15 GRAS "certified'?Does fatty15 have any health benefits in humans?We review the evidence, including all the human randomized controlled trials on fatty15.Slides: https://www.jsdatascience.com/fatty15/

In this episode, we revisit the controversial Keto-CTA study about cholesterol, and arterial plaque progression. After the authors released some of their raw data, our analysis reveals the study's flaws are far worse than first thought. John breaks down the egregious errors, from published charts that incorrectly cut plaque progression in half, to statistical models that violate their own assumptions, directly contradicting the authors' claims and invalidating their conclusions.