A reappraisal of blind bronchial sampling in the microbiologic diagnosis of nosocomial bronchopneumonia. A comparative study in ventilated patients.

To assess the usefulness of fiberscopy for microbiologic diagnosis of nosocomial bronchopneumonia (NBP) in ventilated patients. Data were collected prospectively. We compared the results of semiquantitative cultures obtained by protected specimen brush (PSB), bronchoalveolar lavage (BAL) and blind bronchial sampling (BBS). Positive thresholds were 10(3) CFU ml-1 for PSB and BAL and 10(4) CFU ml-1 for BBS. We also evaluated the diagnostic performance of direct examination of samples obtained by BAL and BBS. We carried out this study in 64 ventilated patients admitted to a medico-surgical ICU. During the study, 85 sets of samplings were obtained. The concordance between the results of specimen cultures obtained with the three techniques was 87 percent. The concordance between BBS and PSB or between BBS and BAL was 91.8 percent. In two of seven patients with discordant results between BBS and PSB, the microorganisms isolated from blood cultures were found on BBS, but not on PSB samples. As for direct examination, the thresholds for the diagnosis of NBP using BBS were as follows: > or = 10 polymorphonuclear neutrophils (PMN)/high-power field (HPF), > or = 1 bacteria/oil immersion field (OIF), presence of intracellular bacterial inclusions. Using BAL, the thresholds were as follows: > or = 1 PMN/HPF, presence of bacteria/OIF, presence of intracellular bacterial inclusions. The specificity of the presence of bacterial inclusions was excellent regardless of the sampling technique, but the sensitivity of this criteria was mediocre (30.8 percent with BBS and 19.2 percent with BAL). Except for the number of PMN on BBS, all the other diagnostic criteria (PMN count on BAL, bacterial count, count of cells exhibiting inclusions) provide a similar prediction of NBP (correctly classified: 61.2 to 81.2 percent). No combination of criteria enabled significantly better classification regardless of the sampling technique. In view of these findings and the high cost and morbidity of fiberscopy, it is arguably better to use a simple, repeatable, and risk-free technique for obtaining culture specimens from mechanically ventilated patients. Obviously, protected brushing techniques remain the most effective for nonintubated patients.