NBI With Optical Magnification Shows Good Interobserver Agreement in Diagnosing Gastritis.

OBJECTIVE

To study the interobserver agreement for diagnosing gastritis using narrow band imaging (NBI) with magnification.

METHODS

This prospective study recruited patients who underwent gastroscopy for dyspepsia in the Department of Gastroenterology, Aster MIMS Hospital, Kannur.361 patients were included in the study. The gastroscopy was performed using high-definition white light endoscopy (WLE) and NBI with magnification. Histopathology and rapid urease test were used to detect HP infection. Endoscopy videos were analyzed by three trained endoscopists who were blinded to each other. NBI patterns were classified into four types (Types 1, 2A, 2B, and 3). Interobserver variability was examined using Kappa Statistics.

RESULTS

164 of 361 patients had HP infection (45.42%). Of 361 people, 199 had Type 1 (55.12%), 54 Type 2a (14.95%), 65 Type 2b (18%), and 43 Type 3 cases (11.91%). There was good interobserver agreement with a kappa value of 0.730 (95% confidence interval (CI) 0.693-0.768). 87% of HP-negative patients had Type 1 NBI pattern. 79.5% of type 2A and 89.2% of Type 2b pattern were HP positive. The Type 1 pattern identified normal gastric mucosa with 84.75% sensitivity, 88.32% specificity and 87.4% negative predictive value. Type 2 and Type 3 NBI patterns had high positive predictive value and specificity for HP infection.

CONCLUSIONS

This study demonstrates excellent interobserver agreement among experienced endoscopists in using NBI to identify gastric mucosal patterns associated with HP infection. While biopsies remain essential for comprehensive evaluation of gastric pathologies, our findings suggest that NBI, with appropriate training and validation, may have the potential to reduce the need for biopsies in specific cases where the primary concern is HP infection. The good level of interobserver agreement seen in our study is encouraging and suggests that NBI has the potential to be a reliable tool for diagnosing HP infection.