Interobserver Variation in the Endoscopic Diagnosis of Gastroesophageal Reflux Disease / 대한소화기내시경학회지

ABSTRACT

BACKGROUND/AIMS: A diagnosis of gastroesophageal reflux disease (GERD) is based on the typical symptoms, such as acid regurgitation and heartburn. However, there is a very high inter-observer variation in the evaluation of GERD patients. METHODS: The endoscopic images of forty-two cases with reflux symptoms (2 still images and 15-second video images per case) were analyzed by 18 experienced endoscopists and 22 trainees. The findings were classified into the following (1) 6 groups (modified LA classification 4 LA groups, minimal, and normal), (2) erosinve and non-erosive, and (3) confluent erosive and others. The level of inter-observer variation is expressed as a kappa value. RESULTS: The level of inter-observer agreement of the 18 experienced endoscopists for classifying the patients into 6 groups was fairly low (kappa=0.364). However, when the findings were classified into the 2 groups suggested in the Genval workshop (NERD, A, or B versus C or D), the level of inter- observer agreement increased substantially (kappa=0.710). The kappa value of the 22 trainees for classifying the patients into 6 groups was 0.402. CONCLUSIONS: Modified LA classification with minimal change lesions showed a fairly low level of agreement. The problem caused by inter-observer variations decreased significantly when the findings were classified into two groups.