为探究肝功能衰竭患者短期预后的影响因素并初步探讨年龄与相关因素交互作用对预后的影响.通过回顾南京中医药大学附属南京医院2020 年6 月—2022 年5 月收治的肝功能衰竭患者作为研究对象,采集患者入院时人口学特征、并发症、住院期间实验室检查指标,采用单因素、二元Logistic回顾分析研究影响肝衰患者预后的相关因素,采用受试者工作特征曲线(ROC)评估模型效能,采用R语言分析年龄与独立危险因素的交互作用对预后影响.结果表明:共纳入 371 例患者,治疗2 周预后良好组257(69.27%),预后不良组114(30.73%),单因素筛选可能的影响因素有:并发症数量、肝性脑病、胆汁酸最低值、胆汁酸变化率、直接胆红素最低和最高值、直接胆红素变化率、肌酐、尿素氮(P<0.1),二元Logistic回归分析结果发现并发症数量[1.287(95%CI:1.049~1.58),P =0.016]、胆汁酸最低值[1.007(95%CI:1~1.013),P =0.007]及直接胆红素最小值[1.008(95%CI:1.001~1.014),P =0.015]、肌酐[1.008(95%CI:1.001~1.016),P =0.028]、CRP[1.012(95%CI:1~1.025),P =0.047]以及肝性脑病[4.509(95%CI:2.469~8233),P<0.001]为肝衰患者的独立影响因素(P<0.05),ROC曲线显示并发症数量临界值2.5 个,AUC 0.713(95%CI:0.656~0.771)、胆汁酸最低值临界值 57.8 μmol/L,AUC 0.607(95%CI:0.546~0.668)、直接胆红素最低值临界值 71.05 μmol/L,AUC 0.705(95%CI:0.648~0.762)、肌酐临界值77.5 μmol/L,AUC 0.585(95%CI:0.522~0.649)、CRP临界值22.99 mg/L,AUC 0.550(95%CI:0.484~0.615),联合预测因子ROC曲线下面积AUC 0.809,敏感性为66.7%,特异性为83.7%,年龄与胆汁酸最低值、直接胆红素最低值及肝性脑病对预后影响存在协同作用.可见基于肝衰竭患者并发症数量、胆汁酸最低值、直接胆红素最低值,肌酐及肝性脑病构建的Logistic回归模型具有较好的预测分类能力,具备高龄和高胆汁酸、高直接胆红素最低值、高肌酐及肝性脑病阳性因素的患者不良预后风险的概率可能更高.
To investigate the factors influencing the short-term prognosis of patients with liver function failure and to explore the in-teraction between age and related factors on prognosis.Patients with liver function failure admitted to Nanjing Hospital Affiliated to Nanjing University of Chinese Medicine were retrospectively analyzed between June 2020 and May 2022.The patients'demographic characteristics,complications and laboratory examination indicators were collected on admission,and univariate and binary logistic re-gression analyses were performed to identify the factors affecting the prognosis of liver failure patients.The model's efficacy was evalua-ted using the receiver operating characteristic(ROC)curve,and R language was used to analyze the interaction between age and inde-pendent risk factors on prognosis.The results show that a total of 371 patients are included,with 257(69.27%)in the good prognosis group after 2 weeks of treatment and 114(30.73%)in the poor prognosis group.Univariate screening identifies several possible influ-encing factors,including the number of complications,hepatic encephalopathy,minimum bile acid value,bile acid change rate,mini-mum and maximum values of direct bilirubin,direct bilirubin change rate,creatinine,and blood urea nitrogen(P<0.1).Binary lo-gistic regression analysis show that the number of complications[1.287(95%confidence interval(CI):1.049~1.58),P =0.016],minimum bile acid value[1.007(95%CI:1~1.013),P =0.007],minimum value of direct bilirubin[1.008(95%CI:1.001~1.014),P =0.015],creatinine[1.008(95%CI:1.001~1.016),P =0.028],C-reactive protein(CRP)[1.012(95%CI:1~1.025),P =0.047],and hepatic encephalopathy[4.509(95%CI:2.469~8233),P<0.001]are independent factors affecting the prognosis of patients with liver failure(P<0.05).The ROC curve shows that the critical value of the number of complications is 2.5,with an area under the curve(AUC)of 0.713(95%CI:0.656~0.771).The critical value of minimum bile acid value is 57.8 μmol/L,with an AUC of 0.607(95%CI:0.546~0.668).The critical value of minimum value of direct bilirubin is 71.05 μmol/L,with an AUC of 0.705(95%CI:0.648~0.762).The critical value of creatinine is 77.5 μmol/L,with an AUC of 0.585(95%CI:0.522~0.649).And the critical value of CRP is 22.99 mg/L,with an AUC of 0.550(95%CI:0.484~0.615).The combined prediction factor has an AUC of 0.809,sensitivity is 66.7%,and specificity is 83.7%,age,minimum bile acid value,minimum value of direct bilirubin,and hepatic encephalopathy has a synergistic effect on prognosis.In conclusion,the Logistic regression model constructs based on the number of complications,minimum bile acid value,minimum value of direct bilirubin,creatinine,and hepatic encephalopathy of patients with liver function failure has good predictive classification ability.Patients with advanced age,high bile acid,low minimum value of direct bilirubin,high creatinine,and hepatic encephalopathy may have a higher risk of poor prognosis.