Published today: A US-based epidemiological analysis reveals large discordance between exclusive dietary recall-based vegetarian status assessment and self-identification in a nationally representative population-based survey
Our paper "A US-based epidemiological analysis reveals large discordance between exclusive dietary recall-based vegetarian status assessment and self-identification in a nationally representative population-based survey" was published today in the journal Scientifica (Wiley).
It might seem obvious to the average rational thinker that just because someone did not eat any non-vegetarian food in a 24-hour window does not make that person a vegetarian.
But that kind of straightforward, rational thinking is not always popular among scientists who write academic journal articles. It is not uncommon to have people who eat meat, fish, dairy, and eggs rarely be described as "vegans". It is even more common - maybe more common that correct definitions - to see non-vegetarians (such as pescatarians and semi-vegetarians/flexitarians) be defined as "vegetarians".
Add to this that many - possibly most - people who describe themselves as "vegetarians" are non-vegetarians, i.e., they do eat animal flesh, often fish and/or "seafood", or even meat (bird or mammal flesh).
In the cited paper, we provide an example from the famous and infamous NHANES dataset (National Health and Nutrition Examination Surveys, USA).
If one were to assume (which our analysis did) that people who provide a 24-hour recall of their diet (i.e., what they ate on one specific day), which includes only vegetarian foods, are in fact (quite a non-fact) vegetarians, then one would come to the conclusion that more than 14% of the people in this dataset are vegetarians. ... But if one were to raise the bar higher for those who can be categorized as "vegetarians", by combining the 24-hour recall data with data on self identification (i.e., answering "yes" to the question "“Do you consider yourself to be a vegetarian?"), the one would come to the conclusion that only about 1% of this dataset are vegetarians. Considering only the second criterion (i.e., self identification) would lead to the conclusion that about 2% of the dataset are vegetarians.
As this mathematical example from this large dataset in the USA shows, these prevalence estimates differ quite drastically from each other: 1% vs. 2% vs. 14%.
And the real vegetarian prevalence is unknown, because we can reasonably assume that the 14% (based 24-hour data only) is obviously (and unfortunately) nonsense. But the self-identification data (2%) is also an overestimate - this can be assumed based on many studies and the combination of the 24-hour data and the self identification (i.e., the 1%). But combining two overestimates only lowers the number (vegetarian prevalence in this sample) towards the "true" vegetarian prevalence. The 1% respresent all the participants who (1) said "yes" "I am a vegetarian" and who also reported (2) only vegetarian foods on that one particular day.
Take home message: We can assume that the vegetarian prevalence in this NHANES dataset (2007–2010 data) is probably lower than 1%.
Which is a shame.
Let's hope one day affluent countries like the USA can boast a true vegetarian prevalence of 10% and more.
All the credit for the paper is due to the first author, Maximilian A. Storz.
