Evaluation of reliability in the use of endoscopic terminology

Variability in the reporting of gastrointestinal endoscopic findings may affect the validity of analyses of data collected from clinical reports of those findings. In this project, images of 10 endoscopic findings were collected from the data repository of the Clinical Outcomes Research Initiative (CORI), all of which had been described by the reporting endoscopist. These images were presented to 52 experienced endoscopists recruited from the clinical affiliates of CORI who were asked to assign each a term from the Minimal Standard Terminology for Digestive Endoscopy. Proportion of agreement with the endoscopist varied by finding from 84.3% to 51.0% (overall 67.6% with 95% CI 63.4-71.8%). Proportion of agreement among the subjects varied by finding from 76.3% to 38.5%. (overall 55.6% with 95% CI 52.4-58.8%). Possible reasons for this lack of agreement are discussed.