摘要
目的建立老年肿瘤患者经外周静脉置入中心静脉导管(PICC)相关深静脉血栓形成的决策树模型。方法选取2017年3月至2021年5月某医院行PICC的400例老年肿瘤患者作为模型组,2021年6月至2023年2月该医院行PICC的120例老年肿瘤患者作为验证组。采用logistic回归方法筛选老年肿瘤患者PICC相关深静脉血栓形成的危险因素,采用SPSS Modeler软件构建老年肿瘤患者PICC相关深静脉血栓形成的决策树模型,并分析该模型的预测效能。结果400例老年肿瘤患者中有74例发生PICC相关深静脉血栓,深静脉血栓的发生率为18.50%。logistic回归分析结果显示,体质量指数、穿刺次数、导管留置时间、糖尿病及慢性肾功能不全是老年肿瘤患者PICC相关深静脉血栓形成的危险因素(P<0.05)。老年肿瘤患者PICC相关深静脉血栓形成决策树模型的分类节点为糖尿病、导管留置时间、慢性肾功能不全、穿刺次数及体质量指数5个临床特征,其中糖尿病是最重要的预测因子。决策树模型的AUC(AUC=0.749,95%CI:0.688~0.811)高于logistic回归模型(AUC=0.701,95%CI:0.633~0.770)(P<0.05),验证组的AUC为0.812(95%CI:0.783~0.841)。结论体质量指数、穿刺次数、导管留置时间、糖尿病及慢性肾功能不全是老年肿瘤患者PICC相关深静脉血栓形成的危险因素,本研究构建的老年肿瘤患者PICC相关深静脉血栓形成决策树模型具有较高的准确性。
Objective To establish a decision tree model of deep vein thrombosis associated with peripheral inserted central catheters(PICC)in elderly patients.Methods A total of 400 elderly cancer patients receiving PICC in a hospital from March 2017 to May 2021 were selected as the model group,and 120 elderly cancer patients receiving PICC in a hospital from June 2021 to February 2023 were selected as the validation group.Logistic regression was used to screen the risk factors of PICC-related deep vein thrombosis in elderly cancer patients.SPSS Modeler software was used to build a decision tree model of PICC-related deep vein thrombosis in elderly cancer patients,and the prediction efficiency of the decision tree model was analyzed.Results Among 400 elderly tumor patients,74 cases developed PICC-associated DVT,and the incidence of DVT was 18.50%.Logistic regression analysis showed that body mass index,puncture times,catheter retention time,diabetes and chronic renal insufficiency were the risk factors for PICC-related deep vein thrombosis in elderly tumor patients(P<0.05).The classification nodes of the decision tree model of PICC-related deep vein thrombosis in elderly tumor patients were diabetes,catheter retention time,chronic renal insufficiency,puncture times and body mass index,among which diabetes was the most important predictor.The AUC of the decision tree model(AUC=0.749,95%CI:0.688-0.811)was higher than that of the logistic regression model(AUC=0.701,95%CI:0.633-0.770)(P<0.05),and that of the verification group was 0.812(95%CI:0.783-0.841).Conclusion Body mass index,puncture times,catheter retention time,diabetes and chronic renal insufficiency are risk factors for PICC-related DVT in elderly cancer patients.The decision tree model of PICC-related DVT in elderly cancer patients established in this study has high accuracy.
作者
杜梦迪
丁娟英
Du Mengdi;Ding Juanying(The First People′s Hospital of Linping District,Hangzhou 311100,China)
出处
《中国医院统计》
2024年第2期81-86,共6页
Chinese Journal of Hospital Statistics
基金
浙江省药学会医院药学专项科研资助项目(2020ZKY08)。
