Risk factors and prediction model construction for peripherally inserted central catheter-related infections

Abstract

Objective: To establish a predictive modeling for the risk of bloodstream infection associated with peripherally inserted central catheter (PICC).

Methods: Patients receiving PICC treatment in Shenzhen People's Hospital from June 2020 to December 2020 were retrospectively enrolled and divided into the infection group and the non-infection group according to the presence and absence of PICC-related infections. Then, relevant clinical information of patients was collected and the predictors of PICC-related infection were screened by the least absolute shrinkage and selection operator regression (LASSO) model. Besides, multivariate logistic regression was used to analyze the influencing factors of PICC-related infection, A nomogram was constructed based on the results of the multivariate analysis. Ultimately, a receiver operating characteristic (ROC) curve was plotted to analyze the application value of influencing factors to predict PICC-related infections.

Results: A total of 505 patients were included, including 75 patients with PICC-related infections (14.85%). The main pathogen was gram-positive cocci. The predictors screened by LASSO included age >60 years, catheter movement, catheter maintenance cycle, insertion technique, immune function, complications, and body temperature ≥37.2 °C before PICC placement. Multivariate logistic regression analysis showed that independent risk factors of infections related to PICC included age >60 years [odds ratio (OR) = 1.722; 95% confidence interval (CI) = 1.312-3.579; P = 0.006], catheter movement (OR = 1.313; 95% CI = 1.119-3.240; P = 0.014), catheter maintenance cycle >7 days (OR = 2.199; 95% CI = 1.677-4.653; P = 0.000), direct insertion (OR = 1.036; 95% CI = 1.019-2.743; P = 0.000), poor immune function (OR = 2.322; 95% CI = 2.012-4.579; P = 0.000), complications (OR = 1.611; 95% CI = 1.133-3.454; P = 0.019), and body temperature ≥37.2 °C before PICC placement (OR = 1.713; 95% CI = 1.172-3.654; P = 0.012). Besides, the area under the ROC curve was 0.889.

Conclusion: PICC-related infections are associated with factors such as age >60 years, catheter movement, catheter maintenance cycle, insertion technique, immune function, complications, and body temperature ≥37.2 °C before PICC placement. Additionally, the LASSO model is moderately predictive for predicting the occurrence of PICC-related infections.

Keywords: Catheter-related infection; Peripherally inserted central catheter; Prediction model.

Fig. 1

Screening predictors of catheter-related infection after peripherally inserted central catheter placement by LASSO regression model A, Cross-validation results. The value between the two dotted lines was the range of positive and negative standard deviation of log (λ). The dotted line on the left indicated the value of harmonic parameter log (λ) when the model error was the smallest. When log (λ) = −5.1 was substituted, 7 variables were selected. B, LASSO coefficient profiles for 13 variables. Vertical lines were drawn at 10 times the value selected by cross-validation. With the decrease of λ value, both the compression degree of the model and the selections of the model for important variables increased. LASSO, least absolute shrinkage and selection operator.

Fig. 2

A nomogram of the risk of catheter-associated infections.

Fig. 3

Analysis of the receiver operating characteristic curve.