Comparison of inter- and intraobserver consistency for grading of esophagitis by expert and trainee endoscopists

Background: This study evaluated intra- and interobserver variability in the endoscopic scoring of esophagitis by using two grading schemes, the Los Angeles and Hetzel-Dent scales. Methods: Three hundred twenty-five endoscopic photographs of esophagitis or healed esophagitis were randomly displayed twice to 9 endoscopist evaluators (4 experts, 5 trainees) by means of a CD-ROM. Evaluators scored each photograph by using both classification schemes. The analysis was based on the kappa (κ) statistic as a measure of consistency. Results: For trainees, intraobserver reproducibility was good (κ = 0.436) when using the Los Angeles classification and marginal (κ = 0.395) by using the Hetzel-Dent classification. For experts, intraobserver reproducibility was better by using both the Los Angeles (κ = 0.545) and Hetzel-Dent classification (κ = 0.552). Interobserver consistency for trainees was good when using both classifications (Los Angeles, κ = 0.459; Hetzel-Dent, κ = 0.427). Again, intraobserver reproducibility was better for experts when using both classifications (Los Angeles, κ = 0.556, HetzelDent, κ = 0.571). By using the Los Angeles classification, only 5.1% and 7.5% of the photographs were interpreted to be more than 1 grade different from the mean grade for, respectively, experts and trainees. This same consistency was found by using the Hetzel-Dent classification scheme (experts 5.1%, trainees 10.2%). Conclusions: Reproducibility in grading esophagitis was good for both expert endoscopists and fellows, although there was greater consistency among the experts. Both the Los Angeles and Hetzel-Dent scoring systems are reproducible.