The endoscopic diagnosis of gastric ulcer. A randomized clinical trial of interobserver variation and bias induced by knowledge of the radiological diagnosis.

A critical evaluation of the endoscopic diagnosis of gastric ulcer was carried out. In a randomized design it was elucidated whether endoscopists are biased by knowledge of the radiological diagnosis when interpreting their endoscopic findings. In addition, the interobserver variation of the endoscopic diagnosis was determined. A total of 156 patients had a barium-meal examination and were subsequently endoscoped by the same two physicians in one session. Before the endoscopy, the patients were randomized in two groups. In 74 patients the physicians knew the radiologic result at the endoscopy; in 82 they did not. One endoscopist was significantly influenced by his knowledge of the radiological diagnosis. The interobserver variation, expressed as the kappa coefficient, was 0.54 and 0.60, respectively, in the two groups. The disagreement especially concerned small ulcers.