Low Carb Diet DATA Distortion

Major scientific journals and media outlets have a pattern of distorting reality when it comes to keto and low-carb diets. Why? This may be in part just because of an incentive structure. Keto bashing has become a sport. You get a lot of media points for it and attention points, which are a valuable currency. But while this video may seem like it's about defending keto and low-carb, it's not. It's about intellectual integrity, raising the bar on intellectual integrity so so that we don't promote confusion and mistrust. And that's the big problem, the big problem I want to address, and I want you to help me combat. Because while these outlets may rationalize their approach as promoting safety and public health, the clear effect of these distorted datasets and manipulation of the data is confusion. And confusion breeds mistrust. And that's a problem for all of us. So in this video, I'm gonna go through 3 examples, and you, by the end, can evaluate for yourself if they're being honest about the data and data communication. And then we can start having a conversation. Welcome to my channel. Stay curious. The first paper we're gonna address dropped yesterday as I record this in Jack Advances, and it's entitled Association of a Low-Carbohydrate, High-Fat Diet with Plasma Lipid Levels and Cardiovascular Risk. I'm gonna start by reading some sections from the abstract, straight from the abstract, provide a little bit of commentary, then we're gonna dig into the data. So, objectives. The purpose of the study was to investigate the association between low-carb, high-fat, LCHF, dietary patterns, lipid levels, and incident major adverse cardiovascular events. In a cohort from the UK Biobank, including ultimately what was 2,000, 34 low-carb, high-fat diet participants who had completed at least one, one 24-hour dietary questionnaire. We'll get to that in a little bit. And the low-carb, high-fat diet was defined as less than 25% calories from carbohydrates and greater than 45% calories from fat. And then they matched the low-carb, high-fat participants 1:4 to standard diet participants as a comparator group. And they found that in the LCHF group, there was higher low-density lipoprotein cholesterol and ApoB. These levels were significantly increased. LDL and ApoB were significantly increased in the LCHF group versus the standard group, p 0.001. Pause here and notice something, or rather notice something that's not there to be noticed. They don't say what the difference was. They don't say what the absolute difference was, and this becomes very relevant momentarily. Anyway, they say that there was a higher rate of incident major adverse cardiovascular events in the LCHF group with a hazard ratio of 2.18, meaning an over twofold higher rate of major adverse cardiovascular events. Again, talk about that large magnitude effect, a relatively large magnitude effect, a twofold increase in MACE versus— again, they don't say what the difference in LDL is, and there are going to be some other factors that complicate this as well. So Bank that in your mind. And then they conclude, "Consumption of a low-carb, high-fat diet was associated with increased LDL and ApoB and increased risk of incident major adverse cardiovascular events." So what they appear to be doing here is saying, you eat low-carb, high-fat, your LDL and ApoB go up, and it more than doubles your risk of major adverse cardiovascular events. That's what they're highlighting in the abstract. Now let's characterize the low-carb, high-fat group. There were 2,034 individuals in the group, and the paper notes, as highlighted in supplemental tables 2 and 3— now it's a little bit weird to put the characteristics, I think, in a supplement, but that aside— the low-carb, high-fat participants were more likely to have diabetes, had much higher obesity rates, 25% or 24.6% in the low-carb, high-fat group versus 18.7% in the standard diet group, and higher BMI. Plus, if you go to the supplement, you'll actually note that the low-carb, high-fat group also had higher smoking rates. So the low-carb, high-fat group was biased towards being much less healthy in general. Higher obesity, higher BMI, higher diabetes, and more smoking. Okay? So that is a huge potential confounder. Now, let's juxtapose that to the magnitude of the difference in the scapegoat here, the LDL cholesterol and the ApoB, because remember, they omitted that information in the abstract. Wonder why. If you actually look at the data, you'll see that the low-carb, high-fat group, yeah, they had statistically significant higher levels of LDL and ApoB, but how much was it? It was a 3.48 mg— that's tiny— difference in LDL and a 0.03 g/L difference in ApoB. You can see the distributions here. In Figure 1. And what's clear is that the ApoB-LDL difference is actually very, very small. So while it might be statistically significantly different, the clinical difference is actually very tiny. So tiny that it's really, really hard to believe