Interobserver agreement on classifying endoscopic diagnoses of nonerosive esophagitis

Background and study aims: The majority of patients with gastroesophageal reflux disease in Japan have low-grade esophagitis, including minimal changes. A modified Los Angeles classification of esophagitis, consisting of erosive esophagitis (grades A-D) and nonerosive esophagitis (grades M and N) has been proposed and is in clinical use in Japan. However, it is unclear whether nonerosive esophagitis with only undemarcated mucosal discoloration (grade M) is clinically significant, since interobserver variations in classification have not been investigated. The aim of the present study was therefore to evaluate interobserver variance and diagnostic agreement in the diagnosis of nonerosive esophagitis (grades M and N). Materials and methods: A total of 84 endoscopists were enrolled to assess the grade of esophagitis in 30 patients by viewing endoscopic images of the gastroesophageal junction. The images were projected onto a screen, and all of the endoscopists reviewed them concurrently. The diagnosis was selected from the following three categories in the modified Los Angeles classification: grades N, M, or A. The endoscopists were grouped according to their experience, whether they had a board license, and whether they had received specialist training in esophagitis. The kappa coefficient of reliability was calculated. Results: The kappa coefficient of reliability for all the endoscopists in the diagnosis of cases of grade M and N nonerosive esophagitis was unacceptably low at 0.22 (95% Cl, 0.21 - 0.24). Endoscopists with a board license and those who had completed a special esophagitis diagnostic course had slightly higher kappa values (0.26; 95% Cl, 0.23 - 0.30 and 0.29; 95% Cl, 0.26 - 0.32), respectively. Conclusions: Interobserver agreement on the endoscopic diagnosis of nonerosive esophagitis (grades M and N) is too low to be of clinical value.