论文部分内容阅读
目的建立和验证预测乙肝后肝硬化患者死亡概率的数学模型,为肝硬化患者选择适宜的治疗方案及合理分配肝移植的肝源提供参考依据。方法收集中南大学湘雅附属第一、二、三医院2000年11月—2012年11月收治的1 386例住院乙肝后肝硬化患者的临床记录和随访资料,应用logistic回归分析方法筛选变量建立患者入院后死亡预测数学模型,并运用受试工作者曲线(ROC)及儿童-特科特-pugh分级(CTP)评分、终末期肝病模型(MELD)评分比较其预测效率。结果经多因素logistic回归分析,凝血酶原时间国际标准化比值(INR)、血清钠(Na)、总胆红素(TBIL)、血清肌酐(CRE)、脾厚度、上消化道出血和肝性脑病等7个指标进入回归方程,对乙肝后肝硬化患者死亡预测的敏感度为91.5%,特异度为92.4%,正确指数为0.839,阳性预测值为84.6%,阴性预测值为96.0%,阳性似然比为12.04,阴性似然比为0.09;患者入院3个月时CTP评分、MELD评分、logistic回归模型ROC的曲线下面积(AUG)及其95%CI分别为0.786(0.762~0.789)、0.825(0.794~0.864)、0.912(0.875~0.931),3种评分方法的AUG差异有统计学意义(Z=2.16,P=0.015)。结论 logistic回归模型能较好判断乙肝后肝硬化患者的短期存活或死亡概率,具有较强的诊断准确度和预后评估价值。
Objective To establish and verify the mathematical model of predicting the probability of death in patients with posthepatitic cirrhosis and provide the reference basis for choosing the suitable treatment plan and rationally distributing the liver of liver transplant recipients. Methods The clinical records and follow-up data of 1 386 hospitalized patients with post-hepatitis B cirrhosis who were admitted to the First, Second and Third Hospitals, Xiangya Affiliated Hospital of Central South University from November 2000 to November 2012 were collected. Logistic regression analysis was used to select the variables to establish the patients The mathematical model of mortality prediction after admission was used and the prediction efficiency was compared by using the ROC and CTP scores and the end-stage liver disease model (MELD) score. Results By multivariate logistic regression analysis, the international ratio of prothrombin time (INR), serum sodium (Na), total bilirubin (TBIL), serum creatinine (CRE), spleen thickness, upper gastrointestinal bleeding and hepatic encephalopathy And other seven indicators into the regression equation, the prognosis of hepatitis B patients with liver cirrhosis mortality prediction was 91.5%, the specificity was 92.4%, the correct index was 0.839, the positive predictive value was 84.6%, the negative predictive value was 96.0%, positive The ratio of negative predictive value (OR) was 12.04, negative likelihood ratio was 0.09; CTP score, MELD score and ROC area under curve (AUG) and 95% CI of logistic regression model at 3 months were 0.786 (0.762-0.889) and 0.825 (0.794-0.864) and 0.912 (0.875-0.931). There was significant difference in AUG between the three methods (Z = 2.16, P = 0.015). Conclusion The logistic regression model can better determine the probability of short-term survival or death in patients with posthepatitic cirrhosis, which has a strong diagnostic accuracy and prognostic value.