肝衰竭预后影响因素的Logistic回归分析

目的 探讨影响肝衰竭预后的因素,拟合回归方程用于肝衰竭预后的判断.方法 回顾性收集肝衰竭患者的临床资料,采用二元Logistic回归法对可能影响肝衰竭预后的因素进行分析并建立回归模型,绘制ROC曲线,确定预后模型的最佳临界值.结果 单因素Logistic回归分析显示,甲胎蛋白、血清总胆固醇升高为肝衰竭预后的保护因素;高龄,总胆红素(total bilirubin,TBIL)、国际标准化比值、尿素氮升高和肝性脑病、上消化道出血、腹水为肝衰竭预后的危险因素.经多因素Logistic回归分析筛选出年龄(OR1=7.207,OR2=21.251,P＜0.001)、血清TBIL(OR=0.347,P=0.002)、总胆固醇(OR=3.769,P＜0.001)、腹水(OR=0.142,P=0.002)4个因素用于回归模型的拟合.ROC曲线下面积为0.893,模型预测准确率为86.96％.结论 研究中拟合的回归方程预测准确率较高,对肝衰竭预后的判断有一定参考价值.

Logistic regression analysis of prognostic factors for liver failure

Objective To investigate prognostic factors for liver failure, and to establish a logistic regression model for predicting the prognosis of liver failure. Methods The clinical data of patients with hepatic failure were collected. The binary Logistic regression analysis was used to explore the associations between mortality and prognostic factors and to establish regression models, and then ROC curve was drawn to determine the optimal threshold of the prognostic model. Results Univariate Logistic regression analysis showed that increased levels of alpha fetal protein （AFP） and serum total cholesterol were the protective factors for the prognosis ; older age, elevated levels of serum total bilirubin （TBIL） and urea, increased international normalized ratio （INR）, hepatic encephalopathy, upper gastrointestinal bleeding and ascites were risk factors for the prognosis. In multivariate logistic regression analysis, the independent factors predicting prognosis were age（ OR： = 7. 207, OR2 = 21. 251, P 〈 0. 001 ）, serum total bilirubin （ OR = 0. 347, P = 0. 002）, serum total cholesterol（ OR = 3. 769, P 〈 0. 001 ）and ascites （OR = 0. 142, P = 0. 002）. The area under the ROC curve was 0. 893 and the predicted accuracy rate of the model was 86. 96%. Conclusions The regression model established in the study can predict the prognosis of patients with hepatic failure with high prediction accuracy.