Healthcare-associated infections are more common in intensive care units than in other areas of care due to the critical condition of patients and the invasive nature of life-support techniques. Surveillance in these units is essential for quantifying the risk of infection, monitoring its progression, and identifying prevention strategies. In France, network-based surveillance of nosocomial infections in intensive care began at the interregional level as early as 1994. Thanks to the standardization of surveillance methodologies within the framework of Raisin (Network for Alert, Investigation, and Surveillance of Nosocomial Infections), this network became national in 2004, with coordination entrusted to CClin Sud-Est. This surveillance focuses on infections associated with invasive devices ("device-related"), for which prevention is a priority. The incidence rate of these infections per 1,000 days of exposure to risk is the preferred indicator: it provides the best measure of their frequency and allows for comparisons between departments. This report is the second produced by REA-Raisin and covers the 2005 French national data from the network-based surveillance of the incidence of nosocomial infections in adult intensive care. From January 1 to June 30, 2005, 151 intensive care units in France voluntarily participated in the REA-Raisin surveillance program, continuously collecting data over a 6-month period on 20,632 patients hospitalized for more than 2 days in intensive care. The level of infection risk must be interpreted in light of various risk factors considered in the surveillance, particularly those related to the patient: age (mean age of 61.7 years), gender (male-to-female ratio of 1.62), diagnostic category (medical 68%, emergency surgery 17%, and elective surgery 15%), trauma status (9%), immunosuppression (12%), patient origin (external 54%, short-stay 38%, medium- and long-stay 4%, other ICU 3%), presence of antibiotic therapy at admission (52%), severity score (mean IGS II of 40.1), and finally length of stay (average of 11.2 days). The use of invasive devices must also be considered: intubation (60.5% of exposed patients), central venous catheterization (57.6%), or urinary catheterization (79.2%); the influence of the duration of exposure to these devices was taken into account. Among the 20,632 patients monitored, 2,569 patients (12.45%) developed at least one infection (pneumonia, central venous catheter-related infection (CVC-related infection), bacteremia, urinary tract infection). Including CVC colonizations, the surveillance recorded a total of 5,159 nosocomial events. Across all departments, the observed incidence rates were as follows: 17.58 cases of pneumonia per 1,000 days of intubation, 5.54 colonizations per 1,000 days of central venous catheterization (or 2.24 when considering only CVI), 3.32 bacteremias per 1,000 days of intensive care hospitalization, and 7.88 urinary tract infections per 1,000 days of catheterization. The characteristics of the intensive care units were highly heterogeneous (size, equipment, organization, practices, patient population), which partly explains the significant inter-unit variation in the observed incidence rates. Comparing them requires optimal adjustment of the indicators based on the use of "patient" risk factors, even if collecting this data increases the workload. A multivariate analysis model enabling this adjustment to construct standardized infection ratios (observed-to-expected infection ratio) to facilitate inter-departmental comparisons is currently under study and will be the subject of a separate publication. This report serves as a national benchmark, improving understanding of nosocomial infection risk in intensive care and enabling the optimization of risk control through the provision of feedback on results to intensivists. Annual analysis of these data will ultimately allow for the evaluation of the impact of prevention measures implemented by participating departments. (R.A.)