Accuracy of Gas Exchange Monitoring During Noninvasive Ventilation

Abstract

Background: Gas exchange monitoring by indirect calorimetry (IC) during noninvasive ventilation (NIV) is desirable but currently not available. Leaks around the mask preclude reliable measurements of carbon dioxide production (VCO₂) and oxygen consumption (VO₂) in this population. We aimed to examine the impact of system leaks and gas flows on the accuracy of gas exchange measurements during NIV using an in vitro metabolic simulation. Materials and Methods: We examined the agreement between VCO₂ and VO₂ measurements by IC (using a novel canopy device) and reference values generated during an in vitro metabolic simulation of NIV at room air. The flow rate of gas sampled by the IC device (V_IC) was set relative to the output flow of the ventilator (V_VENT) to obtain a range of sample factors (SF = V_IC/V_VENT). Linear regression was used to determine the effect of SF on the accuracy of the system. Results: An acceptable agreement between measured and reference values was observed, with mean bias (limits of agreement) of −3.3% (−6.9% to 0.3%) and −10.6% (−14.9% to −6.4%) for VCO₂ and VO₂, respectively. An SF of 1.25 was associated with the highest accuracy of measurement. VO₂ measurement accuracy deteriorated with system leak and at SF >1.25 and was linearly related to sample dilution by ambient air entrainment. Conclusions: A novel canopy device with titration of IC sample flow in relation to the ventilator flow allowed in vitro gas exchange measurements during simulated NIV with acceptable accuracy. This model needs to be tested in clinical settings.