Mambo Number 6 retweeted
Them: ApoB didn't correlate with plaque progression.
Us: Wait...that wasn't your preregistered primary outcome.
Them: Plaque begets plaque
Us: Wait...your study shows LMHRs have historically high plaque progression.
Them: The data is wrong because our vendor sucks.
Us: Why did you go on a historic marketing campaign and produce a documentary?
Them: We retracted our paper. All good.
Us : That's not how this works.
Mambo Number 6 retweeted
KETO-CTA is a case study, not because of results, but communication. There’s no going back.
Right. It's actually one of the most incredible stories of our research.
There's a very strong resistance by critics to openly discussing the anomalous dataset that was provided to us from Keto-CTA, leading to the retraction of that paper.
We're in talks right now on a successor to our documentary and this would definitely be a central theme... (It's a bit like a real life Emperor's New Clothes)
Mambo Number 6 retweeted
The Keto CTA study is back on my feed today, and the authors are reiterating that it was retracted at their request.
That’s fine. I’m actually open to the idea that the Cleerly data was flawed.
My much, much larger problem is with the way the science was communicated. As they say, “Trust takes years to build, seconds to break, and forever to repair.”
At launch, the message was “no association of ApoB with plaque” and “most participants showed NO OR MINIMAL progression.” (Pictures below.)
But when the authors requested the retraction, the exact same data was suddenly the problem. Plaque progression was so high and so one directional, that it was said to be statistically implausible.
So what is it? No or minimal progression or progression so consistent and so bad that the data must be flawed?
Second, it’s worth remembering how science is supposed to be communicated. Plaque progression was the PRIMARY OUTCOME cited for the Keto CTA study.
But plaque progression was not reported.
The authors later apologized (which, to their credit, is good!) for not including the number in the graph they provided.
But while putting the number in the figure would have been nice, that is not how you DISCUSS your primary outcome in the DISCUSSION of a paper.
Let’s compare this to how @BradStanfieldMD discussed the primary outcome of his Rapamycin study, a study he put a tremendous amount of reputation behind and, quite literally put a second mortgage on his house to get over the finish line:
“4.1 Principal Findings
This randomised, double-blind, exploratory trial evaluated whether once-weekly sirolimus (rapamycin) 6 mg, timed to avoid peak training windows, could enhance functional adaptations to a 13-week resistance and endurance programme in sedentary older adults.
Contrary to the ‘cycling hypothesis’ derived from preclinical models, we found no evidence of benefit. Instead, sensitivity analyses (complete-case and per-protocol) revealed a statistically significant blunting of functional gains in the sirolimus arm. Specifically, participants taking the drug performed fewer chair-stand repetitions and showed non-significant trends toward reduced walking distance and grip strength compared with placebo. Safety monitoring identified a higher burden of minor AEs and one serious infection (pneumonia) in the treatment arm.”
He was clear. The study did not meet his primary endpoint, and instead, the intervention appeared to have worsened it. That is how science is communicated.
I asked Grok to write the Discussion the Keto CTA should have had for its primary endpoint, had it been written the way Dr Stanfield did. Here is what Grok wrote:
"Discussion
Principal Findings
This prospective, single-arm observational study evaluated whether long-term exposure to elevated LDL-C and ApoB (in the lean-mass hyper-responder phenotype on a ketogenic diet) was associated with progression of noncalcified coronary plaque over one year, as quantified by coronary CT angiography. The pre-specified primary endpoint was change in noncalcified plaque volume (NCPV).
Contrary to the hypothesis that these elevations would not drive measurable plaque progression in this metabolically healthy cohort, we observed a median increase in NCPV of 18.9 mm³ (IQR 9.3–47.0 mm³), corresponding to a 42.8% relative change. The mean change was 31.5 mm³ (SD 31.5 mm³), reflecting substantial inter-individual heterogeneity. Most participants showed an increase in NCPV."
That is how you discuss your primary endpoint. It’s not supposed to be a number on a chart (which was omitted anyway).
We, the authors, retracted the paper, @Drlipid
This was due to the anomalous @Cleerlyhealth dataset they provided us alongside their repeated refusal to quality control check their work.
Study participants have since resubmitted their scans and have gotten results far more in line with the other independent analyses from Heartflow and QAngio.
So yes, understandably, we moved to withdraw the paper leading to its retraction.
