Interobserver Agreement in Gastric Lesions by Magnifying Endoscopy

Ko, Bong Min; Jang, Jae Young; Kwon, Kwang An; Kim, Seong Hawn; Lee, Sang Ho; Jung, In Seop; Hong, Su Jin; Ryu, Chang Beom; Cho, Young Deok; Kim, Jin Oh; Cho, Joo Young; Lee, Joon Seong; Lee, Moon Sung; Shim, Chan Sup; Kim, Boo Sung

Abstract

BACKGROUND/AIMS
Various magnifying endoscopic classifications have been introduced in esophageal and gastric lesions. However, studies on an agreement between observers according to the form classification of magnifiying endoscopic findings have not been performed yet. The aim of this study was to determine the interobserver agreement for magnifying endoscopic classifications.
METHODS
The patients were divided into 3 groups. The first group of patients had post-EMR ulcer scar (50 cases), the second group, elevated gastric lesion (38 cases), and the third group, gastritis (43 cases). Two mucosal patterns were used in the post-EMR ulcer scar group, 6 mucosal patterns in the elevated gastric lesion group, and 3 patterns of the collecting venule in the gastritis group. Three experienced observers (A, B, C) blinded to the patients' data participated in this study. The agreement between observers was evaluated by calculated kappa. The kappa value of 0.75 or greater was rated excellent, 0.4~0.74, fair to good, and 0.4 or less, poor.
RESULTS
Regarding 2 mucosal pattern in post-EMR ulcer scar, agreements between A and B, A and C, and B and C were 0.896, 0.793, and 0.901, respectively (p<0.01). Regarding 6 mucosal patterns in elevated gastric lesion, agreements between A and B, A and C, and B and C were 0.607, 0.458, and 0.557, respectively (p<0.01). Regarding 3 collecting venule pattern in gastritis, agreements between A and B, A and C, and B and C were 0.822, 0.823, and 0.751, respectively (p<0.01).
CONCLUSIONS
Interobserver agreement is good to excellent in mucosal pattern and collecting venule by magnifying endoscopy. However magnifying endoscopic classification needs to be refinded in order to improve an agreement between observers.