catheter tip and a peripheral venous blood culture, and 53% were based on (semi) quantitative culture of the catheter tip and a blood culture drawn from an arterial catheter or no blood culture. Table I shows the most relevant patient and CVC characteristics, including the infection frequencies. Figure 1 shows the risk of CR‐BSI over time, averaged per four days. Prolonged catheterisation was significantly associated with the risk of developing infection [odds ratio of one extra CVC‐day 1.1, 95% CI 1.07–1.14] in univariate logistic regression (Proc logistic in SAS)). Multi‐level Poisson regression (Proc glimmix in SAS) revealed that both hospital and patient level were not significant. Therefore, Cox regression (Proc phreg in SAS) was used to Figure 1: Daily hazard for the development of catheter‐related bloodstream infection, averaged per four days, with 95% confidence intervals. obtain the presented results. The time‐dependent ‘ICU presence’ was not included in the multi‐variate analysis (hazard ratio 1.3, 95% CI 0.7–2.4), but length of ICU stay prior to insertion was significantly associated with the risk of CR‐BSI. Table II presents the hazard ratios of univariate and multi‐variate analyses. The results of multi‐variate analysis were adjusted for the confounding variable ‘lumen’ (single vs multi‐lumen). When adjusted for lumen, the effect of prolonged ICU stay was more pronounced. Insertion in the jugular (P=0.06) or femoral vein compared with the subclavian vein, (prolonged) ICU‐stay prior to CVC insertion (up to 20 days) and TPN through the CVC were associated with an increased risk of CR‐BSI, whereas administration of antibiotics through the CVC was associated with a lower risk of infection. Daily hazard (average of four days) 0 5 10 15 20 25 1-4 5-8 9-12 13-16 17-20 21-24 25-28 number of days in place 48 Chapter 3
RkJQdWJsaXNoZXIy MTk4NDMw