Accuracy of feedback-controlled oxygen delivery into a closed anaesthesia circuit for measurement of oxygen consumption

Abstract

Oxygen consumption (V*O2) is rarely measured during anaesthesia, probably because of technical difficulties. Theoretically, oxygen delivery into a closed anaesthesia circuit (V*O2-PF; PhysioFlex Draeger Medical Company, Germany) should measure V*O2. We aimed to measure V*O2-PF in vitro and in vivo. Three sets of experiments were performed. V*O2-PF was assessed with five values of V*O2 (0-300 ml min(-1)) simulated by a calibrated lung model (V*O2-Model) at five values of FIO2 (0.25-0.85). The time taken for V*O2-PF to respond to changes in V*O2-Model gave a measure of dynamic performance. In six healthy anaesthetized dogs we compared V*O2-PF with V*O2 measured by the Fick method (V*O2-Fick) during ventilation with nine values of FIO2 (0.21-1.00). V*O2-PF and V*O2-Fick were also compared in three dogs when V*O2 was changed pharmacologically [102 (SD 14), 121 (17) and 200 (57) ml min(-1)]. In patients during surgery, we measured V*O2-PF and V*O2-Fick simultaneously after induction of anaesthesia (n=21) and during surgery (n=17) (FIO2 0.3-0.5). Compared with V*O2-Model, V*O2-PF values varied from time to time so that averaging over 10 min is recommended. Furthermore, at an FIO2 >0.8, V*O2-PF always overestimated V*O2. With FIO2 <0.8, averaged V*O2-PF corresponded to V*O2-Model and adapted rapidly to changes. Averaged V*O2-PF also corresponded to V*O2-Fick in dogs at FIO2 <0.8. V*O2 measured by the two methods gave similar results when V*O2 was changed pharmacologically. In contrast, V*O2-PF systematically overestimated V*O2-Fick in patients by 52 (SD 40) ml min-1 and this bias increased with smaller arteriovenous differences in oxygen content. V*O2-PF measures V*O2 adequately within specific conditions.