Influence of data filtering on reliability of respiratory impedance and derived parameters in children

Forced oscillation impedance (Zrs) measurements are frequently corrupted by noisy data in children. Our aim was to improve the reliability of respiratory resistance (Rrs) and routine identification of its flow (V') and volume (V) dependence during tidal breathing, according to Rrs = K1 + K2 x /V'/ + K3 x V. Zrs was measured at 12 Hz, using a head generator in 23 children aged 4-13 years undergoing methacholine (Mch) challenge. Rrs, K1, K2, K3, and residual mean square error (RMSD), as well as reproducibility of the parameters, were obtained. Calculations were performed separately in inspiration and expiration on the raw data, and after eliminating values lying outside confidence intervals ranging from 99% (F3SD) to 95% (F2SD) of the mean Zrs. Filtering decreased RMSD and Rrs, F3SD improved reproducibility of Rrs in expiration compared to raw data. F2SD increased K1 and decreased the amplitude of K2 and K3 before and after Mch compared to F3SD. Rrs and K2 were significantly larger and K3 significantly more negative in expiration than in inspiration at all levels of filtering, probably because of the upper airways. F3SD appeared efficient in eliminating aberrant data, while allowing the identification of physiological variations of Rrs.