Central venous catheter infections: building a causal model with expert domain knowledge to inform future clinical trials.

Abstract

Aim: Central venous catheters (CVCs) are essential for long-term therapies but carry a high risk of central line-associated bloodstream infections (CLABSIs), which significantly impact patient outcomes and healthcare costs. This study aimed to develop a causal model for CLABSI using expert knowledge to guide future clinical trials and prevention strategies.

Methods: We constructed a directed acyclic graph (DAG) informed by literature and expert knowledge elicitation. A multidisciplinary team of clinicians, including infectious disease and vascular access experts, participated in interviews and workshops to refine the DAG, resulting in a final model with 30 variables representing CLABSI development.

Findings: The expert-elicited DAG identified two main pathways, patient-related and CVC-related, each contributing to CLABSI risk. Variables and relationships in the DAG highlighted key patient characteristics, CVC management practices, and overlapping factors influencing infection. This model serves as a novel framework to understand CLABSI causation and supports trial design by identifying confounding factors, causal pathways, and meaningful endpoints.

Conclusions/implications: Our causal DAG provides a structured representation of CLABSI risk factors, which may support the design of clinical trials examining interventions to reduce CVC-related infections. By clarifying causal mechanisms, the DAG can enhance the specificity of endpoints and improve the rigor of prevention strategies.