Objective: Patients undergoing heart surgery show a high risk of catheter colonization and catheter-related bloodstream infections. We evaluated whether skin insertion site and catheter hub surveillance cultures ("surface cultures") could predict catheter colonization and help establish the origin of bloodstream infections.
Design: : Prospective cohort study.
Setting: An 11-bed heart surgery intensive care unit in a tertiary university hospital.
Patients: Heart surgery patients spending >4 days in intensive care over an 11-month period.
Interventions: All catheters were surveyed. Cultures were obtained from the skin insertion site and all hubs on day 5 after surgery, every 72 hrs thereafter, and on catheter removal. Swabs were processed semiquantitatively by streaking the surface of a Columbia agar plate. Catheters were processed using Maki's method. The observation of > or = 15 colonies/plate was taken to indicate a positive skin or catheter colonization culture result.
Measurements and main results: Over the study period, 561 catheters were inserted in 130 patients. The median time a catheter was in place was 6 days (interquartile range 3-11), and 3,712 surface cultures were obtained (median four per patient). Catheter colonization occurred in 133 catheters, and there were 15 episodes of catheter-related bloodstream infection (incidence density of colonization 29.3 and of catheter-related bloodstream infection 8.8 per 1,000 catheter-days). Validity indexes for the capacity of surface cultures to predict catheter colonization and catheter-related bloodstream infection, respectively, were as follows: accuracy, 71.4, 65.6; sensitivity, 83.5%, 100%; specificity, 67.1%, 64.7%; positive predictive value, 47.6%, 7.2%; negative predictive value, 91.9%, 100%; positive likelihood ratio, 2.5, 2.83; and negative likelihood ratio, 0.2, 0. Surface cultures correctly predicted 77.4% of all bacteremia episodes (catheter-related and non-catheter-related).
Conclusions: Systematic surveillance cultures of catheter hub and skin insertion sites in patients admitted to a heart surgery intensive care unit could help identify patients who would benefit from decontamination and preventive measures and establish whether catheters are the portal of entry of bloodstream infection.