A nonlinear regression analysis of nonlinear, passive-deflation flow-volume plots

When studying lung mechanics of intubated premature infants, by the passive‐deflation technique we noted that in many flow‐volume plots the descending limb was curvilinear with a convexity toward the volume axis. By conventional linear analysis lung mechanics of these patients did not change after the administration of terbutaline, but Mead's tangent‐chord slope ratio method for quantifying the amount of curvature showed that the shape of the flow‐volume plots did change. Because of the limitations of this method, we developed a microcomputer based, reiterative regression algorithm which optimizes a nonlinear function for the best fit to any given set of data. We then studied six very low birth weight infants with clinical evidence of pulmonary gas trapping (weight at study, 1.22 ± 0.29 kg; age, 26 ± 16 days). We measured respiratory system resistance (Rrs), compliance (Crs), and expiratory time constants (TCrs), by the passive deflation technique before and after subcutaneous administration of 0.02 mg/kg of terbutaline. No effect of terbutaline in a dose sufficient to increase heart rate > 25 beatshin was observed. The same data analyzed using the nonlinear regression technique with a function based upon a two compartment model of parallel inhomogeneities revealed one compartment with relatively normal Rrs, Crs, and TCrs, values, and a second compartment with a very high Rrs. The latter fell by 50% after terbutaline. These data suggest that abnormalities of airway resistance in ventilated preterm infants are not easily identifiable by classic linear analysis of lung mechanics. We propose that in such patients the two compartment analysis of passive exhalation mechanics measurements can provide better information about Rrs. © 1993 Wiley‐Liss, Inc.