Summary
Endoscopy of the stomach was performed in 152 consecutive patients alternately by two endoscopists, such that the non-examining endoscopist always followed the procedure via a teaching attachment. A fiberscope with a close-focusing optical system for magnification of the mucosa was used. The interobserver reliability of eleven predefined macroscopic criteria was analysed and these criteria were compared with the histological diagnosis for each gastric region (body, transitional zone, antrum). The percentage of agreement between the two observers ranged between 80 % and 97 % for all criteria except mucosal erythema (74, 68, 62 %) and mucus covering (53, 61, 87 %).
A consideration of all three gastric regions reveals that only one macroscopic criterion (visibility of submucosal vessels) correlated significantly with the histological diagnosis (i.e. atrophic gastritis). Other macroscopic criteria that correlated significantly with the histological diagnosis were most often observed in the gastric body. Their positive predictive value, however, rarely exceeded 50 percent. Thus, even with the use of modern instruments, endoscopy is of limited value for the prediction of gastritis identified by histology.
Key words:
Endoscopy - Gastritis - Histological diagnosis - Diagnostic value - Post-test probabilities (predictive value)
