Bloodstream infection (BSI) is defined by positive blood cultures in a patient with systemic signs of infection and may be either secondary to a documented source or primary-that is, without identified origin. Community-acquired BSIs in immunocompetent adults usually involve drug-susceptible bacteria, while healthcare-associated BSIs are frequently due to multidrug-resistant (MDR) strains. Early adequate antimicrobial therapy is a key to improve patient outcomes, especially in those with criteria for sepsis or septic shock, and should be based on guidelines and direct examination of available samples. Local epidemiology, suspected source, immune status, previous antimicrobial exposure, and documented colonization with MDR bacteria must be considered for the choice of first-line antimicrobials in healthcare-associated and hospital-acquired BSIs. Early genotypic or phenotypic tests are now available for bacterial identification and early detection of resistance mechanisms and may help, though their clinical impact warrants further investigations. Initial antimicrobial dosing should take into account the pharmacokinetic alterations commonly observed in ICU patients, with a loading dose in case of sepsis or septic shock. Initial antimicrobial combination attempting to increase the antimicrobial spectrum should be discussed when MDR bacteria are suspected and/or in the most severely ill patients. Source identification and control should be performed as soon as the hemodynamic status is stabilized. De-escalation from a broad-spectrum to a narrow-spectrum antimicrobial may reduce antibiotic selection pressure without negative impact on mortality. The duration of therapy is usually 5-8 days though longer durations may be discussed depending on the underlying illness and the source of infection. This narrative review covers the epidemiology, diagnostic workflow and therapeutic aspects of BSI in ICU patients and proposed up-todate expert statements.

Empirical broad spectrum antimicrobial therapy prescribed in life-threatening situations should be de-escalated to mitigate the risk of resistance emergence. Definitions of de-escalation (DE) vary among studies, thereby biasing their results. The aim of this study was to provide a consensus definition of DE and to establish a ranking of β-lactam according to both their spectra and their ecological consequences. Twenty-eight experts from intensive care, infectious disease and clinical microbiology were consulted using the Delphi method (four successive questionnaires) from July to November 2013. More than 70% of similar answers to a question were necessary to reach a consensus. According to our consensus definition, DE purpose was to reduce both the spectrum of antimicrobial therapy and the selective pressure on microbiota. DE included switching from combination to monotherapy. A six-rank consensual classification of β-lactams allowing gradation of DE was established. The group was unable to differentiate ecological consequences of molecules included in group 4, i.e. piperacillin/tazobactam, ticarcillin/clavulanic acid, fourth-generation cephalosporin and antipseudomonal third-generation cephalosporin. Furthermore, no consensus was reached on the delay within which DE should be performed and on whether or not the shortening of antibiotic therapy duration should be included in DE definition. This study provides a consensual ranking of β-lactams according to their global ecological consequences that may be helpful in future studies on DE. However, this work also underlines the difficulties of reaching a consensus on the relative ecological impact of each individual drug and on the timing of DE.