Risk Factor Prediction Model for Catheter-Associated Bloodstream Infections (CABSIs) in Midline and Central Venous...

Background: Venous catheter placement is the most common invasive procedure performed in hospitals. Despite their widespread use and importance in healthcare, these devices can cause complications such as catheter-associated bloodstream infections (CABSIs). Although several studies have investigated potential risk factors, including sociodemographic, medical history, and clinical variables, the results remain inconsistent and inconclusive. Objectives: The aim of this study was to identify independent risk factors for CABSIs and to develop and validate a predictive model for CABSIs in patients with midline catheters, centrally inserted central catheters (CICCs), and peripherally inserted central catheters (PICCs). Methods: We conducted an observational cohort follow-up study including hospitalized patients with a CICC, PICC, or midline catheter between January 2016 and March 2022. Devices were randomly assigned to derivation (n = 6036) and validation (n = 1549) cohorts. Candidate predictors with p < 0.25 in univariate analysis entered a multivariable logistic regression model, and final variables were selected by backward stepwise regression. Performance in the validation cohort was assessed by calibration and discrimination using the Hosmer-Lemeshow test and AUC. Results: The prevalence of CABSIs in the derivation cohort was 1.8%. Independent risk factors for CABSIs included tracheostomy, a history of bacteremia within 3 months before catheter placement, the presence of a synchronous central catheter, active oncohematological disease, and having received total parenteral nutrition (TPN). The presence of these five variables increased the probability of CABSIs to 42.1%. The final model demonstrated good predictive performance with an area under the curve (AUC) of 0.73 in the derivation cohort and 0.77 in the validation cohort. Decision curve analysis showed that the predictive model offered a greater net clinical benefit than the "treat-all" or "treat-none" strategies among threshold probabilities between 0.5% and 5%. Conclusions: The model can help identify high-risk patients, guide risk-based clinical decisions, reduce unnecessary catheter use, and support infection prevention and antimicrobial stewardship.