Background and study aims: In resource-limited countries, non-invasive tests for assessing liver fibrosis are a potential alternative to costly endoscopic screening for esophageal varices. We aimed to validate several non-invasive parameters for predicting the presence of varices. Patients and methods: Between September 2006 and August 2017, a total of 46,014 patients who underwent upper gastrointestinal endoscopy as one of the perquisites for receiving hepatitis C virus (HCV) therapy were enrolled and divided into group I (without varices) and group II (with varices). Non-invasive parameters of fibrosis, namely Lok index, Bonacini score, liver stiffness, FIB-4, Baveno, and extended Baveno criteria, were validated. Results: Lok index, Bonacini score, liver stiffness, and FIB-4 had areas under the receiver operating characteristic curve (AUCs) of >0.6 (all P < 0.01 for the null hypothesis that the AUC was 0.5) for determination of the presence/absence of varices, with cutoff values of 0.80, 6.5, 21.9 kPa, and 2.94, and sensitivities of 74%, 74%, 66%, and 83%, respectively. The expanded Baveno VI criteria performed better than the Baveno VI criteria (spared endoscopy rate 81% versus 63%). Conclusion: The use of non-invasive methods is of limited value in predicting esophageal varices. The limited accuracy of ≤60% may delay the use of appropriate primary prophylaxis against variceal bleeding in a large proportion of cirrhotic patients.
