First off, there is most certainly a real medical condition called celiac disease. It involves a measurable immune reaction to foods that contain gluten. Gluten is a protein-starch complex (but often described in non-technical and medical writing as just protein) found in some grass seeds, such as wheat, rye, and barley, but not others like rice and corn. It is not found in any foods other than those made from these specific grains.

Celiac is apparently found in about 1% of the population.

Some medical authorities and lay people also assert the existence of “Non-Celiac Gluten Sensitivity” (NGCS). As you might expect, this describes people who are sensitive to gluten in some way, but not through the (partially) known mechanism of celiac disease. To quote a BBC article by “Doctor Chris” van Tulleken, “Although many people who do not have coeliac disease claim to suffer gut symptoms like bloating and nausea when they eat gluten – and even other things like “brain fog” and tiredness – these have not been linked to any physiological changes that can be measured and hence used to make a clinical description and diagnosis.”

Someone with my background wonders if a claimed syndrome that has such a wide variety of symptoms which correlate to nothing measurable might be a result of pareidolia, the tendency of the human brain to find patterns in noisy data, even when no such pattern is real. Examples of what I have just named “pareidolic diseases” include Morgellon’s Syndrome, chronic Lyme disease, and Candidiasis/Yeast Allergy. They share having relatively common, non-specific symptoms (e. g. fatigue, stomach upset) and not being detectable by any lab tests, even when (as with Lyme disease) there is a well-known, effective test available.

Dr. van Tulleken’s “Trust Me, I’m a Doctor” team conducted a study to test the reality of NCGS. The found 60 people, some of whom claimed to experience NGCS, and put them on a gluten-free diet including (experimenter-supplied) gluten-free pasta, but for two weeks (randomly selected?) of the study period, each was given pasta containing gluten instead. The team then analyzed the subjects’ reported symptoms, and also performed lab tests to show inflammation and IgE (antibody) levels. The study was double-blinded, in that neither researchers nor subjects knew in which two week period the subjects were eating gluten until after the study concluded. The subjects acted as their own controls by experiencing both with- and without-gluten diets.

The results: subjects did in fact report feeling fewer gut-related symptoms like bloating during the gluten-free weeks. Variation in other symptoms (e.g. tiredness) was not statistically significant. There was also no significant correlation between gluten-free diet and the lab values being measured.

Dr. van Tulleken does state in the linked BBC article that it’s likely at least some subjects knew which weeks they were eating gluten-containing pasta. And as he concedes, this can easily explain the slight variation in the digestive symptoms.

A weakness he apparently did not realize in his analysis: the two pastas are very clearly made with different recipes. It’s perfectly possible that the gluten-containing pasta that was used has some other difference from the gluten-free one that explains the gut symptomology, perhaps an herb or a preservative that some people are really slightly sensitive to. It’s  possible that the gluten-containing pasta just has less fiber (for one possible example), and that the additional fiber in the gluten-free food aided the subjects’ digestion. Dr. van Tulleken doesn’t provide the recipes/brands for the two foods, making it hard to speculate beyond that guess.

Also, the subjects were eating food they selected and purchased themselves, aside from the pasta. Since as far as one can tell from the BBC article they weren’t asked to record their consumption, it wouldn’t be difficult for this dietary variation to randomly cause one group to score higher on some symptoms.

I’m surprised that Dr. van Tulleken did not report whether the self-diagnosed NGCS sufferers showed greater incidence of gut symptoms than the non-diagnosed. If NGCS is real (and not universal in the population) then a difference between NGCS patients and the rest of the subjects would certainly exist.

It is not stated in the article, but I suspect Dr. van Tulleken’s group does not plan to publish these results in a peer-reviewed journal, which seems to me to be disappointing. I’d love to read in more detail about the protocol and results.

As published, the results seem completely compatible with NGCS being a pareidolic disease. They are also consistent with NGCS being a real syndrome, but with only a small effect (at least in the experimental circumstances). Note that with its weak protocol and small sample size (n=60) it isn’t strong evidence for anything.

I suggest that Dr. van Tulleken’s group or someone else repeat the experiment but with the subjects eating identical diets (entirely supplied by the experimenters, ideally, not just gluten-free with the subjects buying their own food). However, each subject would be required to swallow 10 identical capsules per day. For the “control” period the capsules would contain a non-gluten protein. For the “experimental” period, the capsules would contain gluten proteins. That would make it impossible for the subjects to detect which weeks include gluten from the flavor/texture of their food, and also eliminate the possible effect of other ingredients in the two different pasta recipes.