Is the gram stain useful in the microbiologic diagnosis of VAP? A meta-analysis

Abstract

In a meta-analysis examining respiratory specimen Gram stain for diagnosis of ventilator-associated pneumonia, absence of bacteria on Gram stain had a high negative predictive value, but a positive Gram stain correlated poorly with organisms recovered in culture. Rapid and accurate diagnosis of ventilator-associated pneumonia (VAP) is a major challenge and no generally accepted gold standard exists for VAP diagnosis. We conducted a meta-analysis to examine the role of respiratory specimen Gram stain to diagnose VAP, and the correlation with final culture results. In 21 studies, pooled sensitivity of Gram stain for VAP was 0.79 (95% confidence interval [CI], .77-0.81; P < .0001) and specificity was 0.75 (95% CI, .73-.78; P < .0001). Negative predictive value of Gram stain for a VAP prevalence of 20%-30% was 91%, suggesting that VAP is unlikely with a negative Gram stain but the positive predictive value of Gram stain was only 40%. Pooled kappa was 0.42 for gram-positive organisms and 0.34 for gram-negative organisms, suggesting fair concordance between organisms on Gram stain and recovery by culture. Therefore, a positive Gram stain should not be used to narrow anti-infective therapy until culture results become available.