30 www.thelancet.com/infection Vol 11 January 2011 Articles Lancet Infect Dis 2011; 11: 30–38 Published Online December 1, 2010 DOI:10.1016/S1473- 3099(10)70258-9 See Comment page 3 Healthcare-associated Infections Unit, Public Health and Surveillance Department, Scientific Institute for Public Health, Brussels, Belgium (M-L Lambert MD, I Morales MD, K Mertens MSc); Surveillance Unit, European Centre for Disease Prevention and Control, Stockholm, Sweden (C Suetens MD); Université Lyon 1, Centre National de la Recherche Scientifique, UMR 5558/ Hospices Civils de Lyon, Centre de Coordination de la Lutte contre les Infections Nosocomiales Sud-Est, Lyon, France (A Savey MD); Universitat Autonoma de Barcelona, Hospital Vall d’Hebron, Barcelona, Spain (M Palomar MD); Department of Anaesthesia, General Intensive Care, and Pain Control, Division of Cardiovascular and Thoracic, Anaesthesia, and Intensive Care, Medical University of Vienna, General Hospital of Vienna, Vienna, Austria (Prof M Hiesmayr MD); Italian Hospital Hygiene Study Group—Italian Nosocomial Infection Surveillance in Intensive Care Units GISIO-SItI, Department of Biomedical Sciences, University of Catania, Catania, Italy (Prof A Agodi PhD); Department of Environmental Health Sciences (Prof U Frank MD), Department of Medical Biometry and Statistics (Prof M Schumacher PhD), and Centre of Data and Medical Informatics—Analysis and Modelling (M Wolkewitz PhD), Freiburg University Medical Centre, Freiburg, Germany Clinical outcomes of health-care-associated infections and antimicrobial resistance in patients admitted to European intensive-care units: a cohort study Marie-Laurence Lambert, Carl Suetens, Anne Savey, Mercedes Palomar, Michael Hiesmayr, Ingrid Morales, Antonella Agodi, Uwe Frank, Karl Mertens, Martin Schumacher, Martin Wolkewitz Summary Background Patients admitted to intensive-care units are at high risk of health-care-associated infections, and many are caused by antimicrobial-resistant pathogens. We aimed to assess excess mortality and length of stay in intensive- care units from bloodstream infections and pneumonia. Methods We analysed data collected prospectively from intensive-care units that reported according to the European standard protocol for surveillance of health-care-associated infections. We focused on the most frequent causative microorganisms. Resistance was defined as resistance to ceftazidime (Acinetobacter baumannii or Pseudomonas aeruginosa), third-generation cephalosporins (Escherichia coli), and oxacillin (Staphylococcus aureus). We defined 20 different exposures according to infection site, microorganism, and resistance status. For every exposure, we compared outcomes between patients exposed and unexposed by use of time-dependent regression modelling. We adjusted results for patients’ characteristics and time-dependency of the exposure. Findings We obtained data for 119 699 patients who were admitted for more than 2 days to 537 intensive-care units in ten countries between Jan 1, 2005, and Dec 31, 2008. Excess risk of death (hazard ratio) for pneumonia in the fully adjusted model ranged from 1·7 (95% CI 1·4–1·9) for drug-sensitive S aureus to 3·5 (2·9–4·2) for drug-resistant P aeruginosa. For bloodstream infections, the excess risk ranged from 2·1 (1·6–2·6) for drug-sensitive S aureus to 4·0 (2·7–5·8) for drug- resistant P aeruginosa. Risk of death associated with antimicrobial resistance (ie, additional risk of death to that of the infection) was 1·2 (1·1–1·4) for pneumonia and 1·2 (0·9–1·5) for bloodstream infections for a combination of all four microorganisms, and was highest for S aureus (pneumonia 1·3 [1·0–1·6], bloodstream infections 1·6 [1·1–2·3]). Antimicrobial resistance did not significantly increase length of stay; the hazard ratio for discharge, dead or alive, for sensitive microorganisms compared with resistant microorganisms (all four combined) was 1·05 (0·97–1·13) for pneumonia and 1·02 (0·98–1·17) for bloodstream infections. P aeruginosa had the highest burden of health-care-acquired infections because of its high prevalence and pathogenicity of both its drug-sensitive and drug-resistant strains. Interpretation Health-care-associated bloodstream infections and pneumonia greatly increase mortality and pneumonia increase length of stay in intensive-care units; the additional effect of the most common antimicrobial resistance patterns is comparatively low. Funding European Commission (DG Sanco). Introduction Patients in intensive-care units are at high risk of health- care-associated infections because of intrinsic (eg, severity of illness or impaired immunity) and extrinsic (eg, mechanical ventilation or central line catheterisation) risk factors. In Europe, 3·0% of patients staying more than 2 days in intensive-care units acquire bloodstream infections, and 6·2% acquire pneumonia. 1 Because of their precarious clinical states, common use of antibiotics, and high prevalence of antimicrobial resistance, patients are at high risk of infection with resistant pathogens. 2 Data for clinical outcomes of infections and antimicrobial resistance are conflicting. 3 A key issue is the contribution of host and infection factors. Data are very dependent on methods used to address differences in patients before infection (such as underlying disease severity), type of infection, and causative pathogen. 3 Large studies are needed to provide detailed information on all these factors. Furthermore, modern statistical techniques are needed to account for the timing and duration of exposure 4 and the competing risks caused by informative censoring (patients leaving an intensive-care unit alive are by definition in better health than are those that stay). 5 We aimed to assess the excess mortality and length of stay in intensive-care units associated with bloodstream infections and pneumonia acquired in intensive care and for infections caused by pathogens with and without some common patterns of antimicrobial resistance. We provide separate estimates for Acinetobacter baumannii, Escherichia coli, Pseudomonas aeruginosa, and Staphylococcus aureus, because of their prevalence, 1 pathogenicity, burden of antimicrobial resistance, and the availability of data for antimicrobial resistance (tracer phenotypes) in our database. This study was part of the