Development and validation of a non-invasive multimodal risk score incorporating clinical, laboratory, ultrasonographic, and fibrosis indices for prediction of esophageal varices in cirrhosis: A cross-sectional analytical study

Background and Aim: Esophageal varices are a major complication of cirrhosis; non-invasive risk stratification could help identify patients at higher risk. Many existing scores require liver stiffness or other specialized testing. This study aimed to develop and internally validate COVUS-F, a multimodal score using only routinely available clinical, laboratory, ultrasonographic, and fibrosis parameters. Methods: This cross-sectional study included 140 participants: 40 healthy controls and 100 cirrhotic patients (60 with esophageal varices, 40 without). All underwent clinical assessment, laboratory testing, ultrasonography, and endoscopy. Logistic regression and ROC analysis developed the COVUS-F score, internally validated with 1,000 bootstrap resamples. Results: Platelet count, portal-vein diameter, APRI, and Child-Pugh classification formed the COVUS-F score (0–7), with an AUC of 0.903. At a cut-off of ≥5, sensitivity was 70.0%, specificity 85.0% (bootstrap-validated AUC 0.857, 95% CI: 0.781–0.920). High-risk varices occurred in 10.3%, 31.8%, and 61.2% of the low-, intermediate-, and high-risk categories, respectively. Conclusions: COVUS-F showed good, internally validated performance but should not be used as a stand-alone rule-out strategy given its 25.0% missed-high-risk-varices rate. It may serve as a preliminary risk-prioritization tool pending external validation.