摘要
目的探讨烧伤患者发生深静脉血栓形成(DVT)的危险因素,并构建烧伤患者发生DVT的决策树模型。方法选取2019年2月至2023年2月进入我院进行治疗的390例烧伤患者作为研究对象,根据DVT的发生情况将患者分为DVT组和无DVT组。采用Logistic回归分析筛选烧伤患者发生DVT的危险因素,采用SPSS Modeler软件构建烧伤患者发生DVT的决策树模型,并分析烧伤患者发生DVT的决策树模型的预测效能。结果390例烧伤患者中有63例患者发生DVT,DVT的发生率为16.15%(63/390)。Logistic回归分析结果显示,年龄≥60岁、下肢烧伤、烧伤总面积≥30%TBSA、糖尿病、卧床时间及创面感染等是烧伤患者发生DVT的危险因素(P<0.05)。烧伤患者发生DVT的决策树模型选择了下肢烧伤、烧伤总面积、糖尿病、卧床时间及创面感染等5个临床特征作为模型的节点,其中下肢烧伤是最重要的预测因子。烧伤患者发生DVT的决策树模型的AUC是0.738(95%CI 0.668~0.807),烧伤患者发生DVT的Logistic回归模型的AUC是0.754(95%CI 0.685~0.822),决策树模型的AUC与Logistic回归模型的AUC差异无统计学意义(P>0.05)。结论年龄≥60岁、下肢烧伤、烧伤总面积≥30%TBSA、糖尿病、卧床时间及创面感染等是烧伤患者发生DVT的危险因素,本研究构建的烧伤患者发生DVT的决策树模型具有较高的准确性。
Objective To investigate the risk factors of deep vein thrombosis(DVT)in burn patients,and construct a decision tree model of DVT in burn patients.Methods A total of 390 burn patients admitted to our hospital for treatment from February 2019 to February 2023 were selected as the study objects,and the patients were divided into DVT group and non-DVT group according to the occurrence of DVT.Logistic regression analysis was used to screen the risk factors for DVT in burn patients,and SPSS Modeler software was used to build a decision tree model for DVT in burn patients,and the prediction efficiency of the decision tree model for DVT in burn patients was analyzed.Results Among 390 burn patients,63 patients developed DVT,the incidence of DVT was 16.15%(63/390).Logistic regression analysis showed that age≥60 years old,lower limb burn,total burn area≥30%TBSA,diabetes,bed time and wound infection were risk factors for DVT in burn patients(P<0.05).The decision tree model of DVT in burn patients selected five clinical features as nodes of the model,including lower limb burn,total burn area,diabetes,bed time and wound infection,among which lower limb burn was the most important predictor.The AUC of the decision tree model for DVT in burn patients was 0.738(95%CI 0.668~0.807),and the AUC of the Logistic regression model for DVT in burn patients was 0.754(95%CI 0.685~0.822),there was no significant difference between AUC of decision tree model and Logistic regression model(P>0.05).Conclusion Age≥60 years old,lower limb burn,total burn area≥30%TBSA,diabetes,bed time and wound infection are risk factors for DVT in burn patients.The decision tree model of DVT in burn patients constructed in this study has high accuracy for clinic procedures.
作者
王献珍
吴晓伟
张科伟
Wang Xianzhen;Wu Xiaowei;Zhang Kewei(Department of Burn and Plastic Surgery,Affiliated Hospital of Qinghai University,Xining,Qinghai 810001,China)
出处
《四川医学》
CAS
2023年第12期1275-1279,共5页
Sichuan Medical Journal
