Abstract
To assess accuracy of prediction of oxygenation index (OI=mean airway pressure*FiO2*100÷PaO2) with a non-invasive alternative, oxygen saturation index (OSI= FiO2*100÷SpO2) in a prospectively collected multicenter cohort of children with congenital diaphragmatic hernia (CDH) with substantial degree of hypoxemia (OI≥10 or OSI≥5).
This secondary analysis of the Milrinone in CDH Trial studied 61 subjects with OI≥10 or OSI≥5 at randomization. Prospectively collected arterial blood gas (ABG) data with PaO2 and preductal and postductal SpO2 were used to compare all OI and preductal OSI, all OI and postductal OSI, preductal OI/OSI only and postductal OI/OSI, by simple linear and quadratic regression modelling.
Indwelling arterial lines were present in 61 of 66 randomized subjects (51 as umbilical arterial lines). There were 572 matched pairs of PaO2-SpO2 data. Repeated measures correlation (95% CI) between all OI and preductal OSI by linear regression was 0.82 (0.79,0.85; n=61 patients, 572 samples), and between OI and OSI with SpO2 matched from the same preductal or postductal sites as the ABG were 0.97 (9.91,0.89) and 0.86 (0.84,0.89) respectively. The relationship between preductal OI and OSI was described best by the quadratic equation, OI=0.3*OSI+0.1*OSI2+2.4 (4 patients, 22 samples) and postductal OI and OSI by OI=0.6*OSI +0.08*OSI2+2.6 (54 patients, 477 samples). There was no difference in quasi-likelihood under the independence model information criterion (QlCu) between linear and quadratic models (461 vs 462 respectively). Use of site-specific (preductal vs postductal) quadratic equations captured the non-linear relationship but did not improve QlCu.
Non-invasive OSI values correlate well with OI. The use of OSI based on preductal SpO2 may be a suitable alternate strategy for clinical trials even if post-ductal arterial access is available.