IntroductionMost data on de-escalation of empirical antimicrobial therapy has focused on ventilator-associated pneumonia. In this retrospective monocentric study, we evaluated de-escalation as part of a global strategy of empiric antibiotherapy management irrespective of the location and the severity of the infection. The goal of this trial was to assess the application of a de-escalation strategy and the impact in terms of re-escalation, recurrent infection and to identify variables associated with de-escalation.MethodsAll consecutive patients treated with empiric antibiotic therapy and hospitalized in the intensive care unit for at least 72 hours within a period of 16 months were included. We compared the characteristics and outcome of patients who have experienced de-escalation therapy with those who have not.ResultsA total of 116 patients were studied corresponding to 133 infections. Antibiotic therapy was de-escalated in 60 cases (45%). De-escalation, primarily accomplished by a reduction in the number of antibiotics used, was observed in 52% of severe sepsis or septic shock patients. Adequate empiric antibiotic and use of aminoglycoside were independently linked with de-escalation. De-escalation therapy was associated with a significant reduction of recurrent infection (19% vs 5% P = 0.01). Mortality was not changed by de-escalation.ConclusionsAs part of a global management of empiric antibiotherapy in an intensive care unit, de-escalation might be safe and feasible in a large proportion of patients.