Protected specimen brush (PSB) is considered to be one of the standard methods for the diagnosis of ventilator-associated pneumonia, but to our knowledge, intraindividual variability in results has not been reported previously.
Purpose: To compare the results of two PSB performed in the same subsegment on patients with suspected ICU-acquired pneumonia (IAP).
Study design: Between October 1991 and April 1992, each mechanically ventilated patient with suspected IAP underwent bronchoscopy with two successive PSB in the lung segment identified as abnormal on radiographs. Results of the two PSB cultures were compared using 10(3) cfu/ml cutoff for a positive result. Four definite diagnoses were established during the follow up: definite pneumonia, probable pneumonia, excluded pneumonia, and uncertain pneumonia.
Population: Forty-two episodes in 26 patients were studied; 60 percent of patients received prior antibiotic therapy. Thirty-two microorganisms were isolated from 24 pairs of PSB. Definite diagnosis was definite pneumonia in 7, probable pneumonia in 8, excluded pneumonia in 17, and uncertain pneumonia in 10 cases.
Results: The PSB recovered the same microorganisms and argued for a good qualitative reproducibility. The distinction of positive and negative results on the basis of the 10(3) cfu/ml classic threshold was less reproducible. For 24 percent of the microorganisms recovered and in 16.7 percent of episodes of suspected IAP, the two consecutive samples gave results spread out on each side of the 10(3) cfu/ml cutoff. Discordance was higher when definite diagnosis was certain or probable than when diagnosis was excluded (p = 0.015). There was no statistical effect of the order of samples between the two specimens for bacterial index and microorganism concentrations.
Conclusion: These findings argue for the poor repeatability of PSB in suspected IAP and question the yield of the 10(3) cfu/ml threshold. In attempting to diagnose IAP, the results of PSB must be interpreted with caution considering the intraindividual variability.