Celiac disease (CD) is an autoimmune disease associated with small intestinal damage and villous atrophy. It occurs in genetically susceptible people of all ages and can involve a variety of symptoms and can also be asymptomatic. Not every genetically susceptible person gets CD however, most likely signifying that environmental factors influence the disease. Infant feeding may very well be one of those environmental factors, research over the past year has shown.

A Swedish study over 20 years in the making with contributions from 15 different doctors and scientists was published online on February 18th of last year in the journal Pediatrics. The article, “Prevalence of Childhood Celiac Disease and Changes in Infant Feeding”, with primary contributors Dr. Anneli Ivarsson and Dr. Anna Myléus, investigated possible causes for the Swedish “epidemic” of CD in children less than two years old that occurred between 1984 and 1996. In that time period, the clinical diagnosis of CD in children less than two years old increased fourfold and then decreased similarly by 1996. Previous studies have documented the association between breastfeeding, gluten introduction, and celiac disease risk; however, whether breast feeding and dietary gluten introduction affect the actual disease occurrence or instead just the disease age onset is not yet known.

As part of a large program known as ETICS (Exploring the Iceberg of Celiacs in Sweden) two groups of 12 year old students were screened for CD. The first group screened between 2005 and 2006, contained children born in 1993 during the epidemic. The second group, screened in 2009-2010, had children born in 1997 after the epidemic had subsided.

The study also involved sending out questionnaires to the parents which concerned how the children were fed as infants. The questions were used to determine how long the child was breastfed and when gluten was introduced to their diet. Out of the thousands of participants, 67% returned the questionnaires fully filled out.

The authors decided on a p-value of 0.05 for the threshold of significance. A p-value greater than 0.05 means that the difference between the two groups was not significant while a p-value less than or equal to 0.05 would be significant.

Out of the 1993 group, the CD prevalence rate was determined to be 2.9% while the 1997 group had a prevalence rate of 2.2%. This difference was found to be significant with a p-value of 0.01. The median age of CD diagnosis was also significantly different between the 1993 and 1997 groups at 1.7 years and 5.5 years respectively (p-value = 0.04).

The authors also found differences between infant feeding among the two groups. The duration of breast feeding among the 1993 and 1997 groups was 7 and 9 months (p-value < 0.001). Dietary gluten introduction was found to be 5 months for both groups; however, the percentage of infants that had breastfeeding after gluten was introduced was 70% and 78% for the 1993 and 1997 groups respectively (p-value < 0.001). The authors also note that participants with CD had comparable infant feeding patterns to those without the disease.

From their findings, the scientists make several conclusions. They confirm the idea that the difference between children born during the epidemic and children born after are based on a change in disease occurrence. They also believe that their data supports the hypothesis that CD can be prevented in at least some genetically susceptible individuals through infant feeding practices. They advocate to gradually introduce gluten to infants starting at 4 months while breastfeeding is ongoing.

There are detractors from the study. Dr. Tom Connell, a pediatrician at Royal Children’s Hospital in Melbourne, Australia, and his colleagues voiced their concern in a letter, published online on April 16th, 2013 by Pediatrics. In that letter, Connell et al noted the lack of information that Ivarsson and Myléus et al provided on the questionnaire, despite how critical the questionnaire was to the study as it established the infant feeding patterns. Connell and his colleagues also suggested that the study’s participants might not be representative of the population, because of those asked to be in the study, only about 2/3rds accepted and only about 2/3rds of those participants also filled out a questionnaire completely.

They also stated that the authors’ conclusion suggesting that gluten be introduced into the diet at 4 months had no support from the study because both 1993 and 1997 groups had the same median age at which gluten was introduced (5 months). Connell admits that the authors did show that there was a significant difference in prevalence of CD between the two groups but he and his colleagues “urge caution…in accepting the conclusion that infant feeding patterns are primarily responsible” due to the limited information given.

Ivarsson and Myléus et al believe that their “findings contribute to the evolving evidence base for infant feeding recommendations” in order to prevent some cases of CD.



The article by Ivarsson and Myléus et al and the response by Connell et al were both posted online by Pediatrics.