(335) A Continuous and Noninvasive Method to Estimate PaO2 and PF Ratio: A Brief Report

Introduction: PaO2 serves as the reference for hypoxemia evaluation, but it is scarcely obtain in the pediatric clinical setting as arterial blood gases are invasive and only allow for occasional assessments. This precludes its use for continuous evaluation of hypoxemia and for the creation of precise detection models. Our team previously developed and validated an equation using only noninvasive data to estimate the PaO2 and oxygenation index. This brief report aimed to validate this method to continuously estimate the PF ratio in critically ill pediatric patients.

Methods: We included any patients admitted in the PICU who had an ABG with coexisting continuous SpO2 values from May 2015 to May 2023. We used our previously tested mathematical model to determine the magnitude of hypoxemia by computing the estimated PaO2/FiO2 ratio (ePF) and comparing it to the PaO2/FiO2 (PF) ratio and the SpO2/ FiO2 (SF) ratio in hypoxemia classification.

Results: We analyzed 17,239 PaO2 where SpO2 was below or equal to 97%, including 13,5% of patient on non-invasive ventilation. Using the estimated PaO2 to estimate the PF ratio showed a significant better hypoxemia classification (kappa) than SF ratio (0.71 vs 0.50; p < 0.001), lower fixed bias (-16.61 vs -74.69; p < 0.001) and lower mean absolute error (61 vs 102.07; p < 0.001). Discrimination for severe hypoxemia (PF < 100) was comparable between the two methods with an AUROC (0.90 vs 0.89).

Conclusions: The estimated PaO2 using non-invasive data allows for a continuous estimation of PF ratio for SpO2 less than or equal to 97%. Estimation of PF ratio may provide decision support to assist in hypoxemia diagnosis and clinical care of critically ill patients.