BACKGROUND: Current strategies for diagnosing ventilator-associated pneumonia (VAP) favor the use of quantitative methods; however, semi-quantitative cultures of endotracheal aspirates are still commonly used.

METHODS: The microbiological results of patients with suspected VAP who had both quantitative cultures with non-bronchoscopic bronchoalveolar lavage (BAL) and semi-quantitative cultures of endotracheal aspirate obtained within 24 hours of each other were retrospectively reviewed and compared, using a quantitative threshold of >or=10(4) colony-forming units/mL as a reference standard. RESULTS: 256 patients with paired cultures were identified. Concordance between endotracheal aspirate (any growth of pathogens) and non-bronchoscopic BAL was complete in 58.2% and completely discordant in 23.8%. The sensitivity and specificity of endotracheal aspirate were 65.4% and 56.1%, which improved to 81.2% and 61.9% when antibiotic management decisions were considered in the analysis. Twenty-six patients had endotracheal aspirate cultures that were falsely negative for pathogens, with 61.5% of these patients demonstrating growth of non-fermenting Gram-negative rods or methicillin-resistant Staphylococcus aureus (MRSA) on non-bronchoscopic BAL. Overall, 45 patients (17.5%) among the entire cohort had false positive endotracheal aspirate cultures, with 19 of these patients (42.2%) demonstrating growth of non-fermenting Gram-negative rods or MRSA.

CONCLUSIONS: Semi-quantitative cultures of endotracheal aspirate are poorly concordant with quantitative cultures obtained via non-bronchoscopic BAL. Although the performance of endotracheal aspirate improves when antibiotic treatment is considered, guiding therapy on the basis of semi-quantitative cultures may still result in failure to identify potentially multiple-drug-resistant pathogens, and would also tend to promote excessive antibiotic usage. Our data support the use of quantitative cultures in diagnosing VAP.