ABSTRACT
Los Angeles classification is widely adopted as a means of unifying endoscopic observation for GERD. We assessed the inter- and intra-observer variability of LA classification. Two-hundred fifty-four still images of the lower esophagus taken by an expert endoscopist [RM] were randomly selected and presented to 9 gastroenterologists [2 females, 6 experts, 3 trainees]. They were asked to report the images according to LA classification. After 2 weeks the images were re-ordered and the same people were asked to report them again. Kappa-statistics was calculated for intra- and inter-observer variability. Mean kappa for intra-observer agreement was 0.54 for experts and 0.55 for trainees [p = NS]. Mean intra-observer kappa was greater among females than males [0.70 vs 0.50 respectively, p=0.05]. Mean inter-observer kappa was 0.20 and 0.31 for experts and trainees respectively [p = NS]. Mean inter-observer weighted kappas were 0.25 and 0.07 [p = 0.007] for males and females respectively. Analyzing data for source of the discrepancy showed that the least reproducible reading was GERD-A both for intra- and inter-observer agreement calculations. According to our data, the LA classification, although a major advance in reporting GERD, has a poor to fair reproducibility. There was no difference between experts and trainees in using the LA classification. Females seem to be more consistent in their readings, but have less agreement with others. Despite the inherent short-comings of kappa statistics and the limitation caused by the possible quality of the still images, revision of the LA system seems to be advisable