BACKGROUND : Accurate, fully automated systems may increase efficiency of healthcare-associated infections (HAI) surveillance.AIMWe aimed to validate the performance of a fully automated surveillance system for central-line-associated bloodstream infections (CLABSI) in critically ill patients in Switzerland.

METHODS : We conducted a multicentre retrospective study across six secondary and tertiary care hospital networks' intensive care units (ICU). A centrally hosted, fully automated algorithm was implemented to detect catheter-related bloodstream infections (CRBSI), CLABSI and ICU-onset bloodstream infections (ICU-BSI). Algorithm performance was validated against an anonymised manual review of random samples of positive blood cultures. Incidence data were computed for each hospital.

RESULTS : From January 2022 to December 2023, we analysed 131,166 patient days, 108,719 catheter days and 7,832 positive blood cultures from 1,931 ICU patients. Median age was 65 years (interquartile range (IQR): 53-73), 458 (23.7%) were female. For CLABSI and CRBSI, the algorithm demonstrated a specificity of 95.3% (95% confidence interval (CI): 92.7-97.0), sensitivity of 86.5% (95% CI: 79.8-91.2), positive predictive value of 87.0% (95% CI: 80.4-91.7) and negative predictive value of 95.1% (95% CI: 92.5-96.8). CRBSI/CLABSI and ICU-BSI incidence rates were 3.23/1,000 catheter days (95% CI: 2.91-3.57) and 2.42/1,000 patient days (95% CI: 2.17-2.70), respectively. Most identified microorganisms for CRBSI/CLABSI were Staphylococcus epidermidis (15.1%; 53/351), Enterococcus faecium (9.1%; 32/351) and E. faecalis (5.7%; 20/351).

CONCLUSIONS : We demonstrate feasibility and external validity of a fully automated system for CLABSI surveillance in critically ill patients, supporting its integration into national HAI prevention and control strategies.