Objective: In an attempt to improve our ability to diagnose the cause of ventilator-associated pneumonia (VAP), we explore the usefulness of the conditional evaluation of bronchoalveolar lavage (BAL) samples from the involved and non-involved areas in patients with suspected unilateral lobar VAP (UL-VAP).
Design: Prospective study.
Setting: University teaching hospital intensive care unit.
Patients: We studied 19 consecutive patients with suspected UL-VAP.
Measurements and main results: Nine of the 12 patients (47%) developed UL-VAP. There was a significant difference between the involved and non-involved areas in UL-VAP patients (P < 0.001) in respect of the quantitative bacterial cultures (QBCs) of BAL samples for each micro-organism, whereas there was no difference in patients without UL-VAP. When we applied the criterion of usual BAL (one micro-organism in concentrations > 10(5) colony-forming units per millilitre) for UL-VAP diagnosis, the sensitivity was 100%, the specificity 70%, the positive predictive value 75%, and the negative predictive value 100%. When we used the conditional evaluation of the BAL results for UL-VAP diagnosis, in the involved and non-involved areas, the sensitivity was 78%, the specificity 90%, the positive predictive value 87.5% and the negative predictive value 82%. A statistically significant difference was found when we compared the difference in QBCs between the BAL samples for each micro-organism, between the involved and non-involved areas in patients with and without VAP (P < 0.001).
Conclusion: These data suggest that utilisation of the conditional evaluation of the QBCs of BAL samples improves significantly our ability to diagnose the cause of UL-VAP.