Data on serologic markers, upper endoscopy, and biopsy with histopathological information were also gathered. We categorized children into three age groups: infant-preschool group, 0–5 years; school-aged group, 6–11 years; and adolescence group, 12–19 years. At our institution, we used cutoff values of 20U/ml for tissue transglutaminase antibody positivity and any positive titers for endomysial antibody positivity. According to the recent ESPGHAN CD guideline,⁴ most commercially available kits for tissue transglutaminase antibody applied the cutoff values between 3 and 20U/ml; we believe that applying 20U/ml cutoff would limit the false positive results. Small intestinal biopsy suggestive of Marsh III CD was based on the Marsh–Oberhuber classification^{9, 10} with a reported degree of villous atrophy: partial, IIIa; subtotal, IIIb; total, IIIc. The medical records were independently reviewed by two physicians. The study was approved by the institutional review board at our institution.

Overall, most children had subjective abdominal complaints or bowel movement changes. More than 40% of the children had weight issues (weight loss or poor weight gain). We found that 12% had neither gastrointestinal (GI) symptoms nor weight issues. The proportions of children with atypical extraintestinal presentations were: fatigue or malaise, 12% short stature or stunted height, 5% iron deficiency (with or without anemia), 3% and elevated liver enzymes, 2%.

Regarding the past medical history, CD children were also found to have type 1 diabetes in 6%, food allergy in 4%, hypothyroidism in 3%, and Down syndrome in 2%. Immunoglobulin A deficiency (with serum immunoglobulin A level of <7mg/dl) was found in 10 children, but all underwent intestinal biopsy that confirmed Marsh III lesion in the duodenum. Proportions of reported positive family history (first- and second-degree relatives) of CD, irritable bowel syndrome, and gastroesophageal reflux are shown in Table 1.

We found that 6% of the children and adolescents avoided dietary gluten before the CD diagnosis. Furthermore, 15% of biopsy-confirmed CD patients had gluten transgression within the first year of diagnosis.

Some previous data exist on the different age-related patterns of CD presentations in children vs. adults. Vivas et al.¹² prospectively examined 66 CD children with the mean age of 3.6 years with two-thirds <2 years of age, and most young children presented with more pronounced symptoms and severe villous atrophy. However, the current evidence showed that pediatric CD are diagnosed at a later age with less apparent clinical manifestations.⁶ To our knowledge, this is the first study to directly examine both the clinical presentations and gluten-related issues among pediatric CD patients of different age groups.

Among 411 biopsy-proven CD children and adolescents with positive CD serology, various age-related characteristics were detected; the previous study had suggested that interobserver agreement is substantial when combining positive serology and positive Marsh III biopsy for the CD diagnosis.¹³ Overall, higher male percentage was found in the infant-preschool age group. A simple explanation for this gender difference is elusive but may relate to dissimilar hormonal and immune responses to injury and inflammation in the presence of gluten introduction during early childhood; related gender differences have been observed in other immune system-mediated disorders.^{14, 15}

Regarding the clinical manifestations, the school-aged and adolescence groups more frequently presented with nonspecific abdominal complaints. Moreover, the absence of GI symptoms was also more commonly noticed in the older children. However, on combining the absence of GI symptoms and absence of weight issues (i.e., silent CD), the statistical significance on the age-related patterns disappeared (Table 1), which differs from earlier studies.^{6, 11} Furthermore, no significant differences were found in atypical extraintestinal CD presentations. The proportions of children with iron deficiency and elevated liver enzymes were comparable to the prior study.¹² The infant-preschool group not only had more prominent clinical presentations, similar to prior studies,^{6, 12} but also were more likely to be diagnosed with biopsy-proven CD before the referral to our institution. Possible explanations to the latter finding include more pronounced manifestations that led to an earlier CD diagnosis and tendency to refer patients with more severe presentations to a tertiary care center. Regarding the bowel movement issues, we found that a relatively high percentage (20%) of CD children presented with constipation, compared with prior studies.^{16, 17} The infant-preschool group not only had more children with diarrhea (without constipation) but also higher proportions of constipation and alterations of both when compared with the older children (Table 1), although the previous review suggested that constipation is more common in older children and adolescents with CD.¹¹ These disparities might possibly be because of various reporting and/or referral issues by the caregivers and health-care providers, a possibility that requires further investigation. Generally, the younger children could likely have limited capability to express subjective symptoms when compared with the older individuals. Natural courses, environmental factors (e.g., breastfeeding and gluten introduction), and disease pathogenesis could be different between younger and older children.

