Objective: To assess prediction of multidrug resistant (MDR) pathogens in ventilator-associated pneumonia (VAP) by systematic surveillance cultures (SC) and to assess the contribution of SC to initial antibiotic therapy.
Design: Prospective cohort study of patients with microbiologically confirmed VAP. Comparison of actual early antibiotic coverage with three hypothetical empirical schemes.
Setting: A 50-bed university hospital ICU. SC consisted of oral, nasal, urinary and rectal samples upon admission, 3-weekly urinary and 1-weekly oral, nasal, and rectal samples in all patients, 3-weekly tracheal aspirates in intubated patients.
Results: MDR pathogens were found in 86 of 199 VAP episodes. Sensitivity of SC to predict MDR pathogens was 69% (tracheal SC) and 82% (all SC); specificity was 96% (tracheal) and 91% (all), respectively. Appropriate antibiotic coverage within 24 h and 48 h following MDR VAP was 77% and 89%, respectively. A carbapenem-based empirical scheme would have been equally appropriate (83% vs. 77% at 24 h; 83% vs. 89% at 48 h), but a beta-lactam-fluoroquinolone empirical therapy would have been less (59% vs. 77% at 24 h; 59% vs. 89% at 48 h) as would have been beta-lactam-aminoglycoside therapy (68% vs. 77% at 24 h; 68% vs. 89% at 48 h). Empirical comparators would have resulted in significantly more prescription of broad-spectrum antibiotics within the first 48 h.
Conclusions: With MDR pathogens highly prevalent, systematic SC predicted MDR pathogens causing VAP in 69% to 82% and may have contributed to high rates of early appropriate antibiotic therapy with limited use of broad-spectrum antimicrobials.