“Is keto bad for the heart?” ie. can Keto cause heart disease – another study in the headlines. What do you think?
Another paper is out, or I should say an abstract, because an abstract doesn’t really give you the whole paper to be able to comment on. It was a news release from the American College of Cardiology: Low-Carbohydrate High-Fat “Keto-Like” Diet Associated With Increased Risk of CDV (cardiovascular disease). What does “keto-like” even mean?
My initial take on this is it’s much ado about nothing again. Let’s just go through and next time maybe you’ll just take my word for it that it’s not worth going through this paper.
Language use is not precise
What did they do? Again, I’m waiting for the randomized trial of taking people on a keto diet, or any other diet for that matter, and seeing if there are heart disease outcomes that are different over time. That would be an experimental study and I would put a lot of emphasis on that. Of course, then, the hallmark of science is replication; you’d always want to see the second study as well. This paper is not a randomized trial or a prospective study looking at a keto diet. I’m not even sure what a “keto-like diet” means. Certainly, this is a group from the UK and they presented at the American College of Cardiology, which you would think would be a reputable organization, but as I’ve found, most physicians lose their minds when it comes to nutrition. Suddenly, they don’t have to be really careful about the methods and you can just use words very flippantly, like “keto-like.”
Data collection and methods
The presentation was done using data from what was called the UK Biobank. Now, we’re getting into this world of nutritional epidemiology. What that means is you follow people over time; you don’t tell them what to do; you just ask them what they do in various ways and sometimes you don’t ask them very much about what they’re doing but you take their word for it. The UK Biobank has 70,000 people there, and what they’ve done is they’ve assembled people with their clinical information, meaning what they’re like and their characteristics, and then they’re also looking at the genetics to try to match up the genes with the health outcomes. This report says they followed people over 11 years, which is consistent with the UK Biobank information.
The major problem is they found 305 people out of 70,684 who met the criteria of a “keto-like diet.” What does that mean? First of all, 305 out of 70,684 is .0043 percent of the entire group in that dataset. How do they figure out what they were eating? There was what’s called a “24-hour diet recall,” which is basically, “Please fill out what you ate and drank over the last 24 hours.” The methods say: “Upon enrollment, a 24-hour diet recall and cholesterol level was taken” – 10 years ago. If you wrote down what you ate and drank and you said that you had less than 25 percent of your calories from carbs and over 45 percent of calories from fat, then, based on that 24-hour diet recall, you were categorized as “keto-like.” A 24-hour diet recall, taken once, 10 years ago is the information they have to categorize people. “Keto-like” is like saying, “You’re pregnant-like.” What was it actually?
If you follow a plan like I teach, and taught at Duke, and published papers on this plan, it’s more like 5 percent or 10 percent of carbs per day. (You don’t have to eat the same amount every day.) What if this person, 10 years ago, had a bad carb day? They may have over-consumed or under-consumed or didn’t eat anything because they may have been sick, it was categorized as “keto-like,” which gets into this whole idea of confounding – meaning interference, or is there something else going on? We don’t have any idea what they were eating. They might not have eaten anything but a stick of butter, for example. They would be categorized as “keto-like.” One major problem is people who weren’t feeling well might have underreported what they were eating.
Key flaws in the data and methods
The term “keto-like” is meaningless to me. It really smacks of this whole idea that people are taking potshots at keto when this wasn’t looked at in a credible way; that just reduces your credibility. If you’re a researcher, you want to be very precise about the measurements you’re taking. A food recall 10 years ago and then seeing what happened to people is just not sufficient. It really doesn’t matter what they found now because they had 305 people out of 70,000. The American College of Cardiology had a control group, a selected group, not a randomized control group. They went back to that original 70,000 people and found 1,100 people who were matched by age and gender. Again, this just makes me wonder, there must be more to this in terms of the filter of total people in the Biobank. How they got assembled into these groups needs to be described. That is what’s called the consort statement for clinical trials – “the filter” of the study. Along every step of a prospective clinical trial, you have to say how many people dropped out here, how many stayed in here, how did people get into the study, and what happened during the study. We have no information here. We go from thousands of people down to 305 “keto-like” followers.
The outcomes around keto causing heart disease seem inconclusive
I want to stop here, but wait, it gets better. Then, they look at the people they’ve assembled, which is like flipping coins or rolling dice, and they found that 9.8 percent of the outcomes – diabetes, heart attack, or stents, called a combined outcome. How you follow people in clinical trials is fascinating. The combined endpoints and overall mortality always go hand-in-hand. You don’t want just these combined endpoints; you also want mortality rates. In some studies, the prospective clinical trials, there’s higher mortality or there’s no change in mortality over time, which is really the endpoint you want to have consistent with these other combined cardiac endpoints.
If you looked at these 305 people following a “keto-like” diet, they had 9.8 percent combined clinical outcomes. Let’s think about that. If you’re taking people who are generally healthy, this is not a great thing to have over 10 years, but if you’re taking people who have diabetes, hypertension, and obesity, 10 percent events over 10 years – everyone didn’t die! The preconceived idea here is that on a keto diet, you’re going to get a heart attack. Of course, I don’t see that in my clinic and we don’t see those signals in clinical trials. But they said they were 9.8 percent compared to a 4.3 percent rate of stents (tubes you put in the coronary arteries), heart attacks, stroke, and peripheral artery disease in those eating other than a “keto-like” diet. Really? That was the end result – saying that there’s twice as much risk of cardiovascular disease when following a keto diet, going from 4.3 to 9.8 percent over 10 years in a group of 305 versus 1100 people from a sample of 70,000 people. I have no idea how these people got assembled. It would be an eye-opener if there were deaths that were found; we don’t have information about mortality here either.
These studies are going to be coming out over and over again. This group also reported the LDL levels being a little high, but by now, the type of LDL really matters. If it’s large LDL particles, it doesn’t have the same risk as the small LDL. If you’re just looking at an LDL you’re not really getting all the detail on that – this is all they did is they looked at LDL. I remember a conversation with a nutritional epidemiologist, a very famous physician-researcher – he said, “That’s the best data we have, these small associations.” That’s not good enough. If you’re going to go on a 24-hour diet recall, taken once or even periodically over that period of time, that means you didn’t go home and follow what they ate. If you’re going to have drug approval, the general consensus is that you count how many pills are taken; you want to make sure that they’re actually taking the pills of the drug that’s being studied to know what is actually happening. Pill counts are commonly part of the study. Here we have no validation of what people were eating. It’s sad how scientists lose their mind in terms of precision and accuracy about information when it comes to nutrition. It’s sad, because good nutrition is more powerful than medications and if you do it right, you can get people off medicines and reverse the medical problems at the same time.
The bottom line is that this study doesn’t show much about people eating a well-formulated, adequate protein keto diet. Someone could have just been eating pork rinds, for example, and they would be in the “keto-like” category in this study when that’s not what we recommend. You have to have whole foods and good foods in a program that we recommend or that’s generally recommended today by reputable people. It’s sad that the scientists who did this thought they were coming up with something important. Maybe they were trying over and over and over to see that keto was bad – that’s the implication in the title. It’s also sad that the American College of Cardiology, a reputable organization, must have had some sort of vetting process for the type of science being presented and that this is really not worthy of a presentation at a national meeting anywhere. You must use great methodology when you’re doing science on diets and nutrition, which deserve the same attention that clinical trials of medications deserve.
Watch the full video here.