Independent review team versus standard infection preventionalist for CLABSI diagnosis: a systematic review of diagnostic error

Emily Larsen 1,2, Nicole Gavin 2, Nicole Marsh 1,2, Joan Webster 1,2

Griffith University, Nathan, QLD, Australia

Royal Brisbane and Women’s Hospital, Herston, QLD, Australia


Central venous access devices are frequently associated with hospital acquired infections known as central line-associated bloodstream infections (CLA-BSI). Classification of CLA-BSI is usually the responsibility of infection preventionalists (IPs) or other infectious diseases professionals, and is based on standardised definitions. Application of these definitions is sometimes inaccurate based on subjective clinical decision-making. The objective of this systematic review was to compare hospital reported CLA-BSI rates with independent review team classification.

MEDLINE, CINAHL, Scopus and ProQuest were searched for eligible articles. Full-text studies comparing the diagnostic validity of historically reported bacteraemia episodes using NHSN CLA-BSI definitions with validation team data were included.

87 studies were identified in the initial search, and 8 papers including 6754 patient records met the eligibility criteria. Of the total sample, validation teams found N=5646 true negatives; N=701 true positives; N=124 false positives and N=283 false negatives. Sensitivity (95% CI) ranged from 0.42 (0.15, 0.72) to 0.88 (0.77, 0.95) and specificity (95% CI) from 0.70 (0.58, 0.81) to 0.99 (0.99, 1.0). This indicates that clinicians are more likely to under-report than over-report. The overall risk of study bias was low using the Quality Assessment of Diagnostic Accuracy Studies 2 tool.

Validation of publicly reported data is essential in order to facilitate benchmarking between facilities, as well as evaluate the success of interventions to reduce the incidence of CLA-BSI. Continuous work is required to ensure classification reliability and further studies are needed to gain true understanding of CLA-BSI reporting bias.

Recent Comments