Predictive value of broncho-alveolar lavage fluid Gram's stain in the diagnosis of ventilator-associated pneumonia: a prospective study.

BACKGROUND: Quantitative broncho-alveolar lavage (qBAL) is increasingly being used for diagnosing ventilator-associated pneumonia (VAP). The current study prospectively evaluates the accuracy of broncho-alveolar lavage fluid Gram's stain (GS) in predicting both the presence of VAP and the class of causative microorganism in patients suspected of VAP.

METHODS: Patients suspected of VAP in a trauma or surgical intensive care unit underwent bronchoscopic qBAL with GS. Presence and class of organisms seen on GS were correlated respectively with the presence of VAP, as diagnosed by qBAL, and class of causative microorganism. VAP was defined as qBAL >10(5) colony forming units/mL. All data were gathered prospectively.

RESULTS: During a 28-month study period, 229 patients underwent 309 qBALs for suspected VAP. Seventy-one (23%) specimens were positive for VAP (qBAL>10(5) CFU/mL). Fifty-four specimens (77%) had one causative microorganism, 13 (18%) had two, 3 (4%) had three, and 1 (1%) demonstrated four microorganisms giving a total of 93 VAPs. Forty-one (62%) of 66 specimens showing moderate or many microorganisms on GS were positive for VAP. However, 7 (4%) of 167 specimens showing none and 23 (30%) of 76 showing few microorganisms on GS were also positive for VAP. Of the 64 qBAL specimens positive for VAP and where the GS showed microorganisms, 6 (23%) of 26 showing only G+ microorganisms on GS had G- VAP (G- alone, 4; G+ and G-, 2), and 1 (8%) of 12 showing G- microorganisms only had G+ and G- VAP. Of the seven qBAL specimens positive for VAP where the GS did not show microorganisms, one had G+ and six had G- VAP. With the threshold of positivity of GS at more than none, the sensitivity, specificity, positive, and negative predictive values of GS for the presence of VAP were 90%, 67%, 45%, and 96% respectively.

CONCLUSIONS: Broncho-alveolar lavage fluid GS is poor in predicting the presence of VAP and predicting the class of causative microorganism. Using GS to determine necessity of and to select class of antimicrobial therapy will result in delayed or inappropriate VAP therapy or both.