Accuracy and Degree of Interobserver Agreement of Small-Caliber Endoscope in Screening for Esophageal Varices

摘要

Background: Small-caliber endoscope has been proposed as an accurate modality for esophageal varices (EV) screening in patients with liver cirrhosis and/or portal hypertension in several small studies. Recent advances in design of endoscope have allowed high-quality video imaging using endoscope with a diameter of less than 5-6 mm. In this study, we aim to evaluate the accuraq' and degree of interobserver agreement of small-caliber endoscope to screen for EV in a large cohort compared to conventional endoscope. Methods: Seventy-five patients with liver cirrhosis and/or portal hypertension (mean age of 56 years; 59% male) were prospectively enrolled between May 2005 and November 2005. Every patient underwent EGD using conventional endoscope (Olympus GIF-160; diameter of 8.6 mm) and small-caliber endoscope (Olympus GIF-N180; diameter of 4.9 mm) back-to-back under the same sedation by 2 different endoscopists who were blinded to the other's results. Using a standard schema, EV were graded both at the timeof EGD and by blinded review of photographs taken by each endoscope. Consensus diagnosis (2 of 3 endoscopists) was used as the final result. Cohen's kappa statistic was used to assess the accuracy and degree of interobserver agreement. Results: Of 75 patients, 22 patients (29%) were Child's class A, 30 patients (40%) were Child's class B, and 23 patients (31%) were Child's class C. Mean MELD score was 13. Twenty-four patients (32%) underwent previous EV banding. Compared to conventional endoscope, the accuracy of small-caliber endoscope for grading of EV was 92% by real-time grading and 93% by consensus review of photographs. All but 1 of the discrepancies in real-time grading were overgrading by small-caliber endoscope compared to conventional endoscope, whereas all of the discrepancies in consensus diagnosis were overgrading. Excellent concordances were demonstrated between real-time grading and consensus diagnosis with kappa of 0.96 for conventional endoscope, 0.94 for small-caliber endoscope, and 0.95 forany type of endoscope. There were excellent interobserver agreements on grading of EV among 3 endoscopists (Table). Child's classification, MELD score, and previous EV banding did not affect the accuracy and degree of interobserver agreement. Conclusion: Small-caliber endoscope is a highly accurate and reliable modality for EV screening in patients with liver cirrhosis and/or portal hypertension.