that the twofold increase in major adverse cardiovascular events is due to this tiny clinical difference. And then again, you juxtapose that to the fact that the low-carb, high-fat diet group had more obesity, more diabetes, and more smoking? And I'm left to conclude that I think most of the effect is due to confounding, if not all of the effect. But that's not all. Let's move on to the measurement tool they use. So let's talk about their data collection tool. For most participants, they used a single, a single 24-hour dietary recall questionnaire, asked participants what they ate yesterday. And they admit, this is a quote from the paper, that this tool is prone to measurement error and recall bias, and is a short-term, yesterday, assessment of dietary pattern which may not accurately reflect longer-term food intake. That's a direct quote. What's more, if you reference dig, go to reference 14, then link to the questionnaire, go to page 13 of that, you can see instructions to read to participants statements like this: Yesterday may not have been a typical day of eating, for you. That's okay. We are still interested in what you ate and drank. So they're asking participants, what did you eat and drink yesterday? It might not have been typical, we don't really care. And then drawing from that conclusions about how their diet relates to their lipid levels and how that relates to cardiovascular risk when it's framed and confounded by the fact that this group has higher rates of obesity, smoking, and diabetes. And if you actually look at the data collected from the questionnaire, there are some suspicious things, like the fact that the low-carb, high-fat diet group reported eating about 500 calories per day fewer calories than the standard diet group, despite having higher obesity and diabetes rates, and of course reporting eating lower carbohydrates, since that's the definition of the low-carb, high-fat group. And irrespective if you're carbohydrate insulin model camp or energy balance model camp, the fact that there were much higher obesity and diabetes rates and reported much, much lower calorie and carb intake should make you question the validity of the data collected, especially when the authors themselves are admitting the tool they're using is prone to measurement error and recall bias and may not reflect what patients are actually eating over the long term. So that is the data collection method they're using for this paper. So in summary, my high-level thoughts about this paper are that the data are based on a poor collection method that has measurement error, recall bias, and may not reflect habitual diets. The low-carb high-fat group was definitely not even keto. The average ketone levels were about 0.1 or 0.14 milligrams per deciliter. They had higher obesity, higher diabetes, and higher smoking rates. The difference between the LDL and ApoB levels of the low-carb high-fat diet group and the standard diet group were tiny clinically speaking. 3.48 milligrams per deciliter difference in LDL and 0.03 grams per liter difference in ApoB, which cannot possibly account for the twofold difference in major adverse cardiovascular events they're reporting. That, framed by the fact that they had— the low-carb, high-fat diet group— higher rates of obesity, diabetes, and smoking, really suggests to me that the results are primarily the results of poor data collection and major confounding. So I really would not make much of these data other than they're more low-carb diet fearmongering. And actually, I find it quite heartening that people outside the, quote, low-carb diet camp are agreeing with me that this is bad data, bad data reporting, and it stirs confusion, which is not productive for any of us, except maybe for the journals and media headlines that get good clicks off of keto bashing. Now, with that, moving on. Now I'm going to go through two more examples, and I'm going to be a little bit more brief on these just to establish a pattern. So for example, a semi-recent headline ran, huge global studies Find low-carb or keto diets could shorten lifespan. So they're claiming, literally in, in the headline, keto diets could shorten lifespan. But if you go to the study, which was published in The Lancet Public Health, they looked at quintiles of carbohydrate intake. And if you go to the lowest quintile of carbohydrate intake, the lowest carb intake group, what is the threshold for kilocalories from carbs? It's 37% kilocalories from carbs. For reference, that's about the same as kilocalories from carbs, as percent, as a Burger King Whopper with fries. I did out the math. The Whopper with the bun and fries is about 36-37% kilocalories from carbs, and that is what they're pinning in the headline as a keto diet. Keto diets can shorten lifespan. In fact, if you go to the paper and Ctrl+F, you will find the word keto or ketogenic diet mentioned zero times, and yet the headline has the audacity to claim ketogenic diet shorten lifespan? I think that's absurd. But