What sets off celiac disease? What the evidence says about infections, surgery, pregnancy and stress

Short answer

In the sources we read, none of these four is an established cause of celiac disease: NIDDK says research suggests it occurs only in people who have certain genes and eat gluten, and that other factors are still being studied3. For infections the studies report associations, and the meta-analysis authors say noncausal explanations cannot be excluded2; an umbrella review found that none of the non-genetic associations it graded received a high level of evidence1. For surgery, pregnancy and stress, we found a registry study, case reports and one cohort that found no association4, 5, 6.

Published by Gluten Verdict · updated · 19 sources, each linked below

Reviewed by Mark Peter, Registered dietitian (RD), on

Kitchen counter half packed into moving boxes with a tape gun, kettle and a single mug

Key points

  • NIDDK: about 30 percent of people have the gene variants DQ2 or DQ8, and only about 3 percent of people with them develop celiac disease7; NIDDK also says researchers are studying other factors3.
  • For infections, a 2020 meta-analysis of 19 observational studies found an association (odds ratio 1.37) with high heterogeneity, and its authors write that noncausal explanations cannot be excluded8, 2.
  • A 2024 umbrella review graded the non-genetic associations it covered: none received a high level of evidence1.
  • One Swedish cohort of children found no association between family stress and later celiac disease; its authors say limited statistical power could explain that6.
  • NIDDK notes that many people who have celiac disease have not been diagnosed9, and the authors of an Italian study attribute surgery before diagnosis to doctors' misdiagnosis and/or poor health status11; the Swedish team set out to test whether misdiagnosis delays diagnosis10.

What is established, and what is still unexplained

NIDDK states that research suggests celiac disease only occurs in people who have certain genes and eat food that contains gluten, and that experts are studying other factors that may play a role3. About 30 percent of people have DQ2 or DQ8, but only about 3 percent of those people develop celiac disease7.

That gap is where the "trigger" question comes from. Coeliac UK, the UK charity, puts it plainly: people with the condition have genes that mean they may develop it, but "we don’t yet know why"12. Our guides on gene carriers who never develop celiac disease and on whether eating a lot of gluten can cause it (listed under related guides) cover the genes and the gluten. This page covers the other proposed triggers.

How to weigh a claimed trigger

We sort the sources below by what kind of study they are, because the same sentence ("X is linked to celiac disease") can rest on very different evidence. This ladder is our own way of sorting; it is not taken from one source.

  • A randomized trial changes one exposure and watches the outcome. None of the proposed triggers on this page were studied this way in the sources we read.
  • A prospective cohort records the exposure first and follows people forward. It shows that one thing came before the other, but not that it was the reason.
  • A registry or case-control study compares groups after the fact. It shows that two things occur together more often than chance would, not why.
  • A meta-analysis or umbrella review pools or grades earlier studies. It is only as strong as the studies inside it.
  • An animal or laboratory model can show a possible mechanism, not that it happens in people.
  • A case report describes one person. It shows that something happened, not how often or why.
Where each proposed trigger stands in the sources we read
Proposed triggerKinds of evidence we foundWhat the studies reportWhere it stops
InfectionsMeta-analysis of observational studies; cohorts of children at genetic risk; a mouse modelAny infection associated with celiac disease, odds ratio 1.378; a gastrointestinal infection followed by celiac autoimmunity within 3 months, hazard ratio 1.33, in children at genetic risk13High heterogeneity; the meta-analysis authors say noncausal explanations cannot be excluded8, 2
SurgeryA Swedish registry study, a smaller retrospective Italian study with controls, one case reportMore abdominal surgery recorded before a celiac diagnosis, highest in the first year after surgery4The Swedish team tested whether misdiagnosis delays diagnosis10; the Italian authors attribute pre-diagnosis surgery to misdiagnosis and/or poor health status11
Pregnancy and childbirthCase reportsA previously healthy woman presented with life-threatening celiac disease after childbirth5One person; the authors call a first presentation in pregnancy or after birth uncommon5
StressOne cohort of Swedish childrenNo association between family stress measures and later celiac disease6Limited statistical power; stress in young children’s families, not adult stress14, 6

Infections: what the pooled and cohort studies show

NIDDK says research suggests that a higher number of infections in early life, and certain digestive tract infections, may increase the risk15.

What the pooled studies show

A 2020 meta-analysis pooled 19 observational studies (15 on infection, six on antibiotic exposure). It found that any infection was associated with a higher risk of celiac disease later in life, with an odds ratio of 1.37, and that the studies disagreed a lot (an I2 of 94 percent, which the authors call high heterogeneity)8. The authors also write that noncausal explanations for these associations cannot be excluded2.

A 2024 umbrella review looked at five publications describing 15 meta-analytic associations on seven non-genetic risk factors16. It found that none received a high level of evidence, that infections at any age were the one association rated as suggestive, and that more prospective research is warranted1.

Prospective studies in children at genetic risk

TEDDY followed 6327 children in the United States and Europe who carried celiac-linked HLA risk genotypes, and tested them for tissue transglutaminase autoantibodies, which the paper uses as its definition of celiac disease autoimmunity17. A gastrointestinal infection was associated with a higher risk of autoimmunity in the following 3 months (hazard ratio 1.33)13. The authors conclude that gastrointestinal infections increase that risk in genetically susceptible children, and that the risk is modified by HLA genotype, infant gluten intake, breastfeeding and rotavirus vaccination18. The outcome here is antibodies in young children, not a diagnosis in an adult17.

A smaller TEDDY analysis compared stool virus profiles in 83 pairs of children who did and did not develop autoimmunity19. Frequent enterovirus exposure between ages 1 and 2 was associated with a higher risk, and the authors report an interaction with higher gluten intake20.

