To our knowledge, this is the first study to evaluate individual provider variation with respect to celiac testing in children with chronic abdominal pain. Though intra-center practice variation amongst providers did not occur, we found that less than half of these children underwent celiac serologic testing based upon the practice of seventeen pediatric gastroenterology providers at one tertiary care center. When combining both serologic testing and EGD, approximately half of the children were tested. This lack of uniform testing amongst the providers for the children evaluated suggests these providers are selective in their approach; therefore, recommendations regarding uniform celiac testing  in this population are not currently being followed.
Interestingly, we found that age, gender, body mass index, and gastrointestinal symptoms at the time of the initial consultation did not predict whether celiac serologic testing was completed. However, being Caucasian did increase the likelihood of serologic testing. We hypothesize that this may be due to the fact that providers recognize that certain high risk human leukocyte antigen alleles are found more commonly in the Caucasian population, with studies suggesting minority populations such as African Americans comprise only a small percentage of the celiac disease population in the United States [7, 8]. Nevertheless, future investigation into whether race and/or ethnicity clearly impact the yield of celiac testing in children with chronic abdominal pain is needed to help guide providers caring for these patients.
There was a trend toward increased frequency of serologic testing (Table 2) in those children with diarrhea, bloating, and flatus. Interestingly, flatus was found to occur significantly more frequently in those with celiac disease. These symptoms are known to occur in celiac disease , and we note that given this trend toward testing in children with these symptoms there may be a higher frequency of celiac disease identified than that seen in a general population of children with functional abdominal pain. Further studies are needed to delineate associated symptoms which may increase the yield of celiac serologic testing in this population.
Our center’s findings with respect to the overall intracenter frequency of celiac testing in children with functional abdominal pain is supported by a previous study identifying that 57% of 122 children underwent celiac serologic testing when being evaluated at a tertiary care medical center for abdominal pain . However, this study by Dhroove, et al. was limited in that children with chronic diarrhea, weight loss, and abdominal pain in particular locations were excluded. Though abdominal pain in certain locations and weight loss are considered “red flags” when evaluating a child with chronic abdominal pain , children and adults with these symptoms may still have a functional gastrointestinal disorder, particularly given a negative evaluation for an organic etiology [11–14]. As such, children with symptoms associated with celiac disease and functional gastrointestinal disorders were excluded in the Dhroove study. Potentially in concordance with these limitations, none of the studied children were subsequently found to have celiac disease, leading to an estimated prevalence of celiac disease in their population of chronic abdominal pain below that of the general population . Nonetheless, the lack of universal celiac serologic testing supports our findings.
The prevalence of celiac disease in those tested in our study was higher than that found in a previous pediatric case–control study using celiac serology alone in the primary care setting . The prevalence in this previous study in both those with chronic abdominal pain and controls was 1% (95% CI: 0-7%) similar to the estimated prevalence of celiac disease of 1% in the United States and European general population . The difference in prevalence between our study and that of Fitzpatrick’s et al. may be due in part to the usage of EGD biopsy findings rather than serologic testing alone to make the diagnosis, and the exclusion of children with other organic contributors in our final study population (hence providing for a more selective group). This is supported by a recent systematic review and meta-analysis that found the pooled prevalence of celiac in adults with IBS in studies based upon biopsy-proved celiac disease to be 4.1% (95% CI:1.9-7.0) versus 1.63% (95% CI: 0.7-3.0) in those studies using serologic testing alone . Moreover, our findings are further supported by a large epidemiologic screening study using primarily endomysial antibody testing with intermittent EGD confirmation which identified a 4.0% (95% CI: 2.99-5.20) prevalence of celiac disease in 1326 tested symptomatic (abdominal pain, constipation, diarrhea) children .
The primary limitation of this study is its retrospective nature. As such, initial medical evaluation, serologic testing laboratories, and endoscopies (e.g. number of duodenal biopsies taken) were not standardized. Another limitation is that children were not classified into a specific Rome III functional gastrointestinal disorder (e.g. irritable bowel syndrome vs. functional dyspepsia), though recent studies suggest this may be difficult to do even with prospective standardized criteria . Finally, this is a single-center study; therefore, a future multicenter study is likely needed to confirm its results with respect to inter-provider variation.
One of the strengths of our study is its basis upon data generated in routine clinical practice at a tertiary medical center amongst numerous providers, which more likely reflects the actual experience in this setting and may help with generalizability of the findings. In evaluating the prevalence of celiac disease in those tested, another strength of this study is the use of endoscopic biopsies as part of the diagnostic criteria for celiac disease, as duodenal biopsy is the gold-standard for diagnosis . All children with positive serologic testing underwent subsequent EGD; therefore, serologic false positives were not part of the prevalence determination. Serologic false negatives influencing the prevalence determination were decreased in our study as some of the children underwent EGD despite having negative celiac serology. In fact, one child in our cohort was identified with celiac disease despite having negative serologic testing.