胃癌化疗患者PICC导管相关性感染易感因素及预测模型的构建

Susceptible factors for PICC-related infection in gastric cancer undergoing chemotherapy and establishment of prediction model

目的探讨胃癌化疗患者经外周静脉中心静脉导管(PICC)置管感染易感因素,建立预测风险模型,为胃癌化疗PICC置管患者感染的预测提供参考。方法选取杭州市临安区第一人民医院2020年1月-2021年12月301例胃癌化疗PICC置管患者,发生导管相关性感染63例纳入感染组,未发生感染的患者238例纳入未感染组,通过Logistic回归分析感染影响因素,构建风险预测模型。另选取同期患者100例建立验证组,对建模组数据进行验证,通过受试者工作特征曲线(ROC)评估模型区分度。结果63例感染患者,培养分离得到病原菌77株,其中革兰阳性菌占58.44%;Logisitic回归分析显示年龄≥60岁、合并糖尿病、穿刺次数≥3次、导管留置时间≥3个月、化疗次数≥6次是胃癌化疗PICC置管患者导管相关性感染发生的独立影响因素;胃癌化疗PICC置管患者发生置管相关性感染概率值回归方程为:P=1/[1+e-^((-3.484+0.593×年龄+1.008×糖尿病+1.103×穿刺次数+2.008×导管留置时间+1.224×化疗次数))],采用Hosmer-Lemeshow检验检测回归方程的拟合优度(P=0.244);以验证组数据对模型进行外部验证,100例患者中发生PICC置管导管相关性感染者14例,模型评估ROC曲线下面积(AUC)为0.824[95%CI(0.706~0.942)],灵敏度为85.71%,特异度为79.07%。结论建立的风险评分模型具有较好的判别效度,可有效将高感染风险和低感染风险的患者区分开,有利于针对性采取预防措施。

OBJECTIVE To explore the susceptibility factors for peripherally inserted central catheter(PICC)catheterization infection in patients undergoing gastric cancer chemotherapy,and to construct the predictive risk model providing reference for infection prediction.METHODS A total of 301 patients undergoing gastric cancer chemotherapy and PICC catheterization in the First People′s Hospital of Lin′an District in Hangzhou between Jan 2020 and Dec 2021 were enrolled,including 63 cases with catheter-related infection in the infection group and 238 cases without infection in the non-infection group.The risk factors for infection were analyzed by logistic regression analysis,and the prediction model was constructed.Another 100 patients during the same period were enrolled as the validation group,which was used to validate the data in the modeling group.Model differentiation was evaluated by receiver operating characteristics(ROC)curves.RESULTS In the 63 patients with infection,77 strains of pathogens were isolated,gram-positive bacteria accounting for 58.44%.Multivariate logisitic regression analysis showed that age≥60 years,diabetes mellitus,puncture≥3 times,catheter indwelling time≥3 months and chemotherapy≥6 times were independent risk factors for catheter-related infection.The regression equation for infection was as the following:P=1/[1+e-^((-3.484+0.593×age+1.008×diabetes mellitus+1.103×puncture times+2.008×catheter indwelling time+1.224×chemotherapy times))].Hosmer-Lemeshow test was used to detect the goodness of fit of the regression equation(P=0.244).The model was internally validated by the data in the validation group.In the 100 patients,14 cases with PICC occurred with catheterization related infection.The area under the ROC curve(AUC),sensitivity and specificity of the model were 0.824[95%CI(0.706-0.942)],85.71%and 79.07%,respectively.CONCLUSION The constructed risk scoring model has good discriminant validity,which can effectively distinguish patients with high-risk and low-risk infection and facilitate targeted preventive measures.