Why Research Often Makes Low-Carb Diets Look Bad

It’s often by design

I got into a discussion with another reader (I’ll call him Mr. R) in the comments section of an article here on Medium. We were discussing the inflammatory effect of saturated fat. Mr. R offered a randomized controlled cross-over study as support for his POV that saturated fat was, indeed, inflammatory.

After looking at the details of the study, I replied that the diets tested were all high-carb diets — and that under low-carb conditions, saturated fat is actually known to be anti-inflammatory.

Mr. R challenged my categorization of the tested diets as being high-carb by first quoting the research report, then adding the details of his diet. He wrote:

“ ‘In a randomized crossover design, 50 men consumed controlled diets for 5 weeks that provided 15% of energy from protein, 39% of energy from fat, and 46% of energy from carbohydrate.’ Let me repeat ‘39% of energy from fat’ and only ‘46% of energy from carbohydrate’. And you call it high-carb diet??? I eat high-carb diet. It is less than 10% fat, 10% protein and 80% carbs.”

For reference, the Standard American Diet (the one recommended by the USDA) recommends between 45% and 65% of daily calories come from carbohydrate. In nutritional science, a “low-carb” diet is presented as an alternative to the SAD. Given that the test-diets all had a carb content within the SAD recommendation, they cannot be classified as “low-carb.” They might not be “high-carb,” as I classified them, but they are certainly NOT low-carb, which was the condition under which I had said saturated fat was anti-inflammatory.

And this got me thinking about how some researchers stack the deck against low-carb diets in their studies. I don’t think the researchers of the paper Mr. R cited were doing anything shady — they don’t actually refer to the diets as being low- or high-carb. They just report the percentage of calories coming from the 3 macronutrients — it was Mr. R who categorized the diets as “low-carb.” Coming from a baseline of his 80% carbs, it makes sense that he saw 46% as being low. He probably didn’t know what the SAD recommends or how “low-carb” is defined.

A low-carb diet is typically considered (at least according to a review conducted by researchers at the Mayo Clinic) to be between 20 and 60 grams of carbohydrate per day. So that would be in the range of 5% — 15% of calories for someone on a 1,700 calorie/day diet; 4% to 12% for someone on a 2,000 calorie/day diet.

But why did Mr. R’s misunderstanding remind me of what I consider “shady” researchers?

Well, the rise in popularity of various low-carb diets (Carnivore, Keto, Banting, and others) has proponents of their own, higher-carb diets in a bit of a frenzy. Some actively push for a plant-based or even vegan diet for philosophical reasons but hide those personal reasons behind claims that sound like they are based on science — research that has led to definite causal conclusions about the inferiority or at best, parity, of the low-carb diet when compared to their own Pet Diet.

There are three basic types of shadiness a research team can engage in to make low-carb diets seem to be inferior to, or at parity with their Pet Diet.

Scare Tactics

Diet and nutrition research is often observational. Thousands of people start the study by answering a long questionnaire asking them to detail what they have eaten in the past six months and various activities in which they have participated. This questionnaire might be given to these same people several times over the subsequent 10 or even 20 years.

At the end of the study (so, maybe 20 years after it started), the researchers look at how many of the thousands who started it died during the period of the survey, and what they died of (cancer, heart attack, hit by a bus, etc.).

Then, the researchers correlate what the dead people did and ate (or failed to do and/or eat) to see if anything stands out. So, they look to see if the dead people ate more meat or less meat than those who didn’t die; how did vegetable consumption compare between the two groups; and so on. This kind of research, unfortunately, can’t lead to a valid conclusion of causation between any nutrient and any particular health outcome. The reason for this is that it’s impossible to isolate the effect of any one food or behavior on the health outcome. This kind of “observational” research is useful in identifying things that might end up causing the health outcome of interest, but on their own, they simply cannot determine causation.

If you ever see research mentioned in an article and it says it was based on even 1,000 people, chances are it was observational research — which cannot lead to a valid conclusion of causation. Studies of this type will often mention that a “link” between Food X and Health Outcome Y was found. But a link is not a cause. And this is where it can (and often does) get tricky: The researchers can overreach the importance of their findings by saying something like ‘it is our opinion that meat causes cancer,’ or ‘given the link between meat and cancer, we cannot recommend people eat meat.’ The media then pick up these quips and announce that researchers found that meat causes cancer.

There is, however, a research design that can determine if Food X causes Health Outcome Y. It’s called a randomized controlled trial (RCT). In an RCT, tight controls are kept on as many variables as possible in the attempt to isolate the effect of a single factor (like the amount of dietary animal protein) on health outcomes. Such research is hugely expensive — imagine needing to ensure that the people in your study eat specific diets over a period of time. It’s likely impossible to conduct an RCT over a period of 20 years, even if one had the funding. So, fewer people are used in the study (maybe 30–50), studies last for shorter periods of time and the researchers measure health “markers” like body mass, cholesterol, blood pressure, etc., instead of deaths.

You’d think that RCT would be immune to shadiness. But they aren’t. Here are two ways RCTs can be manipulated.

Cut Short

It can sometimes take up to six months (26 weeks) for low-carb diets to show their full benefits. For that reason, some Pet vs Low-Carb Diet research will be designed to be only 12 weeks, maybe 24 weeks. With this short duration 2 things happen: The low-carb diet won’t have had time to fully kick in and the Pet Diet won’t have had a chance to stall and revert to weight gain. Researchers can then point to their data and say low-carb did no better than their Pet Diet.

I actually remember seeing a study that said it was a 26-week study, but participants were measured at the start of the study, and then again at Week 24. My guess is that they were also measured at Week 26, but the results made the Pet Diet look bad, or the Low-Carb diet look good, or both and those data weren’t reported.

Play with Definitions

Imagine a shady scientist knows his or her colleagues are on to the “cut short” ploy, so s/he designs a 30-week study or even one lasting a year. In those scenarios, if a true low-carb diet is tested there is a chance (it turns out a good chance) that it will do better than their Pet Diet. If they test a true low-carb diet, the data will show its advantage over the Pet Diet.

But the shady scientist can set the carb in the “low-carb diet” to a level that’s not so low — like 40% of calories. That’s lower than the SAD, but not really a Low-Carb diet.

With this level of carbs, there are no benefits over even the SAD. But the researcher can label all the charts and graphs in the report “Low-Carb Diet Group” and show their health markers to be on par with their Pet Diet. They can also put the term “low-carb” in the title of the paper and discuss their conclusions that the “low-carb” diet did no better than their Pet Diet.

So if you see people citing research that claims Food X is will kill you or that a Low-Carb diet didn’t do any better than the Pet Diet of the day, dig a little deeper: Was it an RCT? If so, was the low-carb diet given enough time to show its full benefits? Was it really a low-carb diet — 60 grams of carbohydrate per day maximum? If not, you are probably looking at shady research.

There are a lot of competing interests at play, even in what we might consider academic research. So it’s up to us to stay alert and to be skeptical. I recommend reading beyond the headlines and into the details before accepting what even articles in prestigious journals publish.

Thank you for reading this article — hopefully it contained something you found useful.

If you aren’t a member of Medium but are thinking of joining, please join through my page! If you do sign up to Medium through my page, some of your membership fee goes to me (but you still pay the normal membership price).

With a paid membership to Medium, you will get to read more of my work plus you get unlimited access to thousands of Medium writers. And it’s only about $5.00 a month!

I wrote a book about nutrition, metabolism and weight management in which I cover topics like this in more depth — but not so much that you need an advanced degree to understand it! It’s available on Amazon. See my profile page for more details, if you are interested.