Direct examination and cultures of bronchoalveolar lavage in pneumonia diagnosis: a comparative experimental study.

OBJECTIVE: To assess the accuracy of direct examination and quantitative cultures of BAL to diagnose pneumonia with or without antibiotic treatment.
DESIGN: Experimental rat models.
INTERVENTIONS: Pneumonia was induced by intratracheal inoculation of S. pneumoniae (10(9) cfu/ml) or P. aeruginosa (10(8)cfu/ml). Controls (n = 10) received sterile inoculum. Study animals received penicillin (n = 19) or saline (n = 18) (pneumococcal model); amikacin (n = 13), ceftazidime (n = 11), or saline (n = 13) (Pseudomonas model). BAL was assessed 48 h after infection. The animals were killed for histopathological analysis.
RESULTS: All study animals developed pneumonia, which was more extensive in the pneumococcal than in the Pseudomonas model. In pneumococcal pneumonia the sensitivity of BAL cultures (10(3) cfu/ml or higher) was 77.8% with saline and 21.0% with penicillin. In the Pseudomonas ceftazidime group all specimens were negative, precluding diagnosis. The sensitivity of cultures with amikacin was 23.1% vs. 30.8% with saline. In the pneumococcal model intracellular organism (ICO) count of 2% or higher had a sensitivity of 100% for detecting pneumonia with saline and 57.9% with penicillin. In the Pseudomonas model the sensitivity of ICO was 69.2% with both amikacin and saline and 36.3% with ceftazidime. The sensitivity of neutrophil count above 50% in pneumococcal pneumonia was 77.8% and 64.7% with saline and penicillin, respectively, and 69.2%, 61.5%, and 81.8% with saline, amikacin, and ceftazidime, respectively, in Pseudomonas pneumonia.
CONCLUSIONS: BAL-positive intracellular organisms were more accurate than cultures for the diagnosis of recent pneumonia, and were less affected by antibiotic treatment.