The Study Of Screening Models And Prediction Models Of NAFLD

Nonalcoholic fatty liver disease(NAFLD)refers to a clinical pathology syndrome with the excessive deposition of fat as a main feature,excluding those caused by alcohol and other specific liver damage.And it is regarded as a metabolic stress liver injury which is closely related to insulin resistance and genetic susceptibility.In recent years,with the changes in living habits and diet,the incidence and prevalence of NAFLD have continued rising.And due to NAFLD not only does harm to liver tissue but also serves as a risk factor to some severe chronic diseases,such as cardiovascular disease,metabolic syndrome and chronic kidney disease,widely concern from scientists has aroused towards it.Therefore,early screening and diagnosis of NAFLD is especially important.In this study,based on cross-sectional and cohort data from "Shandong Multi-center Longitudinal Cohort for Health Management",Logistic model and Cox model were adopted to build screening and 3-years risk prediction model to detect high-risk population and evaluate the 3-years risk of NAFLD.Results:? NAFLD screening model1.A total of 18620 subjects including 10030 males and 8590 females were selected to build the screening model.There were 3811 NAFLD cases in male and the detection rate was 38.00%.For female,1898 cases were diagnosed with NAFLD and the detection rate was 22.10%.The detection rate of NAFLD in male was higher than it in female(?2=550.27,P<0.001).The detection rate was increasing with the age increases both for male and female.2.All variables were significantly different between subjects with NAFLD and without NAFLD in female.For male,all variables were significantly different except for age.3.According to the logistic regression,age,BMI,DBP,ALT,GGT,FBG,TG,HDL-C and LDL-C were finally selected into screening model both for male and female.4.The AUC(95%CI)of the screening model for male and female were 0.800(0.792,0.807)and 0.844(0.836,0.852)respectively.After the external validation,the AUC(95%CI)for male and female were 0.845(0.832,0.858)and 0.868(0.854,0.882),respectively.? NAFLD prediction model1.A total of 3429 subjects including 1847 male and 1582 female were selected into the NAFILD cohort to build the prediction model.During the follow-up,1130 new NAFLD cases were observed including 694 males and 436 females.The incidence density was 13.05 per 100 person-years in total and 13.57 per 100 person-years for male and 10.53 per 100 person-years for female respectively.2.All variables were significantly different between subjects with NAFLD and without NAFLD in baseline for male.For female,all variables were significantly different except for NC.3.According to the result of Cox regression,the variables in the final prediction model include age,BMI,DBP,GGT,TG,HDL-C and LDL-C for male,and age,BMI,DBP,TG,LDL-C and HDL-C for female.4.The AUC(95%Cl)of the 3-year prediction model for male and female were 0.724(0.703.0.744)and 0.773(0.751,0.793)respectively.After external validation,the AUC(95%CI)for male and female were 0.715(0.665,0.765)and 0.712(0.654,0.770)respectively.Conclusions:1.The detection rate of NAFLD in male is higher than it in female.And the detection rate of NAFLD shows an increasing trend with the age increases.2.The screening models for male and female show good competence in discriminating NAFLD and non-NAFLD subjects.3.The 3-year prediction models for male and female do well in predicting the 3-years risk of developing NAFLD.