Recognising coeliac disease: why the diagnosis often arrives years too late

It rarely begins in the gut. Usually it begins with a lab value that refuses to move.

You are prescribed iron. You take it. The value barely rises. So you take it for longer. At the next appointment you hear that you should take it more regularly, ideally on an empty stomach, ideally with vitamin C.

Or it begins with fatigue that nobody can explain. With headaches. With a bone density report that does not match your age. With mildly raised liver values.

Many people know this pattern. Every abnormal finding is treated on its own. And nobody asks whether they belong together.

In some of these people they do belong together. That connection could be made visible with two blood values, if someone thinks of them.

This article walks the path from the beginning. It does not start with the symptoms though, it starts with a warning. Because here the order of steps decides everything.

Three different things that all sound like bread

There is one sentence I hear very often in consultations. It goes: bread does not agree with me.

That sentence describes three completely different processes in the body. They feel similar and they are worked up in completely different ways. Which is why this distinction comes before everything else.

Coeliac disease is an autoimmune condition. Certain fragments of the gluten protein, the gliadin epitopes, set an immune response in motion in genetically predisposed people. That response ends up directed not only against the protein from the bread, but also against one of the body’s own enzymes in the intestinal wall, tissue transglutaminase.

Picture a guard chasing a burglar and kicking in their own front door along the way. That is the difference from an intolerance: measurable damage arises in your own tissue. The villi of the small intestine, through which you take up nutrients, flatten out.

Wheat allergy is an allergy in the classical sense. In the fast, IgE mediated form the body reacts within minutes to a few hours, often in the skin and airways. The route there runs through specific IgE and a medically supervised challenge. It usually begins in childhood and is usually outgrown.

Gluten sensitivity without coeliac disease is a diagnosis of exclusion. To this day there is no biomarker for it, and the methodologically clean route runs through a double blind, placebo controlled challenge. More on that in its own article: gluten and gliadin without coeliac disease.

How many there are, and how many do not know it

Roughly one person in a hundred. That number sounds like a footnote until you apply it to a railway station, an office building or a school.

And then comes the second number, the more important one really: how many of these people know it?

The more interesting question is asked by studies that test a whole region and then look at who already knew. Norway has done this twice.

And now to the time it takes. Unrecognised does not mean unnoticed. Often it means noticed, named, treated, only under a different name.

That leaves the mistaken idea that this is a childhood illness. A review led by Pekka Collin puts it at roughly a quarter of all diagnoses being made from age 60 onwards, and around 4 percent from 80. Symptoms at that age are particularly unremarkable.

Why it gets overlooked: the classic picture is the minority

Ask a few people what coeliac disease looks like. Almost all of them describe the same picture: a child with thin arms, a distended belly and diarrhoea.

That picture exists. It is simply no longer the rule. The German S2k guideline records that 62 percent of adults have symptoms outside the gut at the time of diagnosis. And it writes one sentence that carries this entire section: there is no leading symptom.

Iron deficiency that does not respond to tablets

An iron deficiency that does not respond to oral iron is a reason to think about absorption in the small intestine instead of raising the dose further. How an iron deficiency arises and what in the gut it hangs on is covered in separate articles: iron deficiency and absorption and iron deficiency despite iron tablets.

Osteoporosis at an age where it does not belong

The lining of the small intestine takes up calcium. If it is inflamed and flattened for years, less arrives. The German guideline puts it at over 50 percent of untreated people having reduced bone density, depending on how severe the villous flattening is.

Liver values that nobody can explain

Raised transaminases without a recognisable cause are a common incidental finding. A meta-analysis found biopsy confirmed coeliac disease in 5.7 percent of cases with cryptogenically raised liver values, and in 4.6 percent with cryptogenic liver cirrhosis. Important: with cirrhosis of any cause the frequency was only 0.8 percent, so at the level of the general population. For raised liver values of all causes no pooled figure could be formed in the same paper, there it remains a qualitative overview of four studies. So the association holds for the unexplained. The figures come from Yoosuf 2023, source 14 in the list.

The German guideline adds two figures. In roughly half of those affected the transaminases are raised at the time of diagnosis. In 80 percent they are back within the reference range after around one and a half years of a gluten free diet.

Teeth, mouth and head

A meta-analysis of 22 observational studies found a relative risk of 2.49 for enamel defects and 2.45 for recurrent mouth ulcers in coeliac disease. Oral signs overall were present in 21.4 percent of those affected.

