More than 10% of patients admitted to intensive care units (ICUs) experience a severe, healthcare-associated infection, such as ventilator-associated pneumonia (VAP) or bloodstream infection (BSI). What could be a public health target for prevention is hotly debated, because properly adjusting for intrinsic risk factors in the patient population is difficult. We aimed to estimate the proportion of ICU-acquired VAP and BSI cases that are amenable to prevention in routine conditions.

We analyzed routine data collected prospectively according to the European standard protocol for patient-based surveillance of healthcare-acquired infections in ICUs. We computed the number of infections to be expected if, after adjustment for case mix, the infection incidence in ICUs with higher infection rates could be reduced to that of the top-tenth-percentile-ranked ICU. Computations came from model-based simulation of individual patient profiles over time in the ICU. The preventable proportion was computed as the number of observed cases minus the number of expected cases divided by the number of observed cases.

Data for 78,222 patients admitted for more than 2 days to 525 ICUs in 6 European countries from 2005 to 2008 were available for analysis. We calculated that 52% of VAP and 69% of BSI was preventable.

Our pragmatic, if highly conservative, estimates quantify the potential for prevention of VAP and BSI in routine conditions, assuming that variation in infection incidence between ICUs can be eliminated with improved quality of care, apart from variation attributable to differential case mix.

To identify modes of HCV transmission during an outbreak of HCV infection in a hemodialysis unit.

An epidemiologic study, virologie analysis, assessment of infection control practices and procedures, and technical examination of products and dialysis machines.

A private hemodialysis unit treating approximately 70 patients.

Detection of HCV RNA by PCR was performed among patients receiving dialysis in 2001. Case-patients were patients who had a first positive result for HCV RNA between January 2001 and January 2002 and either acute hepatitis, a seroconversion for HCV antibodies, or a previous negative result. Three control-patients were randomly selected per case-patient.

Of the 61 patients treated in the unit in 2001 and not infected with HCV, 22 (36.1%) became case-patients with onset from May 2001 to January 2002 for an incidence density rate of 70 per 100 patient-years. Phylogenic analysis identified four distinct HCV groups and an index case-patient for each with a similar virus among patients already known to be infected. No multidose medication vials or material was shared between patients. Connection to a dialysis machine by a nurse who had connected an HCV-infected patient “just before” or “one patient before” increased the risk of HCV infection, whereas using the same dialysis machine after a patient infected with HCV did not. Understating, lack of training, and breaches in infection control were documented. Direct observation of practices revealed frequent flooding of blood into the double filter on the arterial pressure tubing set.

During this outbreak, HCV transmission was mainly patient to patient via healthcare workers' hands. However, transmission via dialysis machines because of possible contamination of internal components could not be excluded.