Development and validation study of a non-alcoholic fatty liver disease risk scoring model among adults in China

Abstract

Background: Non-alcoholic fatty liver disease (NAFLD) is one of the most common liver diseases in China. It is usually asymptomatic and transabdominal ultrasound (USS) is the usual means for diagnosis, but it may not be feasible to have USS screening of the whole population. Objective: To develop a risk scoring model for predicting the presence of NAFLD using parameters that can be easily obtain in clinical settings. Methods: A retrospective study on the data of 672 adults who had general health check including a transabdominal ultrasound. Fractional polynomial and multivariable logistic regressions of sociodemographic and biochemical variables on NAFLD were used to identify the predictors. A risk score was assigned to each predictor using the scaled standardized β-coefficient to create a risk prediction algorithm. The accuracy for NAFLD detection by each cut-off score in the risk algorithm was evaluated. Results: The prevalence of NAFLD in our study population was 33.0% (222/672). Six significant factors were selected in the final prediction model. The areas under the curve (AUC) was 0.82 (95% CI: 0.78–0.85). The optimal cut-off score, based on the ROC was 35, with a sensitivity of 76.58% (95% CI: 70.44–81.98%) and specificity of 74.89% (95% CI: 70.62–78.83%). Conclusion: A NAFLD risk scoring model can be used to identify asymptomatic Chinese people who are at risk of NAFLD for further USS investigation.