摘要
目的利用急诊室测得的临床指标建立一个预测患者住院期间死亡预警评分模型,并且评估该评分模型预测急性主动脉夹层(acute aortic dissection,AAD)患者住院期间病死率价值.方法回顾性分析2016年1月1日至2019年1月1日期间就诊于苏州大学附属第一医院抢救室的208例AAD患者的病例资料,包括入院时生命体征、血常规、肝肾功能及凝血功能等指标.根据患者临床结局将其分为生存组(n=167)与死亡组(n=41),并且根据临床结局资料分别进行两组间的比较.使用SPSS Modeler 18.0中的卡方自动交互检测法(CHAID)建立模型,将两组之间差异有统计学意义的指标纳入到模型中,以住院期间死亡为目标事件,当父分支中的最小记录数2%,子分支的最小记录数1%时停止,并且使用Bonferroni法调整重要值,分割及合并的显著性水平均以0.05为准,根据CHAID模型各项预测变量的重要性制定评分表.使用该评分模型计算所有纳入患者的评分,并绘制该模型预测AAD患者住院期间是否死亡的受试者工作特征(ROC)曲线.结果两组患者Stanford A型比例、住院期间接受手术的比例、血白细胞计数(WBC)、中性粒细胞、中性粒细胞/淋巴细胞比值(NLR)、血小板计数、间接胆红素、谷丙转氨酶、谷草转氨酶、尿素、血肌酐、尿酸、乳酸脱氢酶、凝血酶原时间、国际标准化比率(INR)、凝血酶时间、纤维蛋白原及D-二聚体比较差异有统计学意义(P<0.05).最终纳入CHAID模型的指标有血肌酐、血白细胞计数、纤维蛋白原、淋巴细胞、凝血酶时间、Stanford分型,该模型平均正确性0.909.预警评分模型的ROC曲线下面积(AUC)为0.894,95%CI0.849~0.938,P<0.001,根据约登指数计算出最佳临界值为8.5分,此时敏感度0.927,特异度0.671,该评分模型具有较高的预测准确性.结论本研究利用急诊室可快速获取的指标建立的预警评分模型对AAD患者住院期间病死率有较好的预测作用.
Objective To establish a scoring model by using the clinical indicators measured in the emergency room to predict the death of patients during hospitalization,and to evaluate the value of the scoring model in predicting the mortality of acute aortic dissection(AAD) patients during hospitalization.Methods A retrospective analysis of the data of 208 patients with AAD in the emergency room of the First Affiliated Hospital of Soochow University from January 1,2016 to January 1,2019,including vital signs on admission and blood routine,liver and kidney function,coagulation function and other indicators.The patients were divided into survival group(n=167) and death group(n=41),and the comparisons between the two groups were made according to the type of data.The CHAID algorithm in SPSS Modeler 18.0 was used to build a model,statistically different indicators between the two groups were incorporated into the model,and the death during hospitalization was taken as the target event.When the minimum number of records in the parent branch was less than or equal to2%,the child branch stopped when the minimum number of records was less than or equal to 1 %,and the Bonferroni method was used to adjust the important value.The significance level of segmentation and merger was based on 0.05,and the score table was formulated according to the importance of the predictive variables of the CHAID model.The scoring model was used to calculate the scores of all included patients,and the ROC curve of the model was drawn to predict whether AAD patients would die during hospitalization.Results There were significant statistical differences in the proportion of Stanford type A,the proportion of surgery during hospitalization,white blood cells(WBC),neutrophils,NLR,platelet count,indirect bilirubin,alanine aminotransferase,aspartate aminotransferase,urea,creatinine,uric acid,lactate dehydrogenase,prothrombin time,INR,thrombin time,fibrinogen and D-dimer in the patients between the two groups(P <0.05).The final indicators included in the CHAID model were creatinine,white blood cells,fibrinogen,lymphocytes,thrombin time,and Stanford typing.The average accuracy of the model was 0.909.The area under the ROC curve of the early warning scoring model was 0.894,95% CI 0.849-0.938(P <0.001),the optimal cut-off value calculated based on the Youden index was 8.5 points,at this time the sensitivity was 0.927,and the specificity was0.671.The model had high prediction accuracy.Conclusion The early warning scoring model established in this study by using indicators that can be quickly obtained in the emergency department has a good predictive effect on the mortality of AAD patients during hospitalization.
作者
李末寒
陈一欢
吴云
陆士奇
Li Mo-han;Chen Yi-huan;Wu Yun;Lu Shi-qi(Department of Critical Care Medicine,the First Affiliated Hospital of Soochow University,Suzhou 215006,China)
出处
《中国急救医学》
CAS
CSCD
北大核心
2020年第11期1065-1071,共7页
Chinese Journal of Critical Care Medicine