预测成人钝性脾损伤患者手术干预的列线图模型构建  

Establishment of a clinical nomogram model for predicting operative management in adults with blunt splenic injury

作　　者：宁辰[1] 虞俊波[1] 朱东波[1] NING Chen;YU Jun-bo;ZHU Dong-bo(Department of Emergency Surgery,Affiliated Hospital of Nantong University,Nantong,Jiangsu226001,China)

机构地区：[1]南通大学附属医院急诊外科

出　　处：《创伤外科杂志》2019年第10期725-729,共5页Journal of Traumatic Surgery

摘　　要：目的构建成人钝性脾损伤患者手术干预的列线图(nomogram)模型。方法回顾分析2014年1月—2018年12月南通大学附属医院急诊外科收治的257例成人钝性脾损伤患者,其中男性180例,女性77例;年龄17~103岁,平均49.96岁。根据最终是否剖腹手术分为手术治疗组和非手术治疗组,采用单因素分析和多因素Logistic回归法分析影响手术干预的独立危险因素,根据独立危险因素建立预测成人钝性脾损伤患者手术干预的列线图模型,计算可决系数R 2、一致性指数、ROC曲线下面积,检测列线图模型的拟合优度和预测准确度。计算约登指数(Youden index),确定最佳界值。结果多因素Logistic回归分析发现,脾损伤分级AAST(OR:224.92)、腹腔积血≥1 000mL(OR:93.59)、输RBC≥4U(OR:5.52)是成人钝性脾损伤患者手术干预的独立危险因素,应用R软件绘制获得预测模型列线图,用Bootstrap法进行模型内部验证,显示其R 2为0.668(调整R 2 0.667),一致性指数为0.843(95%CI 0.814~0.872),AUC为0.963(95%CI 0.937~0.989)。最大约登指数为0.883,其对应的最佳界值为0.631。结论脾损伤分级AAST(OR:224.92)、腹腔积血≥1 000mL(OR:93.59)、输RBC≥4U(OR:5.52)是成人钝性脾损伤患者手术干预的独立危险因素,该模型对成人钝性脾损伤患者的临床诊疗决策具有一定指导意义。Objective To establish a clinical nomogram model for predicting operative management in adults with blunt splenic injury.Methods Totally 257 adults with blunt splenic injury(BSI)treated from Jan.2014 to Dec.2018 were reviewed retrospectively.There were 180 cases of males and 77 cases of females aged from 17 to 103 years,with an average age of 49.96 years.The patients were then divided into operative and no-operative groups based on whether they received operative management finally.Univariate and multivariate logistic analyses were performed to investigate correlative factors affecting the outcome of operative management in adults with BSI.Independent risk predictors for operative management were identified and the clinical nomogram for operative management in adults with BSI was established.The accuracy of nomogram model for predicting operative management in adults with BSI was detected by consistency coeffcient(C-index)and Area Under Curve(AUC),while the goodness of fit was detected by coefficient of determination(R2).The best cut-off was detected by Youden index.Results Multivariate logistic regression analysis revealed that higher-grade splenic injury[AAST(OR:224.92)],presence of haemoperitoneum(≥1000ml,OR:93.59),and the number of transfusions(≥4 units,OR:5.52)were independent predictors of operative management in adults with BSI.Finally a clinical nomogram model was established by R statistical software.The coefficient of determination was 0.668,the consistency coeffcient(C-index)was 0.843(95%CI 0.814-0.872)and the area under ROC curve was 0.963(95%CI 0.937-0.989).The Youden index was 0.883,and the best cut-off was 0.631.Conclusion Higher-grade splenic injury(AAST),presence of haemoperitoneum(≥1000mL),and the number of transfusions(≥4 units)were independent predictors of operative management in adults with BSI.And the model offers certain guiding significance for treatment strategy in adults with blunt splenic injury.

关 键 词：脾损伤 列线图 模型 成人 干预 

分 类 号：R657.6[医药卫生—外科学]