PaCO2 and alveolar dead space are more relevant than PaO2/FiO(2) ratio in monitoring the respiratory response to prone position in ARDS patients: a physiological study

Introduction: Our aims in this study were to report changes in the ratio of alveolar dead space to tidal volume (VDalv/V-T) in the prone position (PP) and to test whether changes in partial pressure of arterial CO2 (PaCO2) may be more relevant than changes in the ratio of partial pressure of arterial O-2 to fraction of inspired O-2 (PaO2/FiO(2)) in defining the respiratory response to PP. We also aimed to validate a recently proposed method of estimation of the physiological dead space (VDphysiol/V-T) without measurement of expired CO2. Methods: Thirteen patients with a PaO2/FiO(2) ratio < 100 mmHg were included in the study. Plateau pressure (Pplat), positive end-expiratory pressure (PEEP), blood gas analysis and expiratory CO2 were recorded with patients in the supine position and after 3, 6, 9, 12 and 15 hours in the PP. Responders to PP were defined after 15 hours of PP either by an increase in PaO2/FiO(2) ratio > 20 mmHg or by a decrease in PaCO2 > 2 mmHg. Estimated and measured VDphysiol/V-T ratios were compared. Results: PP induced a decrease in Pplat, PaCO2 and VDalv/V-T ratio and increases in PaO2/FiO(2) ratios and compliance of the respiratory system (Crs). Maximal changes were observed after six to nine hours. Changes in VDalv/V-T were correlated with changes in Crs, but not with changes in PaO2/FiO(2) ratios. When the response was defined by PaO2/FiO(2) ratio, no significant differences in Pplat, PaCO2 or VDalv/VT alterations between responders (n = 7) and nonresponders (n = 6) were observed. When the response was defined by PaCO2, four patients were differently classified, and responders (n = 7) had a greater decrease in VDalv/V-T ratio and in Pplat and a greater increase in PaO2/FiO(2) ratio and in Crs than nonresponders (n = 6). Estimated VDphysiol/V-T ratios significantly underestimated measured VDphysiol/V-T ratios (concordance correlation coefficient 0.19 (interquartile ranges 0.091 to 0.28)), whereas changes during PP were more reliable (concordance correlation coefficient 0.51 (0.32 to 0.66)). Conclusions: PP induced a decrease in VDalv/V-T ratio and an improvement in respiratory mechanics. The respiratory response to PP appeared more relevant when PaCO2 rather than the PaO2/FiO(2) ratio was used. Estimated VDphysiol/V-T ratios systematically underestimated measured VDphysiol/V-T ratios.