Reliability and safety of transnasal compared to conventional endoscopy for detecting oesophageal varices in cirrhotic patients.

Abstract

Unsedated transnasal endoscopy may be used for detecting oesophageal varices. However, few studies evaluated feasibility and accuracy of this technique. We aimed to evaluate accuracy, interobserver agreement and safety of the transnasal ultrathin compared to conventional endoscopy in patients with cirrhosis. This cross-sectional study included consecutive patients referred for screening or surveillance of oesophageal varices. Patients underwent unsedated transnasal and sedated conventional endoscopies at the same day, which were recorded in a digital video file and randomly analysed by two double-blinded endoscopists. High-risk varices were defined by the presence of large calibre or red wale marks. Accuracy, interobserver agreement and safety of transnasal were compared to conventional endoscopy. One hundred and thirty-three cirrhotic patients (48% male, aged of 60±5, 34% Child-Pugh B/C and 71% of cases for variceal screening) were included in the study. The prevalence of oesophageal varices and high-risk oesophageal varices were 59% (n=79) and 29% (n=39) respectively. For the presence of oesophageal varices, transnasal GIE yielded sensitivity of 94% [95% Confidence Interval, CI 88-99], specificity of 89% [81-97] as well as positive and negative predictive value of 93% and 91% respectively. A satisfactory interobserver agreement was observed for the presence of oesophageal varices (κ=0.89) and high-risk varices (κ=0.65). No serious adverse events were recorded; transnasal GIE was safe and significantly associated with lower rates of hypoxaemia (P<.0001) and hypotension (P<.0001) compared to conventional endoscopy. Unsedated transnasal endoscopy was safe and had an excellent accuracy and high interobserver agreement for detecting oesophageal varices and for identifying high-risk varices in cirrhotic patients.