Intended for healthcare professionals


Coeliac disease and schizophrenia: population based case control study with linkage of Danish national registers

BMJ 2004; 328 doi: (Published 19 February 2004) Cite this as: BMJ 2004;328:438
  1. William Eaton, professor (weaton{at},
  2. Preben Bo Mortensen, professor2,
  3. Esben Agerbo, assistant professor2,
  4. Majella Byrne, assistant professor2,
  5. Ole Mors, associate professor3,
  6. Henrik Ewald, professor3
  1. 1Department of Mental Health, Bloomberg School of Public Health, Johns Hopkins University, 624 North Broadway, Baltimore, MD 21205, USA
  2. 2National Centre for Register-Based Research, Aarhus University, Taasingegade 1, 8000 Aarhus C, Denmark
  3. 3Institute of Basic Psychiatric Research, Psychiatric Hospital in Aarhus, Skovagervej 2, 8240 Risskov, Denmark
  1. Correspondence to: W W Eaton
  • Accepted 15 October 2003


Dohan proposed that an inherited defect interacting with an environmental trigger of gluten precipitated schizophrenia in some individuals, and provided supportive epidemiological evidence.1 Some clinical trials and case studies showed that a cereal free diet improved remission of symptoms of schizophrenia.2 The most important genetic marker found in the study of coeliac disease (6p23-p22.3)3 is very close to the dysbindin locus, which has been implicated in schizophrenia.4

Participants, methods, and results

The case sample comprised 7997 people older than 15 who were admitted to a Danish psychiatric facility for the first time between 1981 and 1998 with a diagnosis of schizophrenia and known maternal identity. For each case we randomly selected 25 controls from a subsample of all available controls, matched by year of birth and sex.

We searched records of the national patients' register for a history of autoimmune diseases in cases, controls, and their parents, in a manner that protected the anonymity of the participants. Denmark has few private health facilities, and treatment is free of charge, so that coverage of visits is nearly 100% complete. Diagnoses were according to the International Classification of Diseases (8th revision, 1981-94; 10th revision, 1995-8). We included coeliac disease (and closely related dermatitis herpetiformis), on the basis of prior scientific literature, and two autoimmune gastrointestinal conditions (ulcerative colitis and Crohn's disease), for which little or no scientific literature exists that implies an association with schizophrenia. We included major risk factors for schizophrenia because these might be confounders of an association with coeliac disease: socioeconomic position, urban residence, and family history of schizophrenia.5 Four patients, five mothers of patients, and three fathers of patients were being treated for coeliac disease before the patient entered a psychiatric facility (1.5 per 1000 population, table). In a conditional logistic regression model the relation of risk factors for schizophrenia replicated that found in the literature.5 The univariate relative risk for schizophrenia, given coeliac disease, was 3.2 (P < 0.0001), unchanged by addition of the covariates (table). The adjusted relative risks for Crohn's disease and ulcerative colitis, when using the covariates discussed above, were both 1.4 (P < 0.08 for Crohn's disease, and P < 0.03 for ulcerative colitis). When coeliac disease and four additional cases of dermatitis herpetiformis were combined in an adjusted model as described above, the relative incidence for either of the two disorders compared with neither disorder was 3.1 (95% confidence interval 1.8 to 5.2)

Relative risk of schizophrenia for people with autoimmune intestinal diseases

View this table:


A history of coeliac disease is a risk factor for schizophrenia, as shown in this epidemiological study. The risk relation is strong but reflects a small proportion of cases of either disorder, since both disorders are rare.

Coeliac disease is presumably underascertained in this study. Only coeliac disease occurring before onset of schizophrenia in the case was considered, and only cases in hospitals or specialty clinics are included. Clinical symptoms of coeliac disease occur in only a fraction (about one in seven) of people with the pathognomic antibody.

Ascertainment bias does not explain the results because it would exist with equal strength for Crohn's disease and ulcerative colitis, and the relative risk for them is weak. Ascertainment bias would not affect the relation of schizophrenia and coeliac disease in parents, which had similar strength to that in cases.

Removal of gluten from the diet is not dangerous or expensive and is an effective treatment for coeliac disease. Failure of replication in earlier clinical trials of gluten withdrawal may have been the result of sampling fluctuation since coeliac disease is rare. New screening tests for coeliac disease are inexpensive and carry minimal risk and discomfort. An important question is the degree to which removal of gluten from the diet will alleviate symptoms in the small proportion of people with schizophrenia who screen positively for coeliac disease but do not show its classical symptoms.


  • Contributors WWE and PBM designed the study. WWE, EA, and MB conducted the analyses of data. WWE drafted the manuscript. All authors contributed to the design of the analysis, interpretation of data and reporting of results. Noel Rose, Peter Zandi, and Alessio Fasano provided helpful advice. WWE had full access to (anonymous) data and assumes responsibility for the decision to submit the paper for publication.

  • Funding Danish National Research Foundation, Stanley Medical Research Institute, and NIMH grant #53188.

  • Competing interests None declared

  • Ethical approval Danish Data Protection Agency, Committee on Human Research of the Bloomberg School of Public Health, and Single Project Assurance Committee in Aarhus, Denmark, established according to regulations from the US National Institutes of Health.


View Abstract