Correlation between Pulse Oximetry Oxygen Saturation (SpO2) and Measured Arterial Oxygen Saturation (SaO2) and Arterial Oxygen Tension (PaO2) in Neonates

Abstract

Pulse oximetry has recently been introduced as a method for continuous and non-invasive transcutaneous monitoring of arterial oxygen saturation in neonates. The purpose of our study was to determine the reliability, accuracy, and practicability of pulse oximetry for monitoring arterial oxygenation in sick newborn infants. Pulse oximetry oxygen saturation values (SpO2) were compared with simultaneously measured arterial oxygen saturation values (Sa02) by a co-oximeter. The linear regression equation for 65 data pairs was Y=24.53 + 0.726X (r=0.90, p <0.01). The difference between measured Sa02 and pulse oximetry oxygen saturation (Sp02) was 0.44% ± 2.97% (mean ± S.D.). The linear regression equation comparing measured Sa02 minus pulse oximetry saturation (Sp02) was Y=-23.09 + 0.258X (r=0.58, p <0.01). This demonstrated an increasing error of pulse oximetry with decreasing Sa02. Pulse oximetry oxygen saturation (Sp02) also correlated well with measured arterial oxygen tension (Pa02), and the upper reach of the S-shaped hemoglobin-oxygen dissociation curve was clear. With a hypothetical alarm limit set at 92%, all hyperoxemic values (29/29) were correctly identified (sensitivity 100%). Of the 78 paired values, 33 were classified correctly as nonhyperoxemic (specificity 42%), and a total of 62 (29 + 33) out of the 107 paired values were ranged correctly (accuracy 58%). The accuracy increased significantly with an increase in alarm limit up to 94%, and it remained stable thereafter. We conclude that pulse oximetry is a simple, accurate, noninvasive, and continuous method of transcutaneously measuring arterial oxygen saturation, and hyperoxemia can be detected by pulse oximetry with a high sensitivity provided that prior to use, type-specific alarm limits are set.