Francois F, Weinshel EH, Perez-Perez GI, Yee HT, Blaser MJ, Bini EJ. Endoscopic Training, Looking Past The Surface. Gastrointestinal Endoscopy. 2003; AB109,57(5)


Abstract

Endoscopy remains the best modality for the evaluation of suspected upper GI tract lesions. Studies have reported differing degrees of agreement between endoscopic observations and histological results. Often the diagnosis of varying grades of esophagitis or gastritis is made by the morphology and color of the mucosa alone without further confirmation. An important aspect of endoscopic training is reviewing the histology from biopsied lesions. Aim: To determine the agreement between endoscopic and histologic modalities for the diagnosis of upper GI tract findings in an endoscopy training program. Methods: 118 patients undergoing upper endoscopy for clinically indicated reasons were enrolled in the study. GI fellows performed endoscopies under the supervision of attending physicians. Biopsies were obtained from the antrum, body, fundus, and esophagus in all patients for histologic examination. Histologic evaluation was performed blinded to the clinical history and endoscopic findings. The kappa statistic was used to evaluate agreement between the endoscopic and histologic diagnosis of esophagitis, Barrett's esophagus (BE), and gastritis (kappa >0.75 represents clinically acceptable agreement, 0.75-0.40 denotes fair agreement, and <0.40 denotes poor agreement). The operating characteristics for endoscopic testing were determined using histology as the gold standard. Results: There was poor agreement between the endoscopic and histologic diagnosis of esophagitis (4% vs. 13%, kappa = 0.04), and although there was fair agreement between the endoscopic and histologic diagnosis of BE it was not clinically acceptable (20% vs. 9%, kappa =0.43). There was poor agreement between the endoscopic diagnosis of gastropathy in 76% of patients and the histologic diagnosis of gastritis in 73% (kappa = 0.07). Endoscopic agreement with the histology worsened for the subgroup of patients with iron deficiency anemia. Overall endoscopy was 7% sensitive and 96% specific for the diagnosis of esophagitis, 80% sensitive and 86% specific for the diagnosis of BE, and 78% sensitive and 29% specific for the diagnosis of gastritis. Conclusions: The diagnostic accuracy of endoscopic findings varied depending on the lesion being evaluated. There was poor agreement between the endoscopic and histologic finding of esophagitis and gastritis, and clinically unacceptable agreement between the endoscopic and histologic diagnosis of BE. Routine biopsies for the accurate diagnosis of GERD and gastritis should be reinforced in endoscopic training programs.