These numbers are not a test. A relative risk of 2.45 for something as common as mouth ulcers does not mean that mouth ulcers point to coeliac disease. It means that together with other signs they are a reason to ask the question. Which is why the guideline files them as a should ask trigger.

With the head it is similar. For epilepsy there is a Swedish registry study of 28,885 biopsy confirmed cases, which found a roughly 1.4 fold raised risk. For migraine the guideline also names a raised risk. I could not assign the figure behind it to a verifiable primary paper in this research, which is why none appears here.

Unfulfilled wish for a child and repeated miscarriages

The guidelines name an unfulfilled wish for a child and repeated miscarriages as a reason to think of coeliac disease. The figures on this come from the literature and stand here so that you can place them. Whether and when testing happens in your case is decided by the practice that supports you in this question, as a rule gynaecology or the fertility clinic. No offer and no promise follows from this paragraph.

Mechanistically plausible, not proven in humans: the same paper describes how antibodies against tissue transglutaminase might bind to cells of the developing placenta and slow the formation of new blood vessels. That explanation comes from cell cultures and from a mouse model. It is a hypothesis, not an established process in humans. [In vitro] [In vivo, mouse]

The skin as a route to the diagnosis in its own right

Dermatitis herpetiformis is the skin form of coeliac disease: intensely itching blisters and papules on elbows, knees and buttocks. Open bowel symptoms are rare with it. A Finnish review puts the ratio of the skin form to coeliac disease at 1 to 8, with a mean age at diagnosis of 40 to 50 years.

One practical detail often goes wrong: to confirm it, the tissue sample is taken from the skin next to the group of blisters, not from the blister itself. What is looked for are granular IgA deposits, and that finding counts as conclusive. The route runs through a dermatology practice.

The rule from the very top applies here as well. The IgA deposits in the skin and the antibodies in the blood reflect the reaction to gluten. Anyone who changes their diet beforehand can block this route to the diagnosis too. First the work up, then the food.

The paragraph that does the most work

The German S2k guideline records this: at the time of diagnosis, 28 percent of those affected are overweight and 11 percent obese.

I write it that briefly because this one sentence clears away more false ideas than any symptom list. Coeliac disease is not a diagnosis you can see on a person. It has no appearance. Being slim is not why someone has it. Being sturdy is not why someone does not.

The same goes for age, for sex and for the question of whether someone has always tolerated bread. The condition can show up in any decade of life, even after fifty years of eating bread without trouble.

And finally the most common mix up

The irritable bowel diagnosis is a common stop along the way. A meta-analysis of 29 studies with 7,209 people with irritable bowel syndrome found biopsy confirmed coeliac disease in 2 percent and an odds ratio of 4.42 for abnormal antibodies. How many coeliac diagnoses had previously been filed as irritable bowel is not something this paper says. Its authors do conclude, though, that an irritable bowel diagnosis should not be made without ruling out coeliac disease, more on that in its own article: irritable bowel and the search for causes.

Where to search actively, even without symptoms

There are people for whom the test is worth it although nothing is wrong with them. That sounds like an exaggeration. But it follows directly from the numbers in the last section.

If a relevant share of those affected have no symptoms, then waiting for symptoms cannot be the route. You have to search where the probability is higher.

The conditions that often turn up together

With type 1 diabetes the link is best documented: 4.5 percent in a meta-analysis with 293,889 affected people, 5.1 percent in the summary of the German guideline. Two sources, the same order of magnitude.

For autoimmune thyroiditis the German guideline names 4 to 10 percent, and it recommends a TSH measurement at first coeliac diagnosis anyway. What role food can play in Hashimoto thyroiditis is covered in its own article: nutrition in Hashimoto.

To these the German guideline adds autoimmune hepatitis, trisomy 21, Turner syndrome, Williams Beuren syndrome and selective IgA deficiency. For first degree relatives and for trisomy 21 the guideline makes a strong recommendation, for second degree relatives a weak one.

What makes this list remarkable: it does not come from functional medicine, it comes from the guideline of the gastroenterological professional society itself. It names chronic exhaustion, a drop in performance, migraine, depression, an unfulfilled wish for a child, eating disorders and fractures without a matching accident.

The diagnostic work in the right order

Here it gets concrete. And here the order decides the result, not the number of tests.

What follows is the route the German S2k guideline describes. It is here so you can read your own report and ask questions at your next appointment. It does not replace a medical decision.

