Expecting gluten versus eating gluten

Most studies of gluten sensitivity ask one question: does gluten cause symptoms? A multicentre trial published in The Lancet Gastroenterology & Hepatology asked two at once — what happens when people eat gluten, and what happens when they only believe they have — and designed the experiment so the answers could not blur together.

Non-coeliac gluten sensitivity is the label for people who do not have coeliac disease or wheat allergy but report gut symptoms after eating gluten. It has no confirmatory test; it is diagnosed by ruling the other two out. That leaves it unusually exposed to a problem every food study faces. If someone is sure a meal will make them unwell, the certainty itself can produce symptoms. That is the nocebo effect, the mirror image of a placebo, and it is not the same thing as imagining symptoms.

The standard answer is a blinded gluten challenge, where neither participant nor researcher knows what is in the food. Blinding hides the gluten, though, without switching off expectation: participants still know they might be getting gluten. This trial went one step further and manipulated the expectation directly.

Four groups, two variables

The study ran at the University of Leeds, Maastricht University and Wageningen University and Research between October 2018 and February 2022. It enrolled adults aged 18 to 70 with self-reported gluten sensitivity, defined as gut symptoms within eight hours of eating gluten, with coeliac disease and wheat allergy excluded. Everyone had been on a gluten-free or gluten-restricted diet for at least a week and was symptom-free or only mildly symptomatic on it.

Eighty-four people were randomised into four groups. Each was told to expect either gluten-containing or gluten-free oat bread, and each separately received either gluten-containing or gluten-free oat bread — two slices at breakfast, two at lunch four hours later. Nobody involved knew which bread was actually served, and participants were not told that expectation was part of the experiment. Symptoms were scored in millimetres on a visual analogue scale every hour for eight hours.

What the numbers say

The group that expected gluten and received it reported the most symptoms. That score was significantly higher than both groups told to expect gluten-free bread: 9.6 mm above the group that secretly received gluten (p = 0.0010), and 9.1 mm above the group that got exactly what it was promised (p = 0.0016).

The more striking comparison is the one that did not reach significance. People who expected gluten but actually ate gluten-free bread scored 11.7 mm, and the 4.9 mm gap between them and the gluten-and-gluten group was not statistically significant (p = 0.28). Meanwhile the two groups told to expect gluten-free bread scored almost identically — 6.9 mm with gluten, 7.4 mm without (p = 1.0). When people did not expect gluten, in this trial, actually eating it made no measurable difference.

The authors conclude that the combination of expectancy and actual gluten intake had the largest effect, reflecting a nocebo effect, and they add in the same sentence that an additional effect of gluten cannot be ruled out. Both halves matter. The trial does not show gluten is irrelevant; it shows expectation is a large enough force that a study ignoring it can easily misattribute symptoms.

The limits worth keeping in view

This was a single test day with around twenty people per group, 86% of them women, with a median age of 27. It measured short-term symptoms over eight hours, not what happens across weeks of regular eating. It used one food, oat bread, and wheat contains other components besides gluten — fermentable carbohydrates among them — that have been proposed as triggers in this group of people. The trial was designed to test gluten, not to settle what else might be involved.

The funding deserves a mention too. The work was part of a Dutch public–private partnership whose partners included bakery ingredient suppliers, millers, food manufacturers and wheat industry associations, and the study breads came from a bakery innovation centre. The authors disclose all of this. It does not invalidate a randomised, blinded result, but it is the kind of context a careful reader should have.

None of this applies to coeliac disease

People with coeliac disease were excluded, and the distinction is not a technicality. In coeliac disease gluten drives an autoimmune attack on the intestinal lining whether or not the person notices anything, which is why trace exposure counts with or without symptoms. How a meal feels is not a reliable measure of what was in it. Celiac disease versus gluten sensitivity sets out the differences in mechanism, testing and threshold side by side.

The general lesson carries across both conditions: a belief about what is in food is not the same as knowing. Checking the ingredient list, rather than inferring from how you feel afterwards, keeps those two things separate.

Source: de Graaf MCG, Lawton CL, Croden F, et al. The effect of expectancy versus actual gluten intake on gastrointestinal and extra-intestinal symptoms in non-coeliac gluten sensitivity: a randomised, double-blind, placebo-controlled, international, multicentre study. Lancet Gastroenterol Hepatol, 2024;9(2):110–123. ClinicalTrials.gov NCT05779358.

Common questions

Keep reading