We found a borderline significant higher proportion of reported food allergy in the youngest group. Both CD and food allergy are prevalent and increasingly diagnosed in children.^{6, 18} These two conditions can both present with nonspecific GI symptoms such as vomiting and/or diarrhea in infants and young children.¹⁹ Therefore, health-care providers might initially diagnose these children with presumed food allergy before completing CD evaluation in suspected cases. The significant age-related differences in family history of CD were interesting but difficult to explain. School-aged children and adolescents, who were more likely to have no or less typical CD symptoms, might also be more likely to undergo CD screening solely because of positive family members with CD compared with the younger children. The similar reason might explain the borderline significant higher proportion of family history of irritable bowel syndrome among the school-aged children as well. Another possible explanation might be because of differences in parental or patient reports across the age groups. These questions will need to be addressed with future research.

Notably, we found that dietary gluten avoidance before the formal CD diagnosis is common in children (6%), with the higher proportion among school-aged children up to 10%, although the current CD prevalence is ~1% in the general pediatric population.² As the school-aged group more commonly presented with nonspecific abdominal complaints, caregivers or health-care providers might implement dietary gluten restriction as a therapeutic trial to alleviate these symptoms. Recent evidence suggested that nonceliac gluten intolerance exists in adults.²⁰ Contrary, the adolescents were less likely to avoid gluten before CD diagnosis, which might be because of increased identity, independence, and distinctive peer interaction during this stage of development. The similar grounds could also explain the high rates (27%) of dietary gluten transgression within the first year of CD diagnosis in this age group. An investigation on decisions to implement and continue dietary gluten avoidance among patients, caregivers, and health-care professionals would be valuable to explain these findings.

At the initial CD diagnosis, the more prominent histological features (total villous atrophy (Marsh IIIc) and increased antral intraepithelial lymphocytes) at diagnosis were found among the infant-preschool group, similar to the prior reports.^{11, 12} One study showed that children with increased gastric intraepithelial lymphocytes are more likely to be diagnosed at an earlier age, and present with more pronounced laboratory and duodenal mucosal abnormalities.²¹ The recent study also showed decreased proportions of CD children diagnosed in 2000–2006 with total villous atrophy when compared with individuals who were diagnosed during the previous decade.⁶ On the other hand, older children were more likely to have grossly visible and chronic changes in the duodenum, which might be because of diverse immune responses and histopathological changes between age groups (i.e., timing of CD diagnosis).

This study has some potential limitations. The study was conducted in a retrospective fashion at a large teaching hospital. Although we acknowledge the possibility of incomplete information (e.g., anthropometric data), both recall and information biases as well as the inability to control for all of the parent- or patient-reported data with the physicians' interpretation, we believe that our novel findings will provide a platform for the development of age-related approaches for evaluation and management of CD. Prior evidence suggests that delay in CD diagnosis is correlated with the patient's age;¹² in other words, there will tend to be more delay in older children (which might be subject to more recall biases), as compared with younger children. With regard to parental (proxy) reporting, the limited communication abilities of infants and younger children have inevitably restricted expression of subjective symptoms in this group. Therefore, both researchers (and clinicians) need to rely on the parental reports for the clinical presentations of younger children. We are also well aware that cutoff ages for “pediatric” and “adolescent” population can be varied between different geographic areas. We decided to employ 19 years of age as a cutoff year based on social environment and culture on development (according to the Erikson's stage of development) that would mainly be applied to the aforementioned gluten-related issues.

It is noteworthy that all study patients were evaluated before release of the recent 2012 CD guideline,⁴ and we still believe that it was most appropriate to apply the then-current NASPGHAN 2005 guideline, especially in these North American children and adolescents.⁸ We also note that the issue of nonbiopsy diagnosis in CD remains debatable.²² Furthermore, all patients had documentation of both positive CD-specific serology and biopsy, as we believe, in this situation, that genetic markers would not play an important role in assisting CD diagnosis.^{4, 8} The data on gluten-related issues were obtained in a qualitative way (yes/no) from the medical records. As a generally accepted strategy, patients who undergo CD evaluations should consume regular gluten-containing diet while obtaining serology and biopsy to increase the yield of diagnosis and limit potential false negative results.⁸ Gluten avoidance can also result in several disadvantageous effects on the quality of life,²³ self-perceived health,²⁴ psychosocial dynamics,²⁵ nutritional status,²⁶ and increased economic health burden.²⁷ Moreover, the proportions of dietary gluten transgression in our study were comparable to most previous studies.^{28, 29, 30} Currently, strict lifelong gluten-free diet remains the only effective management in CD. As a result, the history of gluten restriction before making a definite diagnosis and any degree of dietary transgression after the diagnosis would be unsuitable in the processes of proper evaluation and management of CD, respectively.

Guarantor of the article: Carlos A. Camargo Jr, MD, DrPH.

Financial support: None.

Potential competing interests: None.