A three year prospective study involved all patients admitted to the 24 bed ICU of Hospital Universitario de Canarias (Tenerife, Spain), between 1 May 2000 and 30 April 2003. The study was approved by the institutional review board.
The catheters used were not antimicrobial coated, but were radiopaque polyurethane catheters (Arrow, Reading, PA, USA). The placement and maintenance of catheters were performed according to the following protocol. The catheters were inserted by physicians with the following sterile-barrier precautions: use of large sterile drapes around the insertion site, surgical antiseptic hand wash, and sterile gown, gloves, mask and cap.
The skin insertion site was first disinfected with 10% povidone-iodine and anesthetized with 2% mepivacaine. The catheters were inserted percutaneously using the Seldinger technique and were fixed to the skin with 2–0 silk suture. After line insertion the area surrounding the catheter was cleaned with a sterile gauze soaked with povidone-iodine, and a dry sterile gauze occlusive dressing then covered the site. No topical antimicrobial ointment was applied to insertion sites.
The percutaneous entry sites were examined for the presence of local inflammation and purulence, and were cared for daily in the same manner by the ICU nurse assigned to the patient. Catheter dressings were changed every 24 hours, or sooner at the discretion of the nurse if the dressing was contaminated. The connecting lines were changed every 48 hours and disposable transducer components were replaced every 96 hours.
The percutaneous entry sites were also examined daily by the ICU nurse assigned to the patient to avoid accidental catheter removals [28
] and to minimize infection risk associated with reinsertion of the catheter.
The decision to remove the catheter was made by the patient's physician. Catheters were removed when they were no longer needed or if a systemic or local complication occurred. Arterial catheters were routinely replaced every seven days. The insertion site for the new catheter was changed when the catheters were replaced. All catheter tips removed were routinely cultured.
The catheters were removed using a sterile technique by the ICU nurse. The distal five cm segment of the catheters was cut with sterile scissors, placed in a sterile transport tube and cultured using the semiquantitative method described by Maki and colleagues [29
The following data were collected: age, sex, diagnosis, Acute Physiology and Chronic Health Evaluation (APACHE) II score, ICU admission and discharge dates, catheter site, catheter insertion and removal dates, cause of catheter removal, and development of CRLI and CRBSI. We studied the following four groups of arterial catheter sites: radial, femoral, dorsalis pedis and brachial.
Catheter-related infection was defined according to catheter-tip colonization, CRLI and CRBSI. We considered catheter-tip colonization as a significant growth of a microorganism (>15 colony-forming units) from the catheter tip. CRLI was considered as any sign of local infection (induration, erythema, heat, pain, purulent drainage) and catheter-tip colonization. CRBSI was considered a positive blood culture obtained from a peripheral vein, and signs of systemic infection (fever, chills, and/or hypotension), with no apparent source of bacteremia except for the catheter, and catheter-tip colonization with the same organism.
The age, APACHE II score and total duration with the index catheter are expressed as the mean ± standard deviation. The total duration of ICU stay is reported as the median (interquartile range). Categorical variables are expressed as percentages.
The CRLI and the CRBSI are reported as follows: the percentage of catheters that developed CRLI and the number of CRLIs per 1,000 catheter-days, and the percentage of catheters that developed CRBSIs and the number of CRBSIs per 1,000 catheter-days.
The comparisons of the arterial catheter sites on age, the APACHE II score, the total duration with the index catheter and the total duration of ICU stay were carried out with Kruskall-Wallis test for independent samples. The chi-square test was used for comparing proportions between the different arterial catheter sites on the diagnosis groups, the order of catheter insertion and diabetes mellitus. The comparisons of the incidence, per catheter-days, of CRLI and CRBSI between the different arterial catheter sites were performed using four Poisson regression analyses.
Eight models were constructed for adjusting by the total duration of ICU stay, the diagnosis group, the order of catheter insertion and sex. The arterial catheter site was the main independent variable. The rates of CRLI and CRBSI were introduced as dependent variables.
An a posteriori comparison was carried out among the four arterial catheter sites. Statistical analyses were performed with SPSS 12.0.1 (SPSS Inc., Chicago, IL, USA) and LogXact 4.1 (Cyrus Mehta and Nitin Patel, Cambridge, MA, USA).
The magnitude of the effects is expressed as the odds ratio and 95% confidence interval. P < 0.05 was considered statistically significant. Inferences are based on exact P values (Poisson analysis) and asymptotic P values (comparisons of arterial catheter sites on age, APACHE II score, total duration with the index catheter and the total duration of ICU stay).