Evaluation of Correlation and Agreement between SpO2/FiO2 ratio and PaO2/FiO2 ratio in Neonates

Abstract

Objectives This article evaluates correlation and agreement between oxygen saturation (SpO2)/fraction of inspired oxygen (FiO2) (SF) ratio and partial pressure of oxygen (PaO2)/FiO2 (PF) ratio. It also derives and validates predictive PF ratio from noninvasive SF ratio measurements for clinically relevant PF ratios and derives SF ratio equivalent of PF ratio cutoffs used to define acute lung injury (ALI, PF < 300) and acute respiratory distress syndrome (ARDS, PF < 200). Methods Retrospective cohort study including neonates with respiratory failure over a 6-year study period. Correlation and agreement between PF ratio with SF ratio was analyzed by Pearson's correlation coefficient and Bland–Altman analysis. Generalized estimating equation was used to derive PF ratio from measured PF ratio and derive corresponding SF ratio for PF ratio cutoffs for ALI and ARDS. Results A total of 1,019 paired measurements from 196 neonates with mean 28 (± 4.7) weeks' gestational age and 925 (± 1111) g birth weight were analyzed. Strong correlation was noted between SF ratio and PF ratio (r = 0.90). Derived PF ratios from regression (1/PF = –0.0004304 + 2.0897987/SF) showed strong accuracy measures for PF ratio cutoffs < 200 (area under the curve [AUC]: 0.85) and < 100 (AUC: 0.92) with good agreement. Equivalent SF ratio to define ALI was < 450, moderate ARDS was < 355, and severe ARDS was < 220 with strong accuracy measures (AUC > 0.81, 0.84, and 0.93, respectively). Conclusion SF ratio correlated strongly with PF ratio with good agreement between derived PF ratio from noninvasive SpO2 source and measure PF ratio. Derived PF ratio may be useful to reliably assess severity of respiratory failure in neonates. Further studies are needed to validate SF ratio with clinical illness severity and outcomes.