3510 COMPARISON OF INTER- AND INTRA- OBSERVER CONSISTENCY OF ESOPHAGITIS GRADING BY EXPERT AND TRAINEE ENDOSCOPISTS.

Abstract

Upper endoscopy is the most sensitive diagnostic test used for the detection of esophagitis. Its primary purpose is to secure a diagnosis and rule out complications. Apart from clinical practice, grading esophagitis severity is also crucial for standardization of pharmaceutical trials. However, little is known regarding the consistency of esophagitis grading among observers. Our study was designed to describe the intra and inter-evaluator variability in scoring esophagitis using two grading schemes, the Los Angeles (LA) and Hetzel-Dent (HD) scales. Methods: 325 selected photographs of erosive esophagitis or normals were randomly displayed to 9 endoscopists ( 4 experts, 5 fellows) via CD-ROM. Each photograph was presented twice for a total of 650 photographs to be graded in random order. Evaluators scored each photograph using the LA criteria and HD classification scale which are each 5 point Likert scales. The analysis was based on the Kappa (κ) statistic as a measure of consistency (κ >.75 denotes excellent reproducibility,.4 ≤ κ ≤.75 denotes good reproducibility, κ <.4 denotes marginal reproducibility). Results: (Table) The individual fellows showed good reproducibility in the LA classification and marginal reproducibility in the HD classification. The individual experts showed better reproducibility in both the LA and HD classification. The fellows showed good internal consistency in both classifications. Again, the experts were somewhat better in both classifications. As another indication of consistency, in the LA classification only 5.1% and 7.5% of cases were interpreted to be more than one grade different from the mean grade for experts and fellows respectively. This same consistency was found when using the HD grading scheme ( 5.1% and 10.2% respectively). Conclusions: 1) There was good reproducibility of grading esophagitis by both expert endoscopists and fellows; there was less intra-group variability and greater consistency amongst the experts. 2) Both the LA and HD scoring systems are reproducible and consistent endoscopic grading schemes for esophagitis. Upper endoscopy is the most sensitive diagnostic test used for the detection of esophagitis. Its primary purpose is to secure a diagnosis and rule out complications. Apart from clinical practice, grading esophagitis severity is also crucial for standardization of pharmaceutical trials. However, little is known regarding the consistency of esophagitis grading among observers. Our study was designed to describe the intra and inter-evaluator variability in scoring esophagitis using two grading schemes, the Los Angeles (LA) and Hetzel-Dent (HD) scales. Methods: 325 selected photographs of erosive esophagitis or normals were randomly displayed to 9 endoscopists ( 4 experts, 5 fellows) via CD-ROM. Each photograph was presented twice for a total of 650 photographs to be graded in random order. Evaluators scored each photograph using the LA criteria and HD classification scale which are each 5 point Likert scales. The analysis was based on the Kappa (κ) statistic as a measure of consistency (κ >.75 denotes excellent reproducibility,.4 ≤ κ ≤.75 denotes good reproducibility, κ <.4 denotes marginal reproducibility). Results: (Table) The individual fellows showed good reproducibility in the LA classification and marginal reproducibility in the HD classification. The individual experts showed better reproducibility in both the LA and HD classification. The fellows showed good internal consistency in both classifications. Again, the experts were somewhat better in both classifications. As another indication of consistency, in the LA classification only 5.1% and 7.5% of cases were interpreted to be more than one grade different from the mean grade for experts and fellows respectively. This same consistency was found when using the HD grading scheme ( 5.1% and 10.2% respectively). Conclusions: 1) There was good reproducibility of grading esophagitis by both expert endoscopists and fellows; there was less intra-group variability and greater consistency amongst the experts. 2) Both the LA and HD scoring systems are reproducible and consistent endoscopic grading schemes for esophagitis.