[Standardized incidence ratio: a risk index for catheter-related infection surveillance in intensive care units (REACAT network) in Northern France].

Background: Nosocomial infection surveillance is one of the major indicators used to compare health care quality in hospital settings. Wards participating in a network with standardized methods can be compared. We propose a risk index adjusted for catheter-related infection (CRI) specific risk factors in the setting of a CRI surveillance network in intensive care units (ICU): the standardized incidence ratio (SIR).

Methods: All central venous catheters (CVC) inserted for more than 48h were prospectively followed until CVC removal or patient discharge in a yearly 4-month surveillance. Standard clinical and microbiological criteria were used to define colonization and CRI. A logistic regression model, developed on a 3-year pooled database, was used as a predictive model of CRI. Expected number of CRI was calculated and compared with the observed number of CRI to estimate SIR for each year and for each ICU per year.

Results: From 2000 to 2003, 108 ICU participated in at least one of the 3 surveillance periods, including 6414 CVC. Overall, 239 CRI were identified (incidence density (ID): 3.6 CRI/1000 CVC-days). At multivariate analysis, duration of CVC placement (1.1 [1.0-1.1]), rank (1.7 [1.1-2.2]) and site of CVC insertion (1.6 [1.2-2.1]), use of CVC for antibiotic therapy (0.5 [0.3-0.7]), organ failure at CVC removal (2.2 [1.5-3.2]), infection at another site at CVC removal ([1.9 [1.4-2.6]) were significantly associated with CRI. During the last period of surveillance, 14 ICU had a DI higher than 5.5 CRI/1000 CVC-days. More CRI than expected were significantly observed in two wards including one which followed less than 20 CVC.

Conclusion: The REACAT surveillance system assesses a novel and reliable risk index which enables identification of ICU with a higher CRI risk and to focus on prevention.