Pulse Oximetry and Skin Tone in Children (POSTer-Child)

To the Editor:

Retrospective studies have found overestimation of oxygen saturation by pulse oximetry in adult and pediatric patients from races that may be associated with darker skin.¹ These retrospective studies share key limitations, including race as a poor surrogate for skin tone and paired pulse oximetry (SpO2) and arterial saturation (SaO2) measurements in the medical record. Limited prospective laboratory-based² and clinical studies^3,4 using measured skin tone in adults have found worse performance among patients with darker skin. There have been very limited prospective studies in children.⁵

We enrolled 320 patients under 21 years of age undergoing cardiac catheterization in 2024. Skin tone was measured using a spectrophotometer. SpO2 was recorded using two pulse oximeters (Nellcor and Masimo) at the exact time of fractional saturation measured by co-oximetry. Pulse oximetry bias (SpO2-SaO2), precision (standard deviation of bias), accuracy root mean square error (ARMS), and occult hypoxemia (SaO2<88% when SpO2≥92%) were calculated. Additional methodological details are available in the Supplementary Appendix. The population was similar to the US population but showed relative underrepresentation of Hispanic children (Table S4). A total of 48 (15.1%) patients identified as Black, and 44 (13.8%) identified as Hispanic.

Average bias for the Nellcor device was 1.32% and for the Masimo device was 1.88% (Table 1). Bias was higher among children with darker skin (ITA categories 5-6) compared to children with lighter skin (ITA categories 1-2) for both pulse oximeters (p<0.001). Precision and ARMS were also higher for children with darker skin. ARMS was substantially higher than the FDA cutoff of three for children in ITA category 5-6. Occult hypoxemia was present in 7.1% of children in ITA category 5-6 for the Nellcor device and 8.3% for the Masimo device. This was higher than the 0.0% and 3.4% seen, respectively, for children in ITA category 1-2.

Table 1:

Pulse oximeter performance by skin tone^*

	ITA 1-2 (Lightest)†	ITA 3-4	ITA 5-6 (Darkest)	Overall
Nellcor	n=132	n=112	n=75	n=319
Bias (%)	0.86 (0.44, 1.27)	1.12 (0.68, 1.55)	2.43 (1.64, 3.21)	1.32 (1.01, 1.62)
Precision (%)	2.41 (2.15, 2.74)	2.35 (2.08, 2.71)	3.47 (2.99, 4.13)	2.74 (2.54, 2.97)
ARMS	2.55 (2.14, 2.96)	2.59 (2.01, 3.17)	4.21 (3.34, 5.09)	3.04 (2.66 3.41)
Occult Hypoxemia‡	0/81 (0.0%)	2/84 (2.4%)	4/56 (7.1%)	6/221 (2.7%)
Masimo	n=126	n=108	n=69	n=303
Bias (%)	1.37 (0.85, 1.88)	1.34 (0.90, 1.79)	3.67 (2.74, 4.59)	1.88 (1.52, 2.24)
Precision (%)	2.97 (2.64, 3.38)	2.35 (2.08, 2.72)	3.89 (3.33, 4.68)	3.15 (2.92, 3.43)
ARMS	3.25 (2.59, 3.92)	2.70 (2.10, 3.30)	5.33 (4.17, 6.48)	3.67 (3.17, 4.17)
Occult Hypoxemia‡	3/88 (3.4%)	3/89 (3.4%)	5/60 (8.3%)	11/237 (4.6%)

^*Bias, precision, and ARMS are presented as value (95%CI); Occult hypoxemia presented as n/N (%)

^†ITA categories are as previously described by Del Bino et al.; 1=very light, 2=light, 3=intermediate, 4=tan, 5=brown, 6=dark

^‡SpO2≥92% when SaO2<88%; denominator is patients with SpO2≥92%

We report a prospective study in children measuring overestimation of saturation by pulse oximetry attributable to skin tone. This overestimation could lead to undertreatment of hypoxemia and contribute to racial inequities in outcomes. Additional studies are needed in poulations with greater proportion of individuals with darker skin tone. Our findings emphasize that current FDA guidance for pulse oximeter validation (https://www.fda.gov/media/72470/download?attachment), which recommends only limited involvement of darkly pigmented individuals without a formal definition, is inadequate in ensuring equity in pulse oximetry accuracy. New guidance is currently under consideration. Based on these results, there is potential for improvement in both bias and precision.