(This has been known for over half year now.)
nitter.cf/realDaveFeldman/status…
Mambo Number 6 retweeted
Replying to @SomerwilTara
It is rather interesting to see that people accuse doctors of focusing on disease (true) and not on prevention (false) and then turn around to claim you should just wait for something nasty to happen before you change your ways for the better.
Bruh you feed your own grandkids processed food. Why are you judging others for giving their kids 10g/chocolate a day over Christmas?
nitter.cf/drjenunwin/status/1463…
I die a little inside when I see huge chocolate advent calendars in the supermarket in September 🥲
A proportion of people are more or less eating themselves to death on sugar and processed foods. I used to be one of them. @FAS_TheCHC @TheFAIOfficial @lowcarbGP @DoctorTro
3
4
500
Mambo Number 6 retweeted
I think Dave realised that Nick's summary doesn't actually reconcile with the manuscript. Which is why Nick didn't link the primary manuscript to begin with.
I suspect not linking the manuscript was an immediate red flag to @StevenQuartz, it certainly was to me.
Good science communication will always link the manuscripts and actively encourage the audience to read the full text.
Nick's audience are the types that will probably never go fishing or actually read a manuscript anyway. They are very happy to be science commodified, while believing they are receiving science communication. Nick is now simply a health content creator, and treats scientific papers as mere theatrical props, while positioning himself as the sole arbiter of truth.
Mambo Number 6 retweeted
Replying to @SomerwilTara
means you´re hitting a nerve. keep going
movements rooted in anecdotes and scientific illiteracy dismiss anecdotes when they´re inconvenient
most people can see through it
Mambo Number 6 retweeted
Replying to @foundmyfitness
It's important to explicitly note that this result holds for the highly selected population of the study - those with severe obesity, prediabetes, and fatty liver disease. That's precisely the metabolic phenotype in which carbohydrate restriction would be expected to have its largest effects. The results don't show that keto is superior to other diets beyond that population. And the authors themselves note that, "we cannot determine whether a less-severe reduction in carbohydrate intake without ketosis would have the same therapeutic effects." That's because a lower carb, non-ketogenic diet was not a compatator in this study [lead author posted that he agrees with this interpretation]
Mambo Number 6 retweeted
Hi Austin,
This study was explicitly marketed as an investigation into a very specific phenotype with extremely elevated LDL-C, and that this extremely elevated LDL-C is benign in this phenotype.
It is a hypothesis that is going against conventional wisdom. It is a hypothesis that is trying to break the establishment. As I recall, it is billed in the documentary as a high stakes David vs. Goliath battle for the truth?
To challenge an established scientific consensus, your baseline data must be fortress like, not a sieve. You cannot parade a cohort as paradigm shifting when roughly one in seven of your participants fail the basic entry (and the most important) criteria.
Yes, in science we witness drift. But this parameter is so important to the hypothesis, that it is one parameter that we really want to keep tidy for the bookkeeping, and of course for peer reviewed credibility and pre-empting obvious refutation.
In study design, we teach students and researchers how to identify the “critical to quality” (CtQ) factors, the handful of make or break attributes that determine whether a study's findings are genuinely interpretable or merely noise. Not every variable deserves equal scrutiny. The variables that define the phenotype being studied certainly do.
So this particular inclusion threshold isn't an ordinary bookkeeping detail, it’s actually the operational definition of the thing being studied.
If there is drift or deviations, particularly in this case, it should be explicitly mentioned in the manuscript. Speaking to two very old wise heads at university over lunch today, if this occurs on the single variable that defines your target phenotype, failing to explicitly document and justify it within the primary manuscript they would see as a breach of scientific accounting they said.
You don't literally have to get everything correct, because no real study ever does. What you have to do is be exceptionally rigorous about the things most central to the hypothesis.
A baseline value of 49 mg/dL is so discrepant from the stated >190 mg/dL eligibility threshold that the manuscript needed to explain it. And think about what this means? LDL-C exposure was defined as baseline LDL-C plus the measured change during follow up. Every one of those sub threshold baseline values, including whatever produced the 49 mg/dL reading is fed directly into the primary exposure variable with no threshold based exclusion or flag attached. Again, you are not truly testing your hypothesis. You have in all actuality “diluted” the hypothesis.
