长期血液透析导管相关性血流感染风险预测模型构建

Construction of risk prediction model of catheter-related bloodstream infection for long-term hemodialysis patients

目的 调查长期血液透析导管相关性血流感染(CRBSI)情况,并构建Logistic风险预测模型。方法 回顾性分析2019年12月-2020年12月潍坊市益都中心医院收治的长期留置导管的180例血液透析患者临床资料,分析CRBSI感染情况及病原菌分布,单因素及多因素Logistic回归分析CRBSI危险因素,采用Logistic回归分析建立CRBSI感染风险预测模型,采用受试者工作特征(ROC)曲线,以曲线下面积(AUC)评估该模型的预测效果。结果 180例长期留置导管的血液透析患者CRBSI感染率为18.33%,感染病原菌以革兰阳性菌为主(63.04%),单因素和多因素Logistic回归分析结果显示,年龄>60岁、合并糖尿病、股静脉导管置管、透析时间≥3年、白蛋白水平<30 g/L是CRBSI感染的独立危险因素;长期留置导管的血液透析患者CRBSI感染的风险预测模型为P=1/1+e^([-(-4.496+1.018×(年龄)+0.824×(合并糖尿病)+1.181×(股静脉导管置管)+1.315×(透析时间≥3年)+0.697×(白蛋白水平<30g/L)]),预测CRBSI感染的AUC为0.887(95%CI:0.819~0.954)。结论 年龄、合并糖尿病、静脉导管置管部位、透析时间、白蛋白水平是长期留置导管血液透析患者CRBSI感染的危险因素,通过建立多因素Logistic风险预测模型可较准确预测CRBSI感染的发生,进而指导临床干预工作。

OBJECTIVE To investigate the prevalence of catheter-related bloodstream infection(CRBSI) in long-term hemodialysis patients and construct Logistic risk prediction model. METHODS A total of 180 long-term hemodialysis patients who were treated in Yidu Central Hospital of Weifang from Dec 2019 to Dec 2020 were enrolled in the study, the clinical data of the patients were retrospectively analyzed, the prevalence of CRBSI and distribution of pathogens were observed, univariate analysis and multivariate logistic regression analysis were performed for risk factors for CRBSI, the risk prediction model for the CRBSI was established based on the Logistic regression analysis, and the effect of the model on prediction of CRBSI was evaluated by area under curve(AUC) of receiver operating characteristic(ROC) curve. RESULTS Among the 180 long-term hemodialysis patients, the incidence rate of CRBSI was 18.33%. The gram-positive bacteria were dominant among the pathogens causing the infection, accounting for 63.04%. The results of univariate analysis and multivariate logistic regression analysis showed that the more than 60 years of age, complication with diabetes mellitus, femoral venous catheter indwelling, no less than 3 years of dialysis and albumin level less than 30 g/L were independent risk factors for the CRBSI. The risk prediction model for CRBSI in the long-term hemodialysis patients was P=1/1+e^[-(-4.496+1.018×(age)+0.824×(diabetes mellitus)+1.181×(femoral vein catheterization)+1.315×(dialysis time≥3 years)+0.697×(albumin level<30 g/L)]), and the AUC was 0.887 in prediction of CRBSI(95%CI:0.819-0.954). CONCLUSION The age, complication with diabetes mellitus, site of venous catheter indwelling, duration of dialysis and albumin level are the risk factors for CRBSI. The multivariate Logistic risk prediction model may accurately predict the CRBSI in the long-term hemodialysis patients so as to take clinical interventions.