Empirical antimicrobial therapy of septic shock patients: Adequacy and impact on the outcome*

Abstract
Objective To assess the adequacy of empirical antimicrobial therapy prescribed in septic shock patients and to evaluate the relationship between inadequate antimicrobial therapy and 30-day mortality. Design Prospective observational study. Setting Medical-surgical (16-bed) intensive care unit in an urban teaching hospital. Patients A total of 107 patients requiring intensive care admission were prospectively evaluated during the 3-yr period of the study. Interventions Prospective patient surveillance and data collection and assessment of antimicrobial therapy according to microbiological documentation. Measurements and Main Results A source of infection associated with a microbiological documentation was identified in 78 of the 107 patients (72%). Empirical antimicrobial therapy consisted of a pivotal antibiotic (β-lactam) associated with an aminoglycoside (59 patients) or a fluoroquinolone (21 patients). Vancomycin was added in 14 patients. Sixty-nine of the 78 patients (89%) received an adequate antimicrobial therapy. The mortality rate of patients receiving an adequate antimicrobial therapy was 56%, and seven of the nine patients (78%) receiving an inadequate antimicrobial therapy died (p = .2). Among the 81 patients who were alive on day 3, antimicrobial therapy was modified in agreement to clinical status and microbiological documentation in 80% of cases, with de-escalation in 64% of cases. De-escalation consisted of withdrawing the nonpivotal antibiotic in 42% of patients or switching to a narrow-spectrum β-lactam antibiotic (22% of cases). Conclusion The prescription of empirical antimicrobial therapy by a senior physician in agreement with practice guidelines made it possible to achieve a crude rate of 89% of adequate antimicrobial therapy in study patients. Inadequate antimicrobial therapy was associated with a 39% excess of mortality. A de-escalation of the empirical therapy was possible in 64% of patients.