Accuracy of PaO2 /FiO2 calculated from SpO2 for severity assessment in ED patients with pneumonia.

Abstract

Assessment of oxygenation in patients with community-acquired pneumonia is critical for treatment. The accuracy of percutaneous oxygen saturation (SpO2 ) determined by pulse oximetry is uncertain, and it has limited value in patients receiving supplemental oxygen. We hypothesized that calculation of partial arterial oxygen concentration/inspired oxygen faction (PaO2 /FiO2 ) from SpO2 by the Ellis or Rice equations might adequately correlate with PaO2 /FiO2 measured by arterial blood gases. We studied 1004 patients with pneumonia in the emergency department with simultaneous measurement of SpO2 and PaO2 from two cohorts from Valencia, Spain and Utah, USA. We compared SpO2 with measured SaO2 , compared the equations' accuracy in calculating PaO2 /FiO2 and determined how often patients would be misclassified at clinically important thresholds. We compared estimated PaO2 /FiO2 to measured PaO2 /FiO2 using the Spearman correlation. Pairwise correlation of SpO2 with SaO2 was moderate (rho = 0.66; P < 0.01). Both equations performed similarly among patients with lower PaO2 /FiO2 ratios. The Ellis equation estimated PaO2 /FiO2 from SpO2 more accurately than the Rice equation in patients with PaO2 /FiO2 ≥200. Simple agreement between calculated and measured P/F was 91% and 92%, respectively. The Ellis equation was more accurate than the Rice equation for estimating PaO2 /FiO2 , especially at higher levels of P/F ratio. Estimation of PaO2 /FiO2 from SpO2 is accurate enough for initial oxygenation assessment. Ellis and Rice equations could misclassify 20% and 30% of patients, respectively, at higher levels of PaO2 /FiO2 . For patients with abnormal oxygenation falling near thresholds for clinical decision making, arterial blood gas measurement preferably on room air is more accurate.