Evaluating predictive equations for deriving PaO 2 from SpO 2 in neonatal intensive care patients Joseph McDonell 1 , Trassanee Chatmethakul 3 , Phillip Levy 4 , Patrick J. McNamara 3 , Michael Tomasson 2 , Craig Wheeler 4 , Melissa L. Bates 1,2,3 Departments of

Abstract

The oxygen-hemoglobin dissociation curve can be used to calculate the arterial partial pressure of oxygen (PaO2) from the peripheral saturation (SpO2)in adults. Our objective was to evaluate whether this calculation could be performed in neonates. We obtained blood gas measurements from 104 de-identified infants at the University of Iowa Hospitals and Clinics (UIHC) neonatal intensive care unit. We plotted PaO2 against SaO2, measured by co-oximetry, and compared the resultant curve to published curves for adults and neonates. The UIHC curve was leftward shifted compared to published curves for neonates and adults. We tested the additional hypothesis that this because of the large proportion of premature infants (72.3%) in our cohort, but gestational age was not a predictor of the shift. Next, we quantified the degree to which pH, PaCO2, and gestational age contributed to sources of error in using this curve to estimate PaO2 from SpO2. We found that deviations in pH and PaCO2 from the physiological ideal contribute up to 3 mmHg and 4 mmHg of error, respectively, to the calculation of PaO2 using our cohort's curve. We demonstrate that data from previously published cohorts may not be universally useful for the calculation of PaO2. In future studies, we will validate the UIHC curve in additional patients with well characterized clinical phenotypes. Additional simulations are needed that quantify the additive contribution of pH, PaCO2, and gestational age to the error in calculated PaO2 in neonates.