In vitro validation of a metabolic monitor for gas exchange measurements in ventilated neonates

Objective: To evaluate the Datex Deltatrac II for measurements in neonates requiring mechanical ventilation. Design: Prospective laboratory evaluation, using a ventilated lung model an d gas injection. During simulation of 79 neonatal respiratory settings, ass essment of oxygen consumption (V(overdot)O-2), carbon dioxide production (V (overdot)CO2) and respiratory quotient (RQ) was compared to a reference met hod (mass spectrometry, wet gas spirometry) using the statistical method of Bland and Altman. Interventions: Respiratory variables, which may influence the accuracy and precision of gas exchange measurements, were varied within the following ra nges: inspired oxygen fraction (FIO2): 0.21-0.8, expired carbon dioxide fra ction ((FECO2) over bar) and inspiratory-expiratory oxygen fraction (DFO2): 0.0032-0.0256, expiratory flow rate: 1.0-2.5 l/min, inspiratory pressure: 10-55 mbar, respiratory rate 25-60/min, constant RQ of 1. This resulted in 79 tests with V(overdot)CO2 and V(overdot)O-2 ranging from 8-64 ml/min. Measurements and results. The coefficient of repeatability for ten single s ubsequent Deltatrac measurements was 8.09 ml/min for V(overdot)O-2 and 9.17 ml/min for V(overdot)CO2 compared to 2.02 ml/min and 0.90 ml/min for V(ove dot)O-2, and V(overdot)CO2 with repeated reference measurements. The coeffi cient of repeatability of the Deltatrac measurements improved considerably when means of subsequent 5 min intervals were compared: 0.68 ml/min for V(o verdot)O-2 and 0.28 ml/min for V(overdot)CO2. The difference between the tw o methods (Deltatrac-reference) was -3.8% (2 s: 11.4%) for V(overdot)O-2, 1 3.2 % (2s: 7.9%) for V(overdot)CO2 and 17.6% (2 s: 16.7%) for RQ. The agree ment between methods deteriorated with smaller ((FECO2) over bar or DFO2 an d increasing FIO2. Conclusions: Considering limits of agreement of less than +/- 20% as clinic ally acceptable, results for V(overdot)O-2 assessment indicate acceptable a ccuracy and precision whereas V(overdot)CO2 and RQ assessments exceed this limit. Limited accuracy and precision result from detection of CO2 followin g dilution of expiratory gases and increased sensitivity to error propagati on by Haldane equations due to the small differences between inspiratory an d expiratory gas fractions.