Observer agreement on the grading of gastric atrophy

Aims: Assessment of lesser or doubtful degrees of gastric atrophy can be di fficult, especially in the antrum, since well established criteria are lack ing. At the Houston Working Party on Gastritis in 1994 a visual analogue sc ale was designed for the grading of histopathological parameters. This was done to promote uniformity in grading by acting as a reference. The purpose of the present study was to measure interobserver variation between pathol ogists familiar with the Houston visual analogue scale in a specifically se lected set of biopsies from patients with lesser or doubtful degrees of atr ophy. Methods and results: Thirty cases with biopsies of the antrum and corpus fr om a long-term follow-up study on Helicobacter pylori gastritis comprised t he current study material. The cases were selected from that study because there had been uncertainty or disagreement on the presence of gastric atrop hy, The study set of haematoxylin and eosin (H & E) slides was circulated a mongst gastrointestinal pathologists familiar with the visual analogue scal e who were unaware of the source of the study set nor had any other clinica l information, Interobserver variability was analysed using kappa statistic s. The overall agreement for the grade of atrophy in antral biopsies was 0. 461; the kappa value was 0.18 (95% confidence limits 0.12-0.24), which is c onsidered poor agreement, The kappa value was nevertheless statistically si gnificant (P<0.01). The overall agreement on the grade of atrophy in the co rpus biopsies was apparently good (0.833), but the kappa which adjusts for chance agreement was only moderate (0.48; 95% confidence limits 0.42-0.55: P<0.001), Conclusion: The studied series comprised a self-selected sample in which th ere was doubt about the grade of atrophy and such a sample will produce low er kappa values than a random sample of gastric biopsies. The results never theless confirm that better guidelines and firm criteria are needed to prop erly diagnose and grade gastric atrophy. It is suggested that the use of tw o grades, low- and high-grade atrophy, akin to that in use for grading infl ammatory bowel disorder (IBD)-associated dysplasia, could Improve agreement . Furthermore optimal biopsy quality with full thickness mucosa and proper orientation appears important for grading gastric atrophy.