The effect of antibiotic therapy on recovery of intracellular bacteria from bronchoalveolar lavage in suspected ventilator-associated nosocomial pneumonia.

Intracellular bacteria (ICB) within recovered cells (> 7 percent) obtained via bronchoalveolar lavage (BAL) have been described as predictive of subsequent positive quantitative protected specimen brush (PSB) cultures in patients not receiving antibiotics. To determine the effect of prior or current antibiotic therapy on ICB relative to subsequent PSB culture, we prospectively evaluated 49 consecutive episodes of clinically suspected ventilator-associated pneumonia in 36 patients. Three patient groups were defined based on antibiotic administration: group 1 (current antibiotics), n = 31, samples obtained from patients currently receiving antibiotics; group 2 (recent antibiotics), n = 5, samples obtained from patients who received antibiotics > 48 h but < 72 h prior to sampling; and group 3 (no antibiotics), n = 13, samples from patients receiving no previous antibiotics within 7 days prior to sampling. Overall, PSB cultures (> or = 10(3) cfu/ml) were positive in 14 of 49 (29 percent) samples. In group 1, 2 of 31 (6 percent) samples were positive while 5 of 5 (100 percent) samples in group 2, and 7 of 13 (54 percent) in group 3 were positive. The presence or absence of ICB accurately predicted both positive and negative PSB cultures in 43 of 49 episodes. Of 43 correct predictions, 34 were negative predictions (negative ICB, negative PSB culture). The vast majority of these (29) were obtained from group 1, patients currently receiving antibiotics. In contrast, of nine positive predictions (+ICB, +PSB) virtually all (seven) occurred in group 3, patients receiving no antibiotics. In group 3, 13 of 13 PSB cultures were accurately predicted, either positive or negative, by the presence or absence of ICB. Of seven positive PSB cultures in groups 1 and 2, only 2 (28 percent) were accurately predicted by ICB. From both samples, the cultured organism was resistant to all administered antibiotics. These data suggest both prior and current antibiotic therapy reduces recovery of ICB from BAL and reduces predictive accuracy of ICB for subsequent positive PSB cultures. However, negative prediction by ICB for subsequent negative PSB cultures was good. In contrast, ICB obtained from patients not receiving antibiotics are highly predictive of subsequent PSB culture results, both positive and negative. We do not recommend BAL for evaluation of ICB in patients currently receiving antibiotics or with a recent history of antibiotic use.