Neonatal Pulse Oximetry Accuracy and Disparities by Skin Pigmentation (NeoPODS): A Prospective Study.

Abstract

To assess pulse oximetry saturation (SpO₂₎ to arterial oxygen saturation (SaO₂) bias relative to skin pigmentation in newborns and to address potentially occult hypoxemia, defined as SaO₂ lower than SpO₂. We conducted a prospective, diagnostic accuracy study at 2 tertiary neonatal intensive care units between 2022-2025 to assess SpO₂-SaO₂ differences in relation to skin pigmentation in newborns. Newborns younger than 10 postnatal days with an arterial catheter undergoing an arterial blood gas collection were enrolled. A pulse oximeter was placed on the extremity corresponding to the arterial sampling site, and SpO₂ values were recorded during blood gas collection. Skin pigmentation was measured noninvasively using individual typology angle (ITA), melanin index, and visual assessments. Mean SpO₂ was calculated from the 30 seconds immediately preceding arterial sampling or from the most stable 30-second average when timestamps were unavailable and compared with SaO₂ to calculate SpO₂-SaO₂ bias. A total of 136 matched SpO₂-SaO₂ pairs from 70 patients were analyzed, of whom 40% were Black. The overall mean bias was -0·98 ± 2·8% (95% CI -1·4 to -0·52), indicating an underestimation of SaO₂ by SpO₂. Across categorical skin-pigmentation measures and race, no statistically significant differences in SpO₂-SaO₂ bias were observed. When analyzed continuously, bias became less negative with lighter skin pigmentation for both ITA and melanin index, reaching significance only for ITA when analyses were restricted to the first measurement per patient (slope +0·016 per ITA unit, 95% CI 0·001-0·032, p=0·041). Occult hypoxemia was rare, occurring in one newborn with light skin pigmentation. In this prospective neonatal study with tightly paired measurements, SpO₂ slightly underestimated SaO₂ but did not demonstrate clinically significant disparities across skin pigmentation.