Search

Background Several non-invasive diagnostic scores for non-alcoholic fatty liver (NAFL) have been developed, but the clinical application is limited because of their complexity. Aim To develop and validate an easy-to-calculate scoring system to identify ultrasound-diagnosed NAFL. Methods 48,489 patients from 2 centers were included in this study. Multivariable logistic regression models were employed for model development. Ultrasonography was applied to diagnose NAFL. The selected variables were assigned an integer score proportional to the estimated coefficient from the logistic regression analysis, namely NAFL Screening Score (NSS). The ability of the NSS to identify NAFL was assessed by analyzing the area under the receiver operating characteristic curve (AUROC) and was tested in an independent validation cohort. Additionally, the performance of NSS was compared with existing models. Results NSS was developed as a basic score comprising of age, body mass index (BMI), triglyceride (TG), ALT/AST, fasting plasma glucose (FPG) and uric acid (UA) in both sexes. NSS showed a relatively good discriminative power (AUROC = 0.825 for males, 0.861 for females in the validation cohort) in comparison with other models. The optimal cut-off point was 32 for males and 29 for females. Conclusion We developed and validated NSS, an easy-to-use score sheet identify ultrasound-diagnosed NAFL. NSS may be clinically useful for initial diagnosing NAFL.