Objectives: The primary endpoint was to determine the sensitivity and specificity of the bronchoalveolar lavage Gram stain in predicting culture results. Secondary endpoints included determining the proportion of Gram stains from bronchoalveolar lavages that accurately identify culture isolates and the duration of antibiotic treatment before bronchoalveolar lavage collection.
Design: Retrospective, observational study.
Setting: Four ICUs at a single academic medical center.
Subjects: Patients at least 18 years old admitted to an ICU with a diagnosis of pneumonia, collection of a bronchoalveolar lavage sample, and receipt of antibiotics.
Measurements and main results: Two-hundred five isolates were included. Gram stains for Gram-positive and Gram-negative isolates showed high specificity, 97.3% and 100%, respectively, but lower sensitivity at 61.9% and 54.2%, respectively. The positive predictive value and negative predictive value were 77.2% and 95.7% for Gram-positive isolates and 100% and 84.4% for Gram-negative isolates, respectively. Gram stains correctly identified isolates on the bronchoalveolar lavage culture in 61.9% of Gram-positive organisms and in 54.2% of Gram-negative organisms.
Conclusions: Gram stains accurately identified causative organisms in a limited number of patients making the utility of the Gram stain an uncertain modality for predicting causative respiratory pathogens from bronchoalveolar lavage samples.
Keywords: Gram stain; antibiotics; bronchoalveolar lavage; de-escalation; diagnostics; pneumonia.
Copyright © 2021 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of the Society of Critical Care Medicine.