Objectives: To compare the safety of a conventional polyurethane transparent dressing and a novel highly permeable polyurethane dressing, as compared with standard gauze and tape, as site dressings for pulmonary artery catheters; and to rigorously determine the sources of bloodstream infections deriving from these catheters.
Design: Prospective, randomized, clinical trial.
Setting: General adult intensive care units (ICUs) in a university hospital.
Patients: A total of 442 adult patients with pulmonary artery catheters were studied. Two thirds of the catheters had been inserted in the operating room and one third had been inserted in an ICU.
Interventions: Patients were randomized at the time of pulmonary artery catheter insertion to have one of three dressing regimens: a) sterile gauze and tape (control), replaced every 2 days; b) a conventional polyurethane dressing, replaced every 5 days; or c) a highly permeable polyurethane dressing, also replaced every 5 days.
Measurements and main results: The origin of each catheter-associated bloodstream infection was sought by quantitatively culturing the skin of the insertion site and all potential sources on the catheter, including the hub and infusate from each lumen of the introducer sheath and the pulmonary artery catheter, and intravascular segments of the introducer sheath and pulmonary artery catheter. Bloodstream infection was confirmed by demonstrating concordance between isolates from the device and blood cultures by pulsed-field electrophoresis of genomic DNA, digested with low-frequency-cleavage, restriction endonucleases. One hundred thirty catheters were randomized to be dressed with sterile gauze and tape (control), 127 with the conventional polyurethane dressing, and 185 with the highly permeable polyurethane dressing. Patients and catheters in the three dressing groups were very comparable. Ninety-six (21.7%) of the 442 catheters studied showed colonization of the introducer sheath or the pulmonary artery catheter, and five (1.1%) catheters caused bloodstream infection. Catheter-related bloodstream infections were associated with concordant cutaneous colonization of the insertion site (n = 2), a contaminated catheter hub or infusate (n = 3), contamination of the extravascular segment of a repositioned catheter beneath the external protective plastic sleeve (n = 1), or hematogenous colonization of the catheter (n = 1). All pulmonary artery catheter-related bloodstream infections occurred with catheters (introducers) in place for > or = 5 days (p < .001). Cutaneous colonization under the dressing at catheter removal was lowest with gauze (10(1.3) colony-forming units), intermediate with the new highly permeable polyurethane dressing (10(1.8) colony-forming units; p < .01), and highest with the conventional polyurethane dressing (10(2.0) colony-forming units; p < .001). There were no significant differences in catheter colonization (20.0 to 25.2 cases per 100 catheters) or catheter-related bloodstream infection (0.8 to 1.6 cases per 100 catheters) between the three groups.
Conclusions: The incidence of pulmonary artery catheter-related bloodstream infection has decreased over the past 5 yrs. Pulmonary artery catheter-related bloodstream infections originate from multiple sources, indicating that measures to prevent bacteremic infections of these devices must focus both on reducing cutaneous colonization at the insertion site and averting contamination of infusate and catheter hubs. Efforts should be made to limit the duration of catheterization with pulmonary artery catheters (including the introducer) to no longer than 4 days. The polyurethane dressings studied appear to be safe for use with pulmonary artery catheters and may be left on for up to 5 days between dressing changes.