1030: OCCULT HYPOXEMIA AND PULSE OXIMETRY PERFORMANCE ACROSS SKIN PIGMENTATION GROUPS

Abstract

Introduction: Pulse oximetry accuracy reportedly varies by race (Henry et al. Crit Care Med 2022; Sjoding et al. N Engl J Med 2021). However, since race correlates weakly with skin pigmentation, it is more precise to evaluate pulse oximeter performance by skin pigmentation (He et al, J Am Acad Dermatol 2014). The purpose of this analysis was to assess Nellcor™ pulse oximetry performance across varying skin pigmentations for blood oxygen saturations between 80-92%, when occult hypoxemia, with errors of ≥4% occur between SpO2 and SaO2, can occur (Henry et al. Crit Care Med 2022). Methods: This retrospective analysis included Medtronic-sponsored healthy adult volunteer studies conducted between July 2003 and February 2017. In each study, a Nellcor™ pulse oximeter (MAXA, MAXN, MAXFAST, or FLEXMAX) measured arterial blood oxygenation, in comparison to co-oximeter-measured SaO2 (ISO 80601-2-61 aligned) during controlled hypoxia. Skin pigmentation was visually assessed using a 4-level scale: very light, olive, dark olive, and extremely dark. Results: 69 very light or olive volunteers (light skin pigmentation) and 19 dark olive or extremely dark volunteers (dark skin pigmentation) were included. For arterial oxygen saturations between 80-92%, the light and dark skin pigmentation groups had bias of 0.05% (95%CI 0.01-0.09) and 0.30% (95%CI 0.22-0.38); precision of 1.39% (95%CI 1.36-1.41) and 1.54% (95%CI 1.49-1.60); and root-mean square deviation (RMSD) of 1.39% and 1.57%. The percentage of data points with SpO2-SaO2 error ≥4%, the minimum required for occult hypoxemia, was 0.79% (N=40/5098) and 0.54% (N=8/1518) for the light and dark skin pigmentation groups. Conclusions: In both groups, RMSD remained within device specifications (±2), with increased error in individuals with dark skin pigmentation. For both skin pigmentation groups, observed errors of ≥4% occurred < 1% between 80-92% SpO2, whereas the literature has reported occult hypoxemia rates of 3.6%-6.6% (Henry et al. Crit Care Med 2022). This discrepancy may be due to differences in patient type, timing of SpO2/SaO2 readings, the amount of motion noise across studies, and the limited sample size in this retrospective analysis. Additional research is needed to characterize pulse oximetry performance and ensure accuracy across all skin pigmentations.