What a positive antibody test means, and what it does not

An abnormal value is a probability, not a diagnosis. How high that probability is depends on who was tested.

For comparison: in two British cohorts with well founded suspicion, the predictive value for a Marsh 3 finding with very high transglutaminase IgA was 98.7 and 100 percent. That is not a contradiction of the 88.1 percent from Norway, it is a question of pre-test probability.

What is written on your biopsy report: the Marsh classification

Pathology assesses three things. How many immune cells sit between the intestinal cells, how deep the crypts are and how tall the villi stand.

Picture the lining of the small intestine as terry towelling. The villi are the loops, and through them you take up iron, calcium, folate and vitamin B12. The crypts are the supply chambers in between. In active coeliac disease the loops flatten out. Over the years a terry towel turns into a linen cloth.

The practically most important point: Marsh 1 and Marsh 2 are not proof. Increased immune cells without villous damage are also found with infections, with medication and with other inflammatory conditions. Which is why the serology stands alongside. And which is why Marsh 1 on its own cannot yet be a reason to change your diet. If the antibodies are clearly positive at the same time, an early form can still be behind it. What follows from that in your case is decided by the practice that knows your report, not by this text.

HLA-DQ2 and DQ8: a tool for ruling out, not proof

Many people understand the HLA test as a genetic test for coeliac disease. It is something else, and the difference matters.

HLA-DQ2 and HLA-DQ8 are tissue markers that are practically necessary for this immune reaction. Without them coeliac disease is very unlikely. The German guideline words it like this: if these markers cannot be detected, coeliac disease can be ruled out with high certainty. Detecting them, by contrast, does not confirm the diagnosis.

The reason lies in how widespread they are: 20 to 40 percent of the general population carry these markers, and the great majority never develop coeliac disease. A positive HLA finding mainly says that the door is open. Not that anyone has walked through it.

The test is practically useful in two places. First, when someone has eaten gluten free for years and a gluten challenge is out of the question: then a negative finding can close the question. Second, in risk groups: roughly a quarter of first degree relatives and almost two thirds of people with trisomy 21 do not carry these markers.

A note on cost that is missing from most guide articles: the German guideline points out that services for risk estimation are not covered by the German statutory health insurers under the genetic diagnostics act.

The separate rule for children

For children a route of their own has applied since 2020, and parents often run into contradictory statements online. So here, briefly and precisely, is what applies.

The European paediatric society ESPGHAN updated its guideline in 2020. It first states the same thing as the adult guidelines: the combination of total IgA and transglutaminase IgA is superior to any other.

Then comes the difference. In children the diagnosis may be made without a tissue sample when three conditions come together:

Two things that used to belong here are explicitly no longer mandatory criteria: HLA typing and the presence of symptoms. A child without symptoms and with a very high value can therefore take this route just as well.

If the value is below ten times the reference limit, the endoscopy stands. At least four tissue samples from the distal duodenum and at least one from the bulb should then be taken. If the sample shows only Marsh 0 or Marsh 1 while the antibodies are unambiguously positive, close follow up is indicated.

One trigger that counts particularly in children is short stature. A meta-analysis of 17 studies with 3,759 children found biopsy confirmed coeliac disease in 7.4 percent with short stature of all causes, and in 11.6 percent with idiopathic short stature. The authors themselves name a considerable selection bias and a wide spread. The number justifies a question, it is not a fixed figure.

The mistake that can make the diagnosis impossible

Now comes the section this article opened with a warning for.

The sequence is almost always the same. Someone has felt unwell for months. Online it says gluten could be to blame. So gluten is dropped, for two weeks, for two months. Things get better. And then, at some point later, comes the question: was that coeliac disease?

From that point on this question can no longer be answered easily. And that is not a formality, it is physiology.

All four guidelines say this. The European version writes word for word that serology and biopsy should be carried out on a gluten containing diet. The German S2k guideline turns this into a strong recommendation with strong consensus and even reverses it: if a gluten free or wheat free diet is started for reasons other than a confirmed coeliac disease, coeliac disease should be ruled out serologically beforehand, particularly in people with symptoms.

What happens in the gut when gluten comes back

The most interesting study on this turned the question around. It did not ask how fast a lining recovers, but how fast it takes damage again under gluten.

Out of this finding comes the apparent contradiction that many people stumble over online. If 14 days are enough, why does the German guideline ask for three months?

