Inter-observer variations on interpretation of capsule endoscopies

Objective Capsule endoscopy is a novel investigation for diagnosing small bowel diseases. However, its interpretation is highly subjective and the potential variability may compromise its accuracy and reliability. Here we studied the potential inter-observer variations on the interpretation of capsule endoscopy. Method Two residents and one specialist in gastroenterology independently reviewed 58 capsule endoscopy studies in the same sequential order. The gastric transit time, small bowel transit time, and the most significant small bowel lesion were independently recorded. The consensus transit time was determined by the joint review of the three gastroenterologists. The ‘gold standard’ for small bowel diagnoses was based on final surgical, endoscopic findings or consensus diagnosis. Results Clinically significant and relevant small bowel lesions were found in 32 (55%) cases by consensus review. The overall mean accuracy in determining gastric emptying time, small bowel transit time and small bowel lesion was 89%, 76% and 80%, respectively. There was a significant difference in the accuracy between the residents and specialist on small bowel transit time (P<0.05) and small bowel diagnosis (P<0.05). The mean kappa values on small bowel diagnosis among the three viewers was 0.56 (range, 0.52–0.59). Among various small bowel diagnoses, small bowel bleeding was more accurately identified than other pathology. Conclusions Our results show that there is moderate degree of inter-observer discrepancies on the interpretation of capsule endoscopy. A second reading by an experienced viewer might improve the diagnostic accuracy of this investigation.