Observer variability in the histologic diagnosis of microscopic colitis

Abstract

Microscopic colitis is diagnosed based on histologic criteria. There has been no investigation of the reproducibility of the histologic diagnosis of microscopic colitis. Our aim was to evaluate interobserver and intraobserver variation in this diagnosis. Colonic biopsies from 90 subjects (20 lymphocytic colitis, 20 collagenous colitis, 20 inflammatory bowel disease, and 30 normal) were blindly and independently reviewed by 4 gastrointestinal pathologists. The biopsies were classified by each pathologist into 1 of 6 diagnostic categories: lymphocytic colitis, collagenous colitis, active chronic colitis, focal active colitis, normal, or other. The slides were then relabeled and blindly reinterpreted 3 months later. The degree of agreement was determined using kappa statistics (lambda). Interobserver agreement with the 6 diagnostic categories was 69% (kappa = 0.76, 95% CI 0.69, 0.83) and 70% (kappa = 0.71, 95% CI 0.61, 0.79) for the first and second observations, respectively. Interobserver agreement with final diagnostic categories of microscopic colitis versus nonmicroscopic colitis was 91% (kappa = 0.90, 95% CI 0.82, 0.96) and 88% (kappa = 0.83, 95% CI 0.73, 0.92), respectively. Mean intraobserver agreement with the 6 diagnostic categories was 83% (kappa = 0.77). Mean intraobserver agreement with the final diagnostic categories of microscopic colitis versus nonmicroscopic colitis was 95% (kappa = 0.89). Both interobserver and intraobserver agreement were good in distinguishing among the 6 diagnostic categories, and excellent in distinguishing between microscopic colitis and nonmicroscopic colitis diagnoses. The histologic criteria for microscopic colitis provide for consistent and reproducible interindividual and intraindividual diagnoses in the evaluation of colonic biopsies.