Randomized, Double-Blind Trial of an Antibiotic-Lock Technique for Prevention of Gram-Positive Central Venous Catheter-Related Infection in Neutropenic Patients with Cancer

Setting and study population.The study was conducted at the hematology ward of a university hospital for adults in Barcelona, Spain. From April 1994 through March 1996, hospitalized patients with a nontunneled, multilumen, polyurethane CVC (Arrow International, Inc., Reading, Pa.) in place and who were to receive chemotherapy designed to produce severe neutropenia (<500 neutrophils per μl) were eligible for participation in the trial. Patients were excluded if they had clinical or microbiologic evidence of infection or had a known allergy to vancomycin. Patients already receiving antibiotics or parenteral nutrition were also excluded. Participants gave informed consent, and the study was approved by the Ethics Committee of our institution.

Catheter insertion and care.Catheters were inserted into the subclavian vein by physicians who wore masks, caps, sterile gloves, and surgical gowns and who used large sterile drapes. Study catheters were not exchanged over guidewires. At the time of catheter insertion, the skin insertion site was disinfected with 4% chlorhexidine gluconate (Hibiscrub; ICI Farma, Pontevedra, Spain), which was applied by scrubbing for at least 30 s. The insertion sites were covered with sterile gauze.

Catheter care included changing of the dressing, stock-cocks, and tubes every 48 h by registered nurses, who followed maximal barrier precautions. The insertion site was softly scrubbed with sterile gauze saturated with normal saline and 4% chlorhexidine gluconate for at least 30 s, rinsed with normal saline, and dried. Catheter lumens were assessed for occlusion and were flushed with 10 ml of 0.9% sodium chloride. In case of occlusion the lumen was flushed with either heparin or urokinase by a soft instillation-aspiration technique to reestablish patency. The insertion site and catheter hubs were protected with 1% chlorhexidine pomade (Hibitane crema; ICI Farma), dressed, and taped securely.

Study design.Patients were randomized, on the day after completion of the corresponding chemotherapy course, to receive in a double-blind fashion either a solution containing heparin (Rovi, SA, Madrid, Spain) at 10 U/ml or heparin at 10 U/ml and vancomycin (Lilly, Madrid, Spain) at 25 μg/ml. For study solution allocation a computer-generated list of random numbers, which was available only to the pharmacist, was used. The stability of the heparin and vancomycin solution has been demonstrated previously (8). The hospital pharmacy prepared the solutions on a horizontal-airflow workbench using aseptic technique by a previously described procedure (25). The two solutions were indistinguishable to medical personnel and were dispensed in 20-ml vials that were numerically coded and that were kept refrigerated. The code list was kept in the hospital pharmacy and was opened only after the study was completed. The neutrophil count was determined daily, and when it was below 500 per μl the administration of the allocated solution was initiated. A 2.5-ml injection of the solution was administered through the stock-cocks in each catheter lumen and was allowed to dwell for approximately 1 h every 2 days. In all cases, the solution was then aspirated with a syringe and was discarded.

Swabs of the skin insertion site and the inner surfaces of the catheter hubs were obtained before randomization and twice weekly thereafter for culture. Swabs of the insertion site and catheter hub as well as blood specimens from a peripheral vein were also obtained for culture before the beginning of intravenous empirical antibiotic therapy for fever and neutropenia. Ceftazidime plus amikacin was the empirical antibiotic regimen most commonly used to treat febrile episodes that occurred during the study period. Vancomycin was added to the regimens of patients in whom infection with gram-positive bacteria was initially suspected and also those who had not improved after initial therapy for 48 h or who worsened before that time. In case of catheter removal, two 5-cm segments, a proximal subcutaneous segment and the tip, were sampled by a sterile technique for culture. Patients were monitored until any of the following criteria was present: recovery from neutropenia, administration of systemic vancomycin, colonization of the catheter hub, bacteremia caused by gram-positive bacteria, administration of parenteral nutrition, catheter removal, or death. The patients’ physicians and nurses, the clinical investigators, and the research microbiologists who processed all cultures were blinded to each study group.

Microbiologic studies.Specimens from the skin at the insertion site (an area of approximately 4 cm²) were obtained for culture with a moistened sterile swab (Biomedics, Madrid, Spain), as described previously (12). The catheter hub samples were taken by repeatedly rubbing the inner surface of the hub with a sterile cotton swab (Eurotubo; Industrias Aulabor, SA, Barcelona, Spain) before administration of the study solution. All samples were sent to the laboratory in transport medium and were plated onto blood agar plates. The outer surfaces of the catheter tips were cultured by the roll-plate technique and were rolled back and forth across the surfaces of blood agar plates at least four times, as described previously (14). The inner surface of the catheter tip was cultured by a modified quantitative method (11); 2 ml of Trypticase soy broth was passed across each lumen and was then diluted 10-fold, and each dilution was plated onto blood agar. Colony counting was carried out after incubation of the plates at 37°C in a 5% CO₂ atmosphere for 48 h. Skin and catheter hub cultures were considered positive when ≥15 CFU was isolated. The criteria for positivity of the catheter tip culture were counts of ≥10³ CFU by the quantitative method and counts of ≥15 CFU by the roll-plate method. Blood cultures were performed by routine methods (BACTEC NR 860 instrument; Johnston Laboratories, Inc., Towson, Md.). The bacteria were fully identified by conventional procedures and by an automated method (MicroScan; Baxter Healthcare, West Sacramento, Calif.). Antibiotic susceptibility studies were performed by the disk diffusion method and by the microdilution method.

Molecular typing.Molecular typing was performed by pulsed-field gel electrophoresis (PFGE) after DNA restriction withSmaI and separation of fragments in a contour-clamped homogeneous electric field (CHEF-DRII) apparatus (Bio-Rad, Richmond, Calif.) with running conditions of 200 V and pulse times ranging from 1 to 30 s for 23 h at 14°C (3).

Definitions.Significant colonization of the catheter hub was defined as the isolation of ≥15 CFU. Bacteremia attributed to luminal colonization was defined as the isolation of identical organisms from the catheter hub and from cultures of separate percutaneously drawn blood specimens; the organism’s identity was proven by molecular typing for patients with Staphylococcus epidermidis bacteremia. Catheter occlusion was defined as the inability to aspirate blood through the catheter and/or the inability to force infusate through the catheter with usual infusion pressure. Patients were considered to have a fever if the temperature was above 38°C.

Statistical analysis.To assess the adequacy of randomization, groups were compared by the uncorrected chi-square test or, when appropriate, Fisher’s exact test for categorical variables and the Mann-Whitney test for continuous variables. The distributions of time until catheter hub colonization and time until catheter-related bacteremia were stimated by the Kaplan-Meier method and were compared across the two groups by the log rank test (Mantel-Cox test). Crude and adjusted hazard ratios could not be estimated because no events were observed in the heparin and vancomycin group. However, since randomization produced two groups of patients with comparable baseline characteristics, no indication of positive or negative confounding needed to be controlled for with multivariate Cox models. Statistical significance was established at an alpha value of 0.05. AllP values are two-tailed.

The sample size of 116 patients was calculated to give an 80% power to detect a 20% difference in the rate of catheter hub colonization by the two treatments at a one-sided 0.05 level of significance. On the basis of our previous experience (11), it was estimated that about 25% of patients in the heparin group would develop catheter hub colonization. The effect of heparin and vancomycin treatment was thought to be clinically relevant if the catheter hub colonization rate was reduced to 5%. We performed an intention-to-treat analysis for all patients who received at least one dose of study solution.