Exploring Celiac Disease Outcomes in the Community: Melting the Celiac Iceberg?: 2010 ACG Auxiliary Award

Abstract

Purpose: Clinically diagnosed celiac disease (CD) patients carry an increased risk of complications and mortality. CD is markedly under recognized. We evaluated the prevalence as well as morbidity and mortality of undiagnosed CD as compared to incident cases of clinically diagnosed CD in Olmsted County, Minnesota Methods: Blood samples from 35,299 Olmsted County adults collected between 1995 and 2006 were tested for immunoglobulin A tissue transglutaminase antibodies (tTGA) in 2006 to 2008. A tTGA test <2.0 U/ml was considered negative. Subjects with tTGA test >2.0 U/ml were further analyzed for endomysial antibodies. Undiagnosed CD was established if tTGA test >2.0 U/ml and subsequent EMA positive. For comparison, we used an incident cohort of adults with clinically diagnosed CD in Olmsted County from 1995 to 2008 identified via the Rochester Epidemiology Project. Medical records were reviewed for laboratory measurements, comorbid conditions, and mortality. Logistic regression was used to test for an association between CD group and clinical characteristics, adjusted for age, gender, and calendar year of CD assessment. Results: We identified 327 (0.93%) subjects with undiagnosed CD. Ten subjects with undiagnosed CD and 8 patients with clinically diagnosed CD were excluded from further analysis due to lack of research authorization of their medical records. Twenty-one subjects with undiagnosed CD had a subsequent clinical diagnosis of CD and were included in the clinically diagnosed group. Thus, a total of 296 undiagnosed CD cases (56% female, median age 45.5 years) and 150 clinically diagnosed CD cases (67% female, median age 44.9 years) were compared. Subjects with clinically diagnosed CD had significantly lower body weight (median 71.3 vs. 78.0, p=.001), body mass index (24.6 vs. 26.9, p<.001), cholesterol (176 vs. 186.5, p=.001), ferritin (18 vs. 30, p=.001), and hemoglobin (12.9 vs. 13.9, p<.001). Lumbar spine bone density (T-score) was lower among undiagnosed CD subjects (-1.6 vs. -0.9, p=.01). The number of deaths in the 5 years following CD assessment was low in both the clinically diagnosed CD and undiagnosed CD group (5-year Kaplan-Meier survival rates 98% and 94%, respectively). Conclusion: The prevalence of undiagnosed CD was 0.93%, suggesting that a substantial number with CD remain unrecognized in the community. Relative to undiagnosed CD, clinically diagnosed CD was significantly associated with some potential markers of morbidity. These findings suggest that screening could detect undiagnosed CD in the community before some morbidity may occur.