Usefulness of Symptoms as a Diagnostic Tool for Celiac Disease


Celiac disease (CD) is the most common genetically-induced food intolerance and is thought to be present in about 1% of the world’s population. Many cases of CD go undiagnosed until adulthood but mass screening for CD for the entire population is controversial. Instead of serological (blood serum) screening for CD markers as a first step, a questionnaire could be an alternative to identifying those at risk for CD.

A recent study in Sweden by Rosen et al. examined strategies for screening children for CD. Published online by Pediatrics on January 13, the study had 12-year-old children and their parents complete questionnaires about symptoms of CD and family history. Prior to taking the questionnaire, the children were screened for CD using serological markers by a separate program known as ETICS (Exploring The Iceberg of Celiacs in Sweden) but did not know the results of the test until after.

The children were selected from the sixth grade across several regions from a population of 10,041 students. Of those, 6,905 non-CD children and 149 screening-detected CD children completed questionnaires before finding out about the results from the CD screening. Parents of the children were given a separate questionnaire that asked about family history and CD related illnesses such as thyroid disease and Type I diabetes. Most parents answered and returned these questionnaires. These two groups were compared to determine the effectiveness of using questionnaires about symptoms, family history, and/or associated conditions in finding children with undiagnosed CD.


Three hypothetical protocols were tested. First, if a child reported one or more of the questionnaire-listed symptoms which included tiredness, poor appetite, nausea, stomach ache, upset stomach, abdominal gas, bloating, hard stools, loose stools, and lactose intolerance. If this were used as a protocol, 2,333 children (34%) would have been screened. The sensitivity of this testing protocol was 34% (50 out 149 CD children being further tested) and a specificity of 66% (4,572 out of 6,905 non-CD children not being further tested).

The second hypothetical protocol would be testing those children with CD-associated conditions, including family history of CD, determined by the questionnaires sent to the parents. This questionnaire testing protocol had a sensitivity of 10% and a specificity of 94%.

The third hypothetical test would be to combine the previous methods and test any child who reported one of the listed symptoms and/or had a CD-related condition. The sensitivity of that test would have been 38% and the specificity 63%.

The authors also looked at the prevalence of symptoms and CD-associated conditions in the CD children and non-CD children. They found that for both groups, the rates of listed symptoms and the associated conditions did not differ significantly.


The results of the study show that a questionnaire asking about symptoms of CD, family history of CD, and/or CD-related conditions is a poor diagnostic tool for identifying those at risk for CD. Just as many children not suffering from undiagnosed CD had CD-associated symptoms as those who had the disease, and none of the three hypothetical tests using the questionnaires was an effective means to identify children with undiagnosed CD.

Susan Baker, MD, PhD, a professor of Pediatrics at the University of Buffalo wrote a commentary in response to this study by Rosen et al. Published in Pediatrics as well, she says that “the large and carefully designed study by Rosen et al in this issue calls into question the validity of using symptoms to identify children for screening.” Current guidelines for children rely on family history and symptoms to initiate screening for CD. Ultimately Dr. Baker suggests that with the results of this study the guidelines for recommending screening for CD should be revisited, as well as the guidelines for prescribing a gluten-free diet before CD has been diagnosed.

The authors admit that there are some limitations to this study. The questionnaire was designed specifically just for this study so an issue with the questionnaire could affect the results. The authors believe that the questionnaire was clear in its questions but because it was designed specifically for this study, comparing it to future studies on questionnaires could be difficult. The ETICS CD screening could also affect the results of the study. For ethical and practical reasons, only the children with significantly elevated CD markers were referred for a small intestinal biopsy to confirm CD, so it is possible that some children in the non-CD group may actually have CD. However due to the large study size, the authors don’t believe that this had a significant effect on the results.

Pediatrics currently has the article by Rosen et al under its Early Releases tab as well as the response by Susan Baker.