Because the two numbers do not answer the same question. Leffler wanted to know how fast damage arises, and concludes that two weeks can already be enough in many adults to meet the criteria. The guideline wants to rule out something else, namely a false negative finding in those for whom it is not enough. Which is why it sets the challenge longer. That is caution, not a contradiction.

What a gluten challenge looks like under the German guideline

These figures come from the S2k guideline and are therefore literature, not a personal recommendation. They are here so you know what is coming if this route becomes necessary.

If the gluten free diet has been running for years already and a challenge is out of the question, the HLA test remains as a way out. But only in one direction. If the markers cannot be detected, the question is practically answered. If they are there, it is as open as before.

And now the reason this is worth it, even when the challenge is demanding. Of 200 people with damage to the lining and no matching antibodies, only 31 percent had coeliac disease. In the rest there was usually an infection, another inflammatory condition or a medication behind it.

Anyone who goes gluten free without a work up can therefore miss something else. And that something else is often treatable.

One point belongs here explicitly, because medication appears in that list. The best known example is the active substance olmesartan, a prescription only angiotensin 2 receptor blocker for high blood pressure, which in rare cases can trigger villous flattening with diarrhoea and weight loss, sometimes only months to years after the start of treatment. Painkillers from the NSAID group and the immunosuppressant mycophenolate are discussed alongside it.

That stands here as context and not as a suspicion against your medication, and no dosage belongs with it. If a medication turns out to be the cause, only the prescribing practice decides about a change or a withdrawal. No guide article and no research online replaces that, and ongoing medication is never altered on your own. Stopping abruptly carries risks of its own, with a blood pressure drug for instance an uncontrolled rise in blood pressure.

What comes after the diagnosis, and why it is good news

There is one objection I understand very well. It goes: I already feel better without gluten. So why all the diagnostic work?

The answer has three parts. It has to do with what you do not get without a diagnosis.

First: what a confirmed diagnosis brings with it in follow up

Coeliac disease does not only affect the gut. Which is why a programme hangs on the diagnosis that nobody sets in motion for you without it.

None of this happens if you simply eat gluten free and leave the question open. That is the practical difference between a dietary decision and a diagnosis.

Second: what the diagnosis brings in numbers

The strongest finding on this comes from a meta-analysis on pregnancy outcomes. It evaluated the diagnosed and the undiagnosed cases separately.

Bone shows the same pattern. The Danish registry work found the raised odds ratio for osteoporosis even before the diagnosis. The German guideline, on the other hand, records that osteopenia can improve on a gluten free diet after as little as one year.

And the Swedish survey shows it for quality of life. The measured value rose from 0.66 in the year before treatment to 0.86 afterwards. The comparison value for the general population was 0.79.

On the subgroup tile: that statement comes from a subgroup analysis within a meta-analysis and refers to growth restriction, stillbirth, preterm birth and birth weight. It can justify a hypothesis, not a proof.

Third: the honest side

It would be dishonest to stop here. Because a gluten free diet is not the end of symptoms for everyone.

The German guideline puts it at up to 30 percent not responding sufficiently to the gluten free diet. The most common cause is not a rare special form, it is gluten taken in without noticing.

The Italian cohort shows the second most common one. Of 79 people with bowel symptoms, 20 did not see their symptoms settle completely on a gluten free diet, and among 17 who kept strictly to the diet there were two cases of lactose intolerance and 15 of overlap with irritable bowel. For that, a time limited FODMAP approach can be a next step, only after the confirmed diagnosis.

And there is refractory coeliac disease, in which symptoms and damage to the lining persist despite a strict gluten free diet. The American and the European guideline describe it as a rare cause of non response with an often unfavourable prognosis. I did not find a robust frequency figure in this research, which is why none appears here.

What contributes nothing to this diagnosis

Finally the boundary, because a great deal is offered around this topic.

A stool test on the microbiome does not diagnose coeliac disease. Nor does a zonulin value: it describes a concept of barrier function and is not a coeliac marker. And IgG antibodies against foods from an intolerance test have nothing to do with transglutaminase serology, even though both contain the letters IgG.

If you are looking for the overall context this diagnostic work belongs in, you will find it in the overview article of the cluster: the gut reset.

Common questions about coeliac disease and its diagnosis

Where this topic connects to the rest of the body

Coeliac disease rarely stands on its own. It hangs on autoimmunity as a whole and on the uptake of nutrients in the small intestine.