An earlier Denver cohort followed children with celiac HLA risk alleles and counted 54 who developed autoimmunity21. Its authors describe the result as the first indication that frequent rotavirus infections may raise the risk22.

The mouse model, and what it adds

A 2017 paper in Science built a viral infection model with two reovirus strains to study how infection affects tolerance to dietary antigens23; NIDDK's summary of it says the scientists first infected mice with two types of reovirus45. Its authors say their study in humans supports a role for reovirus infection in triggering celiac disease24. NIDDK’s summary adds that people with celiac disease had higher levels of antibodies to reoviruses than people without it25. NIDDK describes the finding as a study showing that reovirus infection "may trigger" celiac disease in people who are genetically susceptible26. An antibody difference measured between people with and without the disease is an association, and the mouse work shows a possible mechanism.

Surgery

The largest study we found is a Swedish registry analysis. It identified everyone with biopsy-confirmed celiac disease at Marsh stage 3 in the country’s pathology departments (n=29,096)27. Earlier abdominal surgery was recorded for 14.0 percent of them and 10.9 percent of matched controls (odds ratio 1.36), and the estimate was highest in the first year after surgery (odds ratio 2.00)4.

The researchers had set out to test whether misdiagnosis delays the diagnosis of celiac disease10. They also found that abdominal surgery was more common after the diagnosis (hazard ratio 1.34)28, and their conclusion is that surgical complications associated with celiac disease may best explain these outcomes29. In other words, the study found more surgery both before and after diagnosis, and its abstract does not present surgery as the cause of celiac disease.

An older Italian study compared 244 adults with celiac disease with 232 controls and also found more surgery before diagnosis30. Its authors attribute that to doctors’ misdiagnosis and/or poor health status11.

The case literature includes a 61-year-old woman who developed severe diarrhea soon after bariatric surgery that did not respond to standard treatment; the diagnosis was made from anti-tissue transglutaminase antibodies31, 32. The authors write that, to their knowledge, it is the first reported case of clinical onset of celiac disease following duodenal switch surgery31. A case report cannot show whether the disease was already present before the operation.

Pregnancy and childbirth

The sources we found on pregnancy as a possible trigger are case reports. One 2024 report describes a previously healthy woman whose adult celiac disease had a life-threatening presentation after childbirth, and its authors say it is uncommon for celiac disease to first manifest during pregnancy or the postpartum period5. A first presentation after birth is not the same as a start after birth, and one report cannot tell the two apart.

The overview in a 2019 case report runs in the other direction: it says undiagnosed celiac disease may lead to adverse reproductive outcomes such as intrauterine growth restriction, stillbirth and preterm birth33. Our guide on celiac disease and pregnancy covers that side.

Stress

The one cohort we found that tested stress followed Swedish children from the ABIS study. It measured serious life events in the family, parenting stress and parental worries when the child was 1 and 2.5 years old14. It found no association between psychological stress and later celiac disease in the children, and its authors say they cannot rule out that this is due to limited statistical power6.

That tells you little about stress in adults, because it measured stress in families with young children and followed the children14.

Early-life exposures: antibiotics, feeding, delivery, vaccination

The 2020 meta-analysis found that antibiotic exposure was also associated with new-onset celiac disease, with an odds ratio of 1.234. The 2024 umbrella review rated the antibiotic association as weak35, and reported that the meta-analyses on breastfeeding, time of gluten introduction, rotavirus vaccination and cesarean section did not report statistically significant results36.

In the Denver cohort, frequent rotavirus infections were linked to a possible higher risk of autoimmunity22. A 2026 Norwegian register study of 740 744 children37 did not find an association between rotavirus vaccine and the risk of early-onset celiac disease38.

NIDDK adds two other ideas: that children with a genetic predisposition who eat more gluten in early childhood may have a greater risk39, and that changes in the microbiome may play a role40. See our guides on gluten intake and the gut microbiome.

Why “it started after X” can mean “it was found after X”

NIDDK says many people who have celiac disease have not been diagnosed9, and that most people with celiac disease have one or more symptoms before they are diagnosed and begin treatment41. One possibility is that an illness, an operation or a pregnancy was simply when an existing disease was noticed; that is our reading, not a finding of the studies above.

NIDDK also says celiac disease isn’t diagnosed based on symptoms alone, because some of the symptoms are like those of other digestive disorders42, and that doctors typically diagnose it with blood tests and biopsies of the small intestine43. The American College of Gastroenterology guideline describes antibody detection in serum as very helpful for the initial screening of patients with suspicion of celiac disease, and says intestinal biopsy is required in most patients to confirm the diagnosis44.

For how those tests work, see our guides on the celiac blood test and the diagnosis process.

Questions people ask

Can stress cause celiac disease?

The one cohort we found, of Swedish children, reported no association between family stress measures and later celiac disease, and its authors say limited statistical power could explain that6. NIDDK’s account is that the disease occurs in people who have certain genes and eat gluten, with other factors still being studied3.

Can pregnancy trigger celiac disease?

The sources we found are case reports. Authors of a 2024 report say it is uncommon for celiac disease to first manifest during pregnancy or the postpartum period5.

Can a virus trigger celiac disease?

Observational studies link infections with celiac disease8, and a mouse study and an antibody comparison support a role for reovirus24, 25. The meta-analysis authors say noncausal explanations cannot be excluded2.

Can surgery trigger celiac disease?

A Swedish registry study found more abdominal surgery before a celiac diagnosis4. Its authors were testing whether misdiagnosis delays diagnosis10; the abstract does not present surgery as the cause.

More in Celiac genetics and family risk

Sources