Abstract: Objective: To systematically search and evaluate the risk prediction models for central venous Catheter-Related Bloodstream Infections, to provide reference for constructing a high-quality risk prediction model for central venous Catheter-Related Bloodstream Infections. Methods: We searched for relevant literature on central venous Catheter-Related Bloodstream Infections from Chinese and English databases from their inception to December 11, 2023. Two researchers independently conducted literature screening and data extraction, using PROBAST tool to evaluate the risk of bias and applicability of included studies. Results: A total of 16 articles were included, involving 15 risk prediction models for central venous Catheter-Related Bloodstream Infections. The AUC or C index of 15 models ranges from 0.67 to 0.930; 5 models received internal validation and 8 models external validation. Quality evaluation of the 16 articles showed high bias risk and low applicability risk. The predictive factors with higher frequency of occurrence included: parenteral nutrition (TPN), diabetes, duration of central venous catheterization, ICU patient or hospital history, number of lumens, use of antibiotics, femoral vein insertion, and malignant tumors. Conclusion: The existing risk prediction models for central venous Catheter-Related Bloodstream Infections have good predictive performance to determine the predictive factors which possess certain guiding significance for clinical practice, but the risk of bias is high. Further validation of future response models is needed to provide a basis for developing effective prevention strategies.

Key words: Central Venous Catheter; Catheter-Related Bloodstream Infections; prediction model; systematic review