代偿期乙肝肝硬化进展为慢加急性肝衰竭的危险因素分析和预测模型的构建

Risk Factors Analysis and Prediction Model of Compensatory Hepatitis B Cirrhosis Progressing to Acute-on-Chronic Liver Failure

目的筛选慢加急性肝衰竭(ACLF)的危险因素,建立预测模型并评估其在预测代偿期乙肝肝硬化进展为ACLF中的应用价值。方法收集390例代偿期乙肝肝硬化急性发作入院患者的临床资料,根据起病28 d内是否发生ACLF分为Non-ACLF组(287例)和ACLF组(103例)。对两组年龄、性别、白细胞(WBC)、凝血酶原时间(PT)、纤维蛋白原(Fib)、结合胆红素(CB)、白蛋白(Alb)、丙氨酸氨基转移酶(ALT)、谷氨酰转肽酶(GGT)、总胆汁酸(TBA)、肌酐(Cr)、血清钠离子(Na~+)进行单因素分析。单因素分析后P<0.05的变量用于构建logistic回归模型,并绘制受试者工作特征(ROC)曲线,通过曲线下面积(AUC)评判模型的诊断价值。结果经单因素分析,两组性别、WBC、PT、Fib、CB、Alb、ALT、GGT、TBA、Na~+比较,差异有统计学意义(P<0.05)。PT(OR=1.993,P=0.001)、CB(OR=1.078,P<0.001)、Alb(OR=0.800,P=0.038)、TBA(OR=1.113,P<0.001)是代偿期乙肝肝硬化进展为ACLF的独立危险因素。预测模型的ROC的AUC为0.998,灵敏度为98.6%,特异度为99.0%。结论由PT、CB、Alb、TBA建立的预测模型预测ACLF准确、敏感,具有一定的临床应用价值。

Objective To screen the risk factors of acute-on-chronic liver failure(ACLF),the prediction model was established and its application value in predicting the progression of compensatory hepatitis B cirrhosis to ACLF was evaluated.Methods Clinical data of 390 patients with acute compensatory hepatitis B cirrhosis were collected.According to the occurrence of ACLF within 28 days of onset,the patients were divided into Non-ACLF group(287 cases)and ACLF group(103 cases).Age,sex,white blood cell(WBC),prothrombin time(PT),fibrinogen(Fib),conjugated bilirubin(CB),albumin(Alb),alanine aminotransferase(ALT),glutamyl transpeptidase(GGT),total bile acid(TBA),creatinine(Cr)and serum sodium ion(Na+)of the two groups were performed.After univariate analysis,the variables P<0.05 were used to construct logistic regression model,and the curve of receiver operating characteristic(ROC)was plotted.The area under the curve(AUC)was used to evaluate the diagnostic value of the model.Results Univariate analysis showed that sex,WBC,PT,Fib,CB,Alb,ALT,GGT,TBA and Na^(+)of the two groups had statistical difference(P<0.05).PT(OR=1.993,P=0.001),CB(OR=1.078,P<0.001),Alb(OR=0.800,P=0.038)and TBA(OR=1.113,P<0.001)were independent risk factors for the progression of compensatory hepatitis B cirrhosis to ACLF.The AUC of the ROC of the prediction model was 0.998,and the sensitivity was 98.6%,and the specificity was 99.0%.Conclusion The prediction model established by PT,CB,Alb and TBA is accurate and sensitive in predicting ACLF and has certain clinical application value.