Objective: Linear modeling as a method of exploring respiratory mechanics during mechanical ventilation, was compared to nonlinear modeling for flow dependence of resistance in three distinct groups of patients, those with: (a) normal respiratory function (NRF), (b) chronic obstructive pulmonary disease (COPD), or (c) adult respiratory distress syndrome (ARDS).
Design and patients: Airways opening pressure (Pao), flow (V'), and volume (V) signals were recorded in 32 ICU mechanically ventilated patients, under sedation and muscle relaxation (10 NRF, 11 COPD, 11 ARDS). All patients were ventilated with controlled mandatory ventilation mode at three levels of end-expiratory pressure (PEEPe): 0, 5, and 10 hPa. Data were analyzed according to: (a) Pao = PE + Ers V + Rrs V' and (b) Pao = PE + Ers V + k1V' + k2¿V'¿V'; where Ers and Rrs represent the intubated respiratory system (RS) elastance and resistance, k1 and k2 the linear and the nonlinear RS resistive coefficients, and PE the end-expiratory pressure. The model's goodness of fit to the data was evaluated by the root mean square difference of predicted minus measured Pao values.
Results: NRF data fit both models well at all PEEPe levels. ARDS and particularly COPD data fit the nonlinear model better. Values of k2 were often negative in COPD and ARDS groups, and they increased in parallel with PEEPe. A gradual increase in PEEPe resulted in better fit of ARDS and COPD data to both models.
Conclusions: The model of V' dependence of resistance is more suitable for the ARDS and particularly the COPD groups. PEEP tends to diminish the V' dependence of respiratory resistance during the respiratory cycle, particularly in the COPD group, probably through an indirect effect of the increased lung volume.