Infection developing in the intensive care unit is a common complication of critical illness, but notoriously difficult to diagnose. A definite diagnosis based on the most reliable tests usually is not possible for at least two days. It is unclear what the optimal management approach should be while awaiting the results of diagnostic tests. In some circumstances, broad spectrum antibiotics are started with a plan to adjust them once the results of cultures are available. Observational studies show that this results in greater antibiotic use, and the risk of superinfection and resistance. In other circumstances, antibiotics may be withheld pending the results of cultures, a strategy that leads to a delay in therapy when cultures are positive, and that may be associated with a worse clinical outcome.

We undertook a randomized pilot study to address the question: "In a critically ill patient for whom clinicians are uncertain whether infection may be present, and in whom potential sites of infection have been managed by removing or changing invasive devices, can a policy of delaying antibiotic treatment until cultures are available reduce the risks of excessive antibiotic use, without increasing the risks associated with delayed therapy?"

Recognizing that the question has not been formally addressed before, and that approaches to clinical management are both widely divergent and passionately held, our pilot study tested the feasibility and acceptability of undertaking a larger trial with sufficient power to determine equivalence.

We randomized critically ill patients who had been in hospital for at least 72 hours, and in the ICU for at least 24 hours, and who manifested either a temperature >38.5 degrees, or a temperature>38.0 degrees and a white cell count >12,000, and in whom clinicians entertained the possibility of infection as a diagnosis, to either site-specific broad spectrum empiric antibiotics or the corresponding placebo. All patients underwent a comprehensive series of investigations to identify an infectious focus, and all patients had full source control, including changes of central lines and urinary catheters, and change of nasogastric to orogastric tubes.

Patients were maintained in assigned study arm for seven days, or until culture data were available, at which time they were switched to culture-guided narrow spectrum therapy

Inclusion Criteria:

• In hospital > 72 hrs and in ICU > 24hrs, and
• Core temperature ≥38.5°C, or temperature ≥ 38.0°C with a WBC>12,000/mm3, or temperature ≤ 36.0°C with a WBC > 12,000/mm3
• Suspicion of infection

Exclusion Criteria:

• Age < 18 years

• Imminent death (within 24 hrs) or withdrawal of aggressive therapy

• Prosthetic heart valve or vascular graft

• Neutropenia (Absolute neutrophil count < 1000/mm3)

• Received > 16 hours of a broad spectrum antibiotic in the last 24 hours (3rd gen cephalosporin, fluoroquinolone, carbapenem, anti-pseudomonal penicillin) or any combination therapy

• History of allergic reaction to both study medications

• New physical findings consistent with infection:

  • Meningeal signs
  • Peritonitis + free air on Abdo x-ray
  • Soft tissue infection / cellulitis
  • Murmur & suspicion of endocarditis

• Newly available (within past 24 hours) culture results consistent with infection