目的 以决策树法构建深静脉置管（deep venous cetheterization ,DVC）并发导管相关性感染的风险预测模型，为医护人员制订相关防控策略提供依据。方法 2019年1月至2021年12月，采用便利抽样法选取深圳市某三级综合性医院就诊的DVC患者2202例为研究对象，根据有未发生导管相关性感染将其分成感染组与非感染组，以二元Logistic回归分析法与决策树法分别创建DVC并发导管相关性感染的预测模型，并评估上述两种模型的风险预测效果。结果 年龄≥65岁、糖尿病、长期置管、重复置管、白蛋白＜30 g/L等均是DVC患者发生导管相关性感染的独立危险因素（P＜0.05）；决策树模型中，重复置管是DVC患者并发导管相关性感染最重要的预测因子；DVC患者并发导管相关性感染决策树模型的受试者特征曲线下面积（area under curve，AUC）值为0.808（95%CI：0.758~0.858）。结论 构建的DVC患者并发导管相关性感染的决策树模型的应用价值较高，可作为医护人员筛选潜在的导管相关性感染患者的依据。

Objective To construct a risk prediction model of deep venous catheterization (DVC) complicating catheter-associated infection based on decision tree model,and to provide reference for medical staff to formulate relevant prevention and control strategies.Methods A total of 2202 patients with DVC who were treated in tertiary hospital from January 2019 to December 2021 were selected by the convivence sampling method,and were divided into infected group and non-infected group according to whether they had catheter-associated infection.Binary Logistic regression analysis and decision tree method were used to establish prediction models for catheter-associated infection after DVC,and the risk prediction effects of the two models were evaluated.Results Age ≥65 years old,diabetes,long-term catheterization,repeated catheterization and albumin ＜ 30 g/L were independent risk factors for catheter-related infection in patients with DVC (P＜0.05).In the decision tree model,repeated catheterization was the most important predictor of catheter-associated infection in patients with DVC.The AUC value of the decision tree model for patients with catheter-associated infection with DVC was 0.808 (95%CI:0.758-0.858).Conclusion The decision tree model of catheter-associated infection in patients with DVC is of high application value and can be used as a basis for medical staff to screen potential patients with catheter-associated infection.

R473.54