moving on, the third and final example has to do with research myself and colleagues have been doing on the lean mass hyper-responder phenotype, which has definitely been controversial. It was an n 1 case study published in the prestigious journal Circulation and presented at an American Heart Association conference. It's entitled "Rapid progression of coronary artery disease after stopping a statin and starting a ketogenic diet in a phenotypically lean mass hyper-responder." Now, for a little bit of background, as to my initial reaction upon reading just the title, I was thinking, "Oh, you know, n 1 case study, and they're claiming a lean mass hyper-responder went on a ketogenic diet, had rapid plaque progression." And I thought, well, if there are thousands to tens of thousands of lean mass hyper-responders, This has probably happened in at least one. So this is a cautionary tale. Okay, that's fair enough, and I will be more than happy to amplify this as a cautionary tale. But then I read the case, and I was appalled because the title turned out to be a lie, frankly. So let's break it down. First of all, they say in a phenotypic lean mass hyperresponder. The patient was not a lean mass hyperresponder, so that's lie one. Now, I don't really care that much about the physiological cutoff, so you could say maybe he was approaching lean mass hyperresponder status, So we'll even bucket that as no biggie. But they're claiming in the title he stopped a statin and then went on a ketogenic diet and then had rapid progression of coronary artery disease. But if you look at the actual case, what really happened in this patient was this patient, a 51-year-old male at the time, had a prior, prior to starting a ketogenic diet, intervention for severe coronary artery disease. They had to go in because his left anterior descending artery, one of the major arteries that comes over the front of the heart, was occluded. So they went in, and they were intervening. And while they were intervening, they noticed, OK, not only is his LAD occluded, but he has moderate disease, which they didn't quantify, but moderate disease in his right coronary artery. So at that point, they start treating him with statin therapy, for which he's on for a couple of years. So he had major cardiovascular disease, coronary artery disease. They put him on medication and let him progress for a couple of years with no more measurements during that time. They're not measuring plaque progression. And then they say at some point in time he stopped the statin and started a ketogenic diet. They don't say what the ketogenic diet looked like. They don't say even how long he was on it. So this could be a case of a guy saying, I went keto, but he was really just eating a lot of bacon for a week. And then he had obstruction of his RCA. And then they pinned that all on the, quote, ketogenic diet, saying he stopped statin, then went keto, then had rapid plaque progression, but I think you can see from the case that is not at all the reality. He was a non-lean mass hyper-responder who was on multiple years of statin therapy without any further measurement after he had a major event, and then he tried a, quote, "ketogenic diet" without any formal definition or duration, and they're blaming his RCA occlusion on the ketogenic diet? That's absurd. It's frankly a lie, and that this got published Even as an abstract in a major cardiovascular journal, Circulation, is appalling. I consider it frank misinformation. And actually, I tried to write a letter to the editor, but they don't accept them for abstracts. So I had to go towards just talking about this on social media, which is not my preferred method. But again, this is one of the things that I showed it to cardiologists and lipidologists, even those who are not low-carb, or one could consider even anti-low-carb, and they agreed, "Yeah, this is garbage, and it's stirring confusion." So those are the 3 examples I wanted to leave you with. Now I want to reemphasize a major point. This is not about just defending keto or low-carb. The issue of data distortion and manipulation does translate to other fields, and I think in general what we need to do is hold data to a high bar of integrity, and when we see data distortion and misinformation, call it out. And I try to do this even within my, quote, "own camp." This video in particular is the topic of distortion of low-carb and ketogenic diet. Datasets and dissemination of information, because quite honestly, I do think that there is an unequal bias against low-carb and ketogenic diets, and I think they have tremendous potential as metabolic therapy. So it's where my emphasis is. But again, in general, this is about intellectual integrity and holding us all to account for a high level of intellectual integrity so there isn't confusion and frustration, which is just in general poor for public health understanding and trust in medicine. So thank you for listening. I look forward to your comments and have a good day.