A study of 18,150 adults from the National Health and Nutrition Examination Surveys (NHANES) between 2009 and 2018 were analyzed for reported calorie intake, comparing it to predictive equations. Underreporting occurred almost twice as often in participants reporting low-carb and low-calorie diets as compared to the general population. The underreporting was worst among the low-carb dieters.

Overview

• What did they test? The researchers used a predictive equation from 6,497 doubly-labeled water measurements (an accurate measure of free-living daily energy expenditure) to detect erroneous self-reported calorie intake in 18,150 people from the National Health and Nutrition Examination Surveys (NHANES) from 2009-2018. The researchers compared underreporting prevalence between people without a special diet, people reporting low-calorie diets, and people reporting low-carb diets.
• What did they find? Underreporting occurred almost twice as often in people reporting low-calorie diets and low-carb diets. Both diets were significantly associated with higher odds of underreporting even after adjustment for sociodemographic factors. Results were similar when looking at people who denied any weight loss attempts and people with stable body weight. The largest disagreements between actual and reported intakes occurred in the low carb group.
• What does it mean for you? There is no “metabolic advantage” to low carb diets. Studies that rely on self-report of calorie and/or carbohydrate intake are not reliable for determining whether any metabolic advantage might exist (and tightly controlled metabolic ward studies show that it doesn’t). Individuals who report low calorie intakes, and especially individuals reporting low carb diets, are more likely to be underreporting their calorie intake compared to people not on these diets. Thus, always be wary when individuals claim to “not eat anything” or are cutting carbs yet are not able to lose weight. They are likely just underreporting and eating more than they think. Finally, given the prevalence of underreporting, the conclusions of any large scale studies on nutrition that involve self-report need to be viewed with some caution.

What’s the Problem?

In 2006, I published a meta-analysis on the effects of variation in protein and carb intake while dieting 1. After controlling for calorie intake, I found that lower carb intakes were associated with greater weight and fat loss compared to higher carb intakes. At the time, I concluded that perhaps low carb diets had some sort of metabolic advantage over higher carb diets.

However, science is always gathering more data, and as a scientist, I will change my mind when presented with new evidence that overturns my previous thinking. In a 2010 blog post, I discussed a series of well-controlled studies that indicated that there was no metabolic advantage to low carb diets. That conclusion became further cemented with well-designed research by Dr. Kevin Hall 2 3 4 .

So why did my meta-analysis show a metabolic advantage when further, better controlled studies did not? One problem, as I outlined in that blog post, was I relied a lot on studies that involved self-report of calorie intake. If there’s any difference in the accuracy of self-report between different types of diets, then that would bias the results of a meta-analysis.

We know from a large body of research that self-report of calorie intake can be unreliable, with people having a tendency to underreport their calorie intake 5. The degree of underreporting can be quite extreme; with ranges of 764 calories per day 5 to over 2000 calories per day in certain individuals 6. Even professional dietitians underreport their calorie intake by over 200 calories per day on average 7. With so many coaches and dietitians that rely on the self-report of their clients, and, on a larger scale, so many nutrition studies that rely on self-report, you can see how this can be a problem.

How do we know that people are inaccurate in assessing calorie intake? We know that in weight stable people, energy intake will equal energy expenditure. So, scientists use a technique called doubly labeled water. With this technique, people drink water that has two stable isotopes.  The elimination of these two isotopes from the body over time (usually a 1-2 week period) is used to determine carbon dioxide production.  This carbon dioxide production can then be used to calculate the number of calories you burn.  The doubly labeled water technique was actually invented in the 1950's, but its use was limited to small animals.  It wasn't until the 1980's that technology had advanced enough, and that isotope costs had decreased enough, where it could be performed in humans.  The method can accurately measure your 24-hour energy expenditure within ±5%.

So, if we have a weight stable person claiming to be eating 1200 calories per day, and doubly labeled water comes back and tells us his/her energy expenditure is 2200 calories per day, we know the person is underreporting calorie intake by 1000 calories.

The question then becomes as to how to apply this to large scale studies, like epidemiological studies, where we already have a bunch of self-reported data. A massive group of scientists led by Rania Bajunaid developed a prediction equation from 6,497 doubly labelled water measurements to allow for detection of errneous self-reported calorie intake 8. This prediction equation can then be applied to large epidemiological data sets like the National Diet and Nutrition Survey (NHANES). In fact, when this has been applied to NHANES, it’s been found that the level of calorie intake misreporting exceeded 27% 9. Bajunaid also found that the macronutrient composition from dietary reports was “systematically biased as the level of misreporting increased.” This means you can get potentially false associations between diet components and body weight in epidemiological studies.

We know that undereporting occurs more frequently with unhealthy foods, and is also more prevalent in obese individuals and in females 10. The question then becomes as to whether underreporting might systematically vary between groups on special diets, like low-carb diets. Researchers from Germany and Uruguay decided to investigate.

Purpose

The purpose was to investigate the prevalence of underreporting in people reporting low-calorie and low-carb diets relative to the general population.

Hypothesis