Empiric antimicrobial therapy in critical illness: results of a surgical infection society survey.

Background: Antibiotics are prescribed commonly in the intensive care unit (ICU). Often, therapy is initiated empirically; practice patterns are not well characterized. We documented approaches to empiric antibiotic therapy among members of the Surgical Infection Society (SIS).

Methods: We sent a scenario-based questionnaire to all SIS members. The hypothetical cases addressed empiric broad-spectrum therapy for a patient with pyrexia and leukocytosis and the use of vancomycin for central venous catheter infection.

Results: The 113 respondents were primarily surgeons (96%) with a university-based practice (92%). Most attended in the ICU (72%), and they had practiced for 14 +/- 8 years. Whereas 63% of the respondents identified overuse of antibiotics as a problem in their ICU, only 19% said inadequate treatment of infection was a concern. For a febrile patient with negative cultures who was receiving antibiotics, estimates of the likelihood of infection increased across the three scenarios as the degree of organ failure increased (p < 0.0001; chi-square test). Deteriorating organ function was associated with a decision to broaden empiric therapy (58% vs. 33%; p < 0.0001) and to initiate anti-fungal therapy (27% vs. 9%; p < 0.0001) rather than to stop antibiotics and re-culture (15% vs. 51%; p < 0.0001). There was considerable variability in management strategy across the scenarios: Even in the face of organ dysfunction, 58% of physicians would add or change empiric therapy, whereas 30% would not. For each scenario, 23 to 25 antibiotic regimens were designated as optimal therapy. Only 45% of the respondents would initiate empiric vancomycin for suspected central line infection. Variability in approach was not explained by critical care practice, academic position, or country.

Conclusions: Clinical deterioration is a strong determinant of a decision to initiate or broaden empiric antibiotic therapy during critical illness. The substantial variability in approach suggests a state of clinical equipoise that calls for more rigorous evaluation through a randomized controlled trial.