What they should have done is either exclude that 49 mg/dL participant, along with a few others (two others at least), or retest those participants again, maybe a few days later? This may have been a simple lab or transcription error. It could have been 249 mg/dL and someone had fat fingers?
And in truth, because this parameter is so important to the hypothesis the participants should have had their LDL-C levels tested regularly for the duration of the study. Maybe every three months or so? You simply needed vendor receipts to prove the exposure was consistently present. Cheap as chips and makes your hypothesis a little more watertight. This study is not watertight. It’s leaking like the Titanic. The KETO-CTA study is actually resembling the Titanic story. Much pomp and prestige at launch, only to be exposed for its many flaws and shortcuts when tough questions were really asked of the ship.
At times your response reads less like an explanation of what happened, and more like a damage control exercise.
The rebranding of the “pilot study” is a good example of this post hoc damage control, because you are fully aware that all the grown-ups know this is a “prospective observational cohort study”. It was actually billed as a “trial” in the original published manuscript if you care to remember.
Respectfully, if you're not sure whether this was a pilot study, you should not be responding with such enthusiasm and authority.
Finally, we also have the small matter of the N problem. And the N problem is worse than the single 97 vs. 100 gap as seen in this slide. The preprint gives three different denominators across its own results, and none are reconciled anywhere in the text. I cannot find anything in the manuscript that tells me why the denominator shifts between analyses? Again, if a study purports to be paradigm shifting, basic cohort accounting isn't an optional extra. It requires full, line by line transparency.
Wish you a good day, and greetings from the University of St Andrews, Scotland.
@_earthling74 @StevenQuartz @drgarymcgowan @JohnSloughII @dylanarmbruste3 @Marijn3D @ChristofferBN
*chef’s kiss*
“Don’t know why you’re making such a big deal about the cholesterol, it’s only the title of the documentary.”
2
10
1,104
“I’m proud to have worked on the Bailey. But please, let’s remember I live in this Motte now.”
Mambo Number 6 retweeted
Replying to @MichaelMindrum
One of the greatest travesty’s in modern nutrition science and politics is the premature end to Dr Hall’s work at the NIH.
Mambo Number 6 retweeted
I've updated my paper on the health impacts of seed oils with causal analyses!
I found lots of family data in the NHANES, so used family fixed-effects. I found that...
Serum linoleic acid was associated with BETTER health, even within families!
I can grant that PUFA oxidises more readily than MUFA and SFA. What I don’t understand is why that would lead us to think butter was healthier than canola.
Or do you not think that - I appreciate you said canola was one of the better ones?
Let me preface this by saying that there is a big diffference between milk that comes 'happy cows' in pasteur and stressed out cows in a concrete enclosures. Different lipid profile.
Canola is not inherently bad, it's not one of the worst ones. Here is just one angle --it has to do with the amount and ratio of polyunsaturated:monounsaturated:saturated fats. The more polyunsaturated fats in an oil, the more prone to spoilage and oxidation. Oils with a high percentage of polyunsaturated fats 'go bad' quickly. Many are rancid before you even cook with them. They are chemically filtered and de-odorized to remove the rancid odors so consumers don't notice. Every cell in your body is surrounded by a lipid-layer. You are what you eat.
Cottonseed oil is actually it's own, separate evil. It's inherently toxic to us and it's the starting ingredient to a common pesticide. It's also very unstable so they partially hydrogenate it. Then, they mix it into vegetable oil. Want to go down a rabbit hole? Look into the history of cottonseed oil.
1
144
Mambo Number 6 retweeted
The fact the @BenBikmanPhD, a researcher and professor would agree, support, spread, or associate with Dr. Ammous in anyway tells you all that you need to know. Ben has swiftly been moving into influencer - grift zone but he is now clearly and unapologetically in the land of quackery.
Mambo Number 6 retweeted
2.3% absolute reduction in 5-year mortality for the ARR teeth gnashers out there.
jamanetwork.com/journals/jam…
Home truths right here
This is what a 5’11’ ~200lb genetically-gifted-by-the-gods NFL running back looks like in street clothes.
In other words, about the best or “biggest” most ~6’ guys can be naturally.
Most of you are chasing a physique that simply doesn’t exist outside of heavy steroid/PED use.
109




























