Risk factors of catheter-related bloodstream infection in ICU patients and construction of risk prediction nomogram model
Objective To analyze the risk factors of catheter-related bloodstream infection(CRBSI)in intensive care unit(ICU)patients,and construct a risk prediction nomogram model and validate it.Methods The case data of patients who underwent central venous catheter(CVC)in ICU of Inner Mongolia People's Hospital from January 1,2020 to April 30,2023 were collected,and the modeling group and validation group were established according to the EPV method.In the modeling group,there were 408 males and 252 females,aged 66.5(54.0,76.0)years,with the duration of catheter retention of 10(5,17)days and the ICU stay of 12(6,22)days.In the validation group,there were 114 males and 65 females,aged 66.0(55.0,75.0)years,with the duration of catheter retention of 10(5,16)days and the ICU stay of 10(5,22)days.Multivariate logistic regression analysis was used to analyze the risk factors of CRBSI in ICU patients of the modeling group,and the risk prediction nomogram model was drawn by R4.3.0 software.The model was verified internally and externally.The receiver operating characteristic curve(ROC),Calibration curve,and Hosmer-Lemeshow goodness of fit test were used to evaluate the discrimination,calibration,and fitting degree of the model.Non-parametric Wilcoxon test and x2 test were used.Results Diabetes mellitus,hypoproteinemia,times of intubation ≥2,duration of catheter retention,and duration of antibiotic use were independent risk factors for CRBSI in ICU patients(all P＜0.05).Hosmer-Lemeshow goodness of fit test results showed that the nomogram model fit well in the modeling group(x2=3.466,P=0.902)and the validation group(x2=1.723,P=0.988).Calibration curve analysis showed that the incidence of CRBSI predicted by the nomogram model was basically consistent with the actual incidence of CRBSI in the modeling group and the validation group.ROC analysis results showed that the area under the curve(AUC)of the nomogram model for predicting CRBSI was 0.872[95％CI(0.824-0.920)]in the modeling group and 0.937[95％CI(0.890-0.984)]in the validation group,respectively.Conclusion The established nomogram model for predicting risk of CRBSI had good prediction value and application value,and could identify the high risk group of CRBSI in ICU patients effectively.
Intensive care unitCentral venous catheterCatheter-related bloodstream infectionRisk factorsRisk prediction
万方数据
维普
