ABSTRACT
BACKGROUND: The lack of prospective, randomized, controlled trial data to guide optimal antibiotic use in bacteraemic critically ill patients has led to a wide variety of strategies and major issues with drug resistance. We therefore prospectively investigated the epidemiology of bacteraemia and fungaemia in intensive care units (ICUs); and the impact of timing, type and appropriateness of antibiotic intervention. METHODS: We conducted a multinational, multicentre, prospective observational study in 132 ICUs from 26 countries with no interventions. RESULTS: 1702 patients [European (69.6%), Australasian (12.2%), South American (8.3%) and Asian (9.9%)] developed 1942 bacteraemic episodes over the study period. Mortality rates were similar for those receiving empirical (40.5%), semi-targeted (37.6%) or fully targeted (33.3%) antibiotic therapy (P=0.40), and in those initially receiving broad- (39.3%) or restricted-spectrum (39.1%) therapy (P=0.94). First-line therapy was effective in terms of the antibiogram (where available) in 70.4% of cases. This in vitro susceptibility ranged from 76.3% for broad-spectrum antibiotics to 46.3% for restricted-spectrum antibiotics (P<0.0001). However, no antibiotic policy-associated variable, including in vitro susceptibility (odds ratio 0.89, 95% confidence interval 0.61-1.30), was a statistically significant predictor of mortality. CONCLUSIONS: We could not show an impact of antibiotics on mortality in critically ill patients, despite in vitro activity and early commencement. Randomized, multicentre trials are urgently needed to establish the appropriate duration, timing and combinations of antibiotics that will both optimally treat infection and minimize development of resistance and other complications.