Abstract
BACKGROUND: Since few studies have assessed the repeatability of non-bronchoscopic bronchoalveolar lavage (NB-BAL), we compared cellular counts and cytokine concentrations in fluid obtained by standardised NB-BAL from each side of 20 intubated infants receiving extracorporeal membrane oxygenation (ECMO). METHODS: Total cell counts were obtained from 95 paired lavages and 77 pairs were suitable for differential counts and measurement of cytokine concentrations. RESULTS: Moderate correlation was noted between the two sides for most cell types including total cell counts and percentages of neutrophils and macrophages (R=0.70-0.84) and for cytokine concentrations (IL-8 R=0.78, IL-6 R=0.75, TNF-alpha R=0.64, all p< or =0.001). Using Bland-Altman analysis the mean difference between the two sides approached zero for cellular constituents (total cell counts mean difference 1.7, limits of agreement -187.5 to +190.9 x 10(4)/ml; percentage neutrophils -3.9%, -41.5% to +33.6%; percentage macrophages 3.9%, -33.8% to +41.6%) but tended to be greater on the right for logarithmically transformed cytokine measurements (IL-8: left/right ratio 0.74, limits of agreement 0.12 to 5.45, IL-6: 0.93, 0.09 to 5.87, and TNF-alpha: 0.93, 0.27 to 3.16). Using linear regression with random effects to assess the variability, only the infant's age appeared to influence the cellular results but, for cytokines, only the volume retrieved affected the variability. The magnitude of the measurements, the underlying disease, the operator's experience, days on ECMO, or survival did not affect the variability. CONCLUSION: Measurements obtained by NB-BAL need to be interpreted with caution and strongly suggest that normalisation for the dilutional effects of saline is essential.