Interobserver variability in application of the revised Sydney classification for gastritis

The Sydney classification for gastritis provides guidelines for histological grading of gastric biopsies. In an ongoing study of gastric preneoplastic lesions in Chiapas, Mexico, 7 biopsies from 150 patients (4 from the antrum and 3 from the body) were obtained during endoscopy and studied histologically. The first 74 endoscopy specimens were read independently by 2 general surgical pathologists. We assessed diagnostic concordance using kappa statistics. The 2 pathologists then jointly reviewed biopsies about which they had disagreed to reach a final diagnosis. A second group of 76 endoscopies was subsequently evaluated independently by the 2 pathologists, and concordance was again assessed. In the first group of biopsies, we found low concordance rates (Heliobacter pylori 0.59, acute inflammation 0.22, intestinal metaplasia 0.60, and atrophy 0.04). In the second group, of independently reviewed cases, there was better concordance (H pylori 0.77, acute inflammation 0.50, intestinal metaplasia 0.70, and atrophy 0.64). We presumed that use of the Sydney classification would result in minimal interpretational differences achieving ideal kappas greater than 0.80. Because pathology results are based on subjective interpretation of this classification, complete diagnostic agreement is practically impossible. Concordance by general surgical pathologists after joint review of cases was similar to that obtained by gastrointestinal pathologists.