Objectives: We aimed to describe the infectious risk during the dwell time for different catheter types. Furthermore, we wanted to identify risk factors for infections from catheters in place for >10 days.

Methods: We performed a post hoc analysis using prospectively collected data from four randomized controlled trials. First, we evaluated the infectious risk after 10 days of analysing the significance of the interaction between dwell time and catheter type in a Cox model. Second, we investigated risk factors for infection in catheters in place for >10 days using multivariable marginal Cox models.

Results: We included 15 036 intravascular catheters from 24 intensive care units. Infections occurred in 46 (0.7%) of 6298 arterial catheters (ACs), 62 (1.0%) of 6036 central venous catheters (CVCs) and 47 (1.7%) of 2702 short-term dialysis catheters (DCs). The interaction between dwell time beyond 10 days and catheter type was significant for CVCs (p 0.008) and DCs (p < 0.001), thus indicating an increased risk of infection after 10 days. The interaction was not significant for ACs (p 0.98). Therefore, we selected 1405 CVCs and 454 DCs in place for >10 days for further analyses. In the multivariable marginal Cox model, we observed an increased hazard ratio (HR) for infection for femoral CVC (HR, 6.33; 95% CI, 1.99-20.09), jugular CVC (HR, 2.82; 95% CI, 1.13-7.07), femoral DC (HR, 4.53; 95% CI, 1.54-13.33) and jugular DC (HR, 4.50; 95% CI, 1.42-14.21) compared with subclavian insertions.

Discussion: We showed that the risk of catheter infection for CVCs and DCs increased 10 days after insertion, thus suggesting routine replacement for nonsubclavian catheters in situ for >10 days.