摘要
目的 分析血浆置换治疗肝衰竭后发生不良反应的影响因素,构建预测血浆置换治疗肝衰竭后发生不良反应的列线图预测模型.方法 选取2019年3月至2022年3月淮安市第二人民医院收治的132例肝衰竭患者为研究对象,依据血浆置换治疗后是否发生不良反应分为发生不良反应组(50例)和未发生不良反应组(82例),比较两组患者性别、年龄、临床分期、血白细胞、总胆红素(total bilirubin,TBil)、直接胆红素(direct bilirubin,DBil)、白蛋白(albumin,ALB)、丙氨酸氨基转移酶(alanine aminotransferase,ALT)、天门冬氨酸氨基转移酶(aspartate aminotransferase,AST)、凝血酶原活动度(prothrombin activity,PTA)、血肌酐、肝性脑病等的差异.采用多因素Logistic回归分析血浆置换治疗肝衰竭后发生不良反应的影响因素.采用R软件构建预测血浆置换治疗肝衰竭后发生不良反应的列线图模型,绘制受试者工作特征(receiver operator characteristic,ROC)曲线评估列线图预测血浆置换治疗肝衰竭后发生不良反应的区分度,采用Hosmer-Lemeshow拟合优度检验与校准曲线评估其一致性.结果 发生不良反应组患者晚期肝衰竭比例[56.00%(28/50)比32.93%(27/82)]、血白细胞[(17.51±3.61)×109/L比(13.64±2.32)×109/L]、TBil[(228.49±40.27)μmol/L比(100.65±26.26)μmol/L]、DBil[(120.52±31.82)μmol/L比(74.26±21.06)μmol/L]、AST[(178.64±56.32)U/L比(79.06±17.08)U/L]、ALT[(216.51±53.95)U/L比(84.62±17.64)U/L]、血肌酐[(156.85±26.72)μmol/L比(127.75±22.96)μmol/L]、并发肝性脑病比例[60.00%(30/50)比12.20%(10/82)]均显著高于未发生不良反应组(P均＜0.05),ALB[(29.63±8.27)g/L比(50.26±10.19)g/L]和PTA[(28.63±8.09)%比(41.68±7.06)%]水平显著低于未发生不良反应组(P均＜0.05).多因素Logistic回归分析表明,中期肝衰竭(OR = 2.706,95%CI:2.234～14.576,P＜0.001)、晚期肝衰竭(OR = 4.532,95%CI:1.762～11.628,P = 0.002)、TBil(OR = 1.028,95%CI:1.016～1.040,P＜0.001)、并发肝性脑病(OR = 5.602,95%CI:1.332～23.562,P = 0.019)是血浆置换治疗肝衰竭后发生不良反应的危险因素,PTA为保护因素(OR = 0.902,95%CI:0.852～0.954,P＜0.001).血浆置换治疗肝衰竭后发生不良反应的列线图预测模型具有较好的区分度(ROC曲线下面积为0.958,95%CI:0.926～0.991)和一致性(Hosmer-Lemeshow拟合优度检验χ2 = 8.555,P = 0.381).结论 本研究构建的预测血浆置换治疗肝衰竭后发生不良反应的列线图模型可识别血浆置换治疗肝衰竭后发生不良反应的高风险患者.
Abstract
Objective To analyze the influencing factors of adverse reactions after plasma exchange for liver failure and to construct a nomogram prediction model for predicting adverse reactions after plasma exchange for liver failure.Methods Total of 132 cases with liver failure in Huai'an Second People's Hospital from March 2019 to March 2022 were selected and divided into adverse reaction group(50 cases)and non-adverse reaction group(82 cases)according to whether adverse reactions occurred after plasma exchange therapy.The differences of gender,age,clinical stage,blood leukocytes,total bilirubin(TBil),direct bilirubin(DBil),albumin(ALB),alanine aminotransferase(ALT),aspartate aminotransferase(AST),prothrombin activity(PTA),blood creatinine and hepatic encephalopathy of patients in two groups were compared.Multivariate Logistic regression analysis was applied to analyze the influencing factors of adverse reactions after plasma exchange for liver failure.R software was applied to construct a nomogram model for predicting adverse reactions after plasma exchange therapy for liver failure.Receiver operator characteristic(ROC)curve was used to assess the discrimination of column plots predicting the occurrence of adverse effects after plasma exchange for liver failure,and Hosmer-Lemeshow goodness-of-fit was used to assess the agreement with calibration curves.Results The proportion of advanced liver failure[56.00%(28/50)vs.32.93%(27/82)],blood white blood cells[(17.51±3.61)×109/L vs.(13.64±2.32)×109/L],TBil[(228.49±40.27)μmol/L vs.(100.65±26.26)μmol/L],DBil[(120.52±31.82)μmol/L vs.(74.26±21.06)μmol/L],AST[(178.64±56.32)U/L vs.(79.06±17.08)U/L],ALT[(216.51±53.95)U/L vs.(84.62±17.64)U/L],serum creatinine[(156.85±26.72)μmol/L vs.(127.75±22.96)μmol/L]and the proportion of hepatic encephalopathy[60.00%(30/50)vs.12.20%(10/82)]of patients in adverse reaction group were significantly higher than those of non-adverse reaction group,and the levels of ALB[(29.63±8.27)g/L vs.(50.26±10.19)g/L]and PTA[(28.63±8.09)%vs.(41.68±7.06)%]were significantly lower(all P＜0.05).Logistic regression analysis showed that middle stage liver failure(OR = 2.706,95%CI:2.234～14.576,P＜0.001),advanced liver failure(OR = 4.532,95%CI:1.762～11.628,P = 0.002),TBil(OR = 1.028,95%CI:1.016～1.040,P＜0.001)and hepatic encephalopathy(OR = 5.602,95%CI:1.332～23.562,P = 0.019)were risk factors for adverse reactions after plasma exchange,while PTA was a protective factor(OR = 0.902,95%CI:0.852～0.954,P＜0.001).The constructed nomogram prediction model for adverse reactions after plasma exchange therapy of liver failure had good discrimination(area under the ROC curve was 0.958,95%CI:0.926～0.991)and consistency(Hosmer-Lemeshow goodness of fit test χ2 = 8.555,P = 0.381).Conclusions The nomogram model constructed for predicting adverse reactions after plasma exchange therapy for liver failure can identify patients with high-risk of adverse reactions after plasma exchange therapy.
NETL
NSTL
万方数据