People Are Bad At Estimating Calorie Intake (New Study)

A new study set out to investigate the level of “misreporting” in Calorie intake. In this context, misreporting means under- or over-estimating intake by a large enough degree to mess with a study’s results.

Various different instrument exist to assess intake based on self-report, e.g. food frequency questionnaires. Each method has some degree of expected error that, when combined with the day-to-day variability in Calorie intake, give researchers an idea for how many subjects they need to include in their study in order to “make up” for that expected error. This is what people are talking about when they say a study is “adequately powered.” If the data collected is far worse than expected however, these calculations would change.

This new study set out to see how much people are likely to mess up when reporting their Calorie intake. To do so, researchers compared doubly labelled water (DLW) measurements of over 7,500 individuals to predicted energy expenditure based on self-reported energy intakes.

The DLW technique measures energy expenditure directly from the elimination of isotopes of oxygen and hydrogen introduced into the body in water and has an analytical error of ~ 7%. An equation was used to predict total energy expenditure of 18,567 individuals based on their reported intake and objectively measured variables like activity levels tracked by an accelerometer, body composition measured by DXA or similar, and so on.

This approach created a “predicted range” for viable dietary intakes. Ideally, any study measuring dietary intake would have the majority of its data fall into the predicted range, otherwise we should be suspicious of misreporting.

So, what did they find? A brief summary:

  • People primarily underreport energy intake to the tune of ~ 400-800 Cal/day on average
  • The degree of under-reporting is higher in kids (kids probably need easier to understand tools for nutrition)
  • The degree of under-reporting is higher in those with higher BMI’s
  • The degree of under-reporting is higher in those with lower protein intakes and higher fat intakes (desirability bias?)
  • Over half of the self-reported data collected is suspicious for misreporting

The authors suggest that using this information, we can reduce the amount of misreported data included in nutrition studies.

I would like to see a few examples to see how this works. I suspect that many nutrition studies would become underpowered after elimination of poor quality data post-hoc, though I’m not sure any general heuristic in nutrition would change, e.g. low carb diets NOW produce more weight loss! (They don’t)

I think that using this study’s equation for predicted energy expenditure during research to identify misreporting on the spot, which could inform the researchers to include more people to ensure adequate powering.

Finally, this study confirms that people are not very good at reporting their dietary intake. This is true even in trained professionals. For example, one study showed registered dietitians underreported their intake by 223 Calories per day, whereas the controls underreported their intake by 429 Calories per day. Champagne 2002 That individuals with higher BMIs, lower protein intake, and higher fat consumption tend to misreport more also corroborates previous data.

I don’t think this affects weight management, but rather serves as a reminder to be skeptical of strange nutrition-related results, e.g. unexplainable weight change (good or bad).