163 flours, and where the gluten actually was

Corn is gluten-free. Rice is gluten-free. Buckwheat, despite the name, is gluten-free. All three statements are true about the grain and none of them is a statement about the bag on the shelf, and a 2025 market study put a number on the distance between the two.

Researchers at Hacettepe University in Ankara bought 163 flour products off the Turkish market and measured what was in them. Not a laboratory simulation of cross-contact, not a survey of what shoppers believed — the actual flour, tested with the R5 ELISA method that the EU recognises for gluten quantification. The results were published in Frontiers in Nutrition in December 2025.

The sample was built to answer one specific question: does the certification do anything? So it split three ways. Fifty-four flours were commercially produced and labelled gluten-free. Fifty-six were naturally gluten-free grains with no gluten-free claim, sold packaged. Fifty-three were the same naturally gluten-free grains sold loose, scooped from open bins. Four grain types ran through all three groups: oat, buckwheat, corn and rice.

The gradient

Gluten was detected in 16.67% of the commercially produced gluten-free flours. In the packaged natural flours, 50%. In the unpackaged ones, 84.91%. The gap between the certified group and the natural groups was significant at p < 0.001.

Read from one end, that is a strong case for certification: a process built to control gluten controlled gluten, roughly five times better than no process at all. Read from the other, the authors are explicit that it is not a guarantee. They report that 16.67% of the commercially produced gluten-free flours exceeded the 20 mg/kg threshold, and conclude in as many words that gluten-free labelling alone does not always ensure safety. Both halves of that sentence are load-bearing.

Oat and buckwheat, again

Across the whole sample, oat and buckwheat flours carried significantly higher contamination levels than corn and rice, at p < 0.001. This is not a surprise to anyone who has followed the oat literature — oats are grown, harvested, transported and milled in rotation with wheat and barley often enough that the crop itself is the exposure route, not the recipe.

Buckwheat sitting alongside it is the more interesting result, because buckwheat is not a cereal at all. It is a pseudocereal, botanically closer to rhubarb than to wheat, and it is the grain whose name most reliably confuses both people and software. Its appearance at the top of the contamination table has nothing to do with its botany and everything to do with where it is milled and how it is handled.

The shop is part of the supply chain

The most actionable part of the paper concerns the loose flours, where the researchers also recorded how the product was stored and served. Two findings stand out. Gluten levels in unpackaged naturally gluten-free flours rose significantly when shared scoops were used to serve them, at p = 0.001. And there was a significant association — also p < 0.001 — between the measured gluten and how the product sat in relation to gluten-containing goods on the shelf.

Flours kept in dedicated sections did measure lower, but that particular difference did not reach statistical significance, and the authors say so rather than rounding it up. It is worth noticing which claims a paper makes firmly and which it hedges; this one hedges in the right places.

The practical implication is that a scoop is a vector. Between a sealed bag of unlabelled rice flour and the same rice flour from an open bin, the study found roughly a 35-point difference in how often gluten turned up at all.

What this changes about reading a label

The mental shortcut this data undermines is the one that says this ingredient is naturally gluten-free, so I do not need to check the pack. On these numbers the shortcut fails half the time for packaged natural flours and five times out of six for loose ones. The ingredient list tells you what was deliberately put in. It says nothing about what arrived by accident, and for these four grains what arrives by accident is the whole question.

What is left is a rough ordering rather than a rule: a tested and certified product, then a sealed packaged one, then anything sold loose. That is a preference ranking, not a safety guarantee, and it applies to a Turkish retail market in a particular year. What a “may contain” warning predicts covers the labelling side of the same problem.

Source: Öcal HY, Özel HG. Naturally gluten-free flours are commonly contaminated, while commercially produced gluten-free flours are relatively safe: a market-based study in Turkey. Frontiers in Nutrition, 2025;12:1707584. Open access.

Common questions

Keep reading