Risk prediction models for central venous catheter-related complications in children: a systematic review.

backgroundCentral venous catheters (CVCs) are widely used in pediatric care; however, their indwelling procedure can lead to severe complications such as catheter-associated infections and thrombosis. These complications significantly increase the suffering of pediatric patients, prolong hospitalization, and pose life-threatening risks. Although multiple risk prediction models have been developed to identify high-risk pediatric patients, a systematic and comprehensive evaluation of these models is still lacking.

objectiveWe systematically reviewed published risk prediction models for central venous catheter-related complications ( CLABSI and CRT ) in children, both domestically and internationally. The review assessed their predictive performance, validation status, and methodological quality, aiming to provide a basis for the selection of appropriate models in clinical practice.

methodsFollowing the PRISMA guidelines, a systematic search was conducted in databases including PubMed, Web of Science, Embase, Cochrane Library, CINAHL, CNKI, Wanfang, VIP, and CBM for relevant literature from their inception to December 31, 2025. Two researchers independently screened the literature, extracted data, and assessed the risk of bias and applicability using the PROBAST tool. A descriptive analysis was performed on the model construction methods, performance metrics, and validation status.

resultsA total of 15 studies were included in this analysis. Among them, three studies developed prediction models for CLABSI, all constructed using logistic regression(LR), with modeling AUCs ranging from 0.793 to 0.812; only one model underwent external validation. Twelve studies developed prediction models for CRT, employing diverse modeling methods (such as LR and machine learning). The modeling AUCs ranged from 0.74 to 0.957, of which four models underwent external validation (AUC 0.745–0.922). The PROBAST assessment indicated that most studies had a high risk of bias, primarily due to issues such as single-center retrospective design, highly subjective predictive factors, and inconsistent outcome evaluation criteria. The applicability of the models was generally satisfactory.

conclusionExisting risk prediction models for central venous catheter-related complications in children demonstrate potential in terms of discrimination. However, models specifically for CLABSI are limited in number and insufficiently validated. Both types of models commonly exhibit methodological flaws such as a high risk of bias. Currently, these models primarily serve as exploratory tools for risk screening and cannot yet be used as standalone evidence for clinical decision-making. Future efforts should focus on conducting multicenter prospective studies, standardizing diagnostic criteria and model development/validation procedures, and promoting the integration of models into clinical systems to enhance their reliability and clinical applicability.