Antibiotic management of suspected nosocomial ICU-acquired infection: Does prolonged empiric therapy improve outcome? Journal Articles uri icon

  • Overview
  • Research
  • Identity
  • Additional Document Info
  • View All


  • OBJECTIVE: To characterize empiric antibiotic use in patients with suspected nosocomial ICU-acquired infections (NI), and determine the impact of prolonged therapy in the absence of infection. DESIGN AND SETTING: Multicenter prospective cohort, with eight medical-surgical ICUs in North America and Europe. PATIENTS: 195 patients with suspected NI. METHODS: The diagnosis of NI was adjudicated by a blinded Clinical Evaluation Committee using retrospective review of clinical, radiological, and culture data. RESULTS: Empiric antibiotics were prescribed for 143 of 195 (73.3%) patients with suspected NI; only 39 of 195 (20.0%) were adjudicated as being infected. Infection rates were similar in patients who did (26 of 143, 18.2%), or did not (13 of 52, 25.0%) receive empiric therapy (p = 0.3). Empiric antibiotics were continued for more than 4 days in 56 of 95 (59.0%) patients without adjudicated NI. Factors associated with continued empiric therapy were increased age (p = 0.02), ongoing SIRS (p = 0.03), and hospital (p = 0.004). Patients without NI who received empiric antibiotics for longer than 4 days had increased 28-day mortality (18 of 56, 32.1%), compared with those whose antibiotics were discontinued (3 of 39, 7.7%; OR = 5.7, 95% CI 1.5-20.9, p = 0.005). When the influence of age, admission diagnosis, vasopressor use, and multiple organ dysfunction was controlled by multivariable analysis, prolonged empiric therapy was not independently associated with mortality (OR = 3.8, 95% CI 0.9-15.5, p = 0.07). CONCLUSIONS: Empiric antibiotics were initiated four times more often than NI was confirmed, and frequently continued in the absence of infection. We found no evidence that prolonged use of empiric antibiotics improved outcome for ICU patients, but rather a suggestion that the practice may be harmful.


  • Aarts, Mary-Anne W
  • Brun-Buisson, Christian
  • Cook, Deborah
  • Kumar, Anand
  • Opal, Steven
  • Rocker, Graeme
  • Smith, Terry
  • Vincent, Jean-Louis
  • Marshall, John C

publication date

  • August 2007