Accurate pulse oximetry ‘challenging’ in critically ill patients; skin pigmentation a factor
Key takeaways:
- Real-world study looked at relationship between skin pigmentation and pulse oximeter bias in critically ill patients.
- Ability to detect occult hypoxemia may be complicated in patients with darker skin pigmentation.
CHICAGO — In a new study, pulse oximeters underestimated arterial oxygen saturation among patients in the ICU, and the ability of the devices to detect hypoxemia was especially hindered in patients with dark skin pigmentation.
The results of EquiOx, a trial designed to assess differences in pulse oximeter performance across a range of objective and subjective skin pigmentations, were presented at the American College of Cardiology Scientific Session.
“We’ve known for decades that these devices performed differently across races and across skin pigment groups,” Carolyn Hendrickson, MD, associate professor of medicine at the University of California, San Francisco, said during a press conference.
Hendrickson cited a 2020 letter to the editor and retrospective analysis published in The New England Journal of Medicine, in which researchers at the University of Michigan showed that occult hypoxemia was undetected by pulse oximetry in Black patients at a rate threefold higher than among white patients.
“This spawned more research in the subsequent 2 years than had been done in the previous 2 decades. Although we did know about this problem before, there were further results that showed these differences were linked to health care disparities,” Hendrickson said. “We designed a prospective observational study with research staff at the bedside directly observing the readings on the pulse oximeter and the blood gas results.”
The EquiOx trial tested the skin pigment hypothesis — different from the race hypothesis of pulse oximeter bias, Hendrickson said — and utilized objective data from spectrophotometry-measured melanin content and the subjective Monk Skin Tone Scale.
The trial included 631 patients (mean age, 62 years) in the ICU at a level 1 trauma center. Seventy-six percent of patients required mechanical ventilation in the ICU and 57% required vasopressors. The median perfusion index was 1 and the median oxygen saturation was 98%. In total, the researchers compiled 1,760 total bedside observations of pulse oximetry and compared those with functional oxygen saturation from blood gas analysis.
Overall, 20% of the patients were self-identified Asian, 21% Black, 20% Hispanic/Latino, 25% white and 12% other. Spectrophotometry identified light skin pigment in 33% of patients, medium pigmentation in 53% and dark pigmentation in 14%.
The researchers reported a negative pulse oximetry bias across all skin pigmentations compared with blood gas analysis (median bias, 1.7%; interquartile ratio, –2.8% to –0.5%; average root mean square, 3.9; 95% CI, 3.2-4.6) and noted that bias was less negative in those with dark skin pigmentation. Hendrickson said negative bias is a pulse oximeter reading that may be too low.
Twenty percent of observations were of positive bias, which Hendrickson said indicates a pulse oximeter reading that may be too high. Hendrickson said positive bias may be falsely reassuring; in retrospective studies, there has been more positive bias in Black patients compared with white patients. Factors associated with positive error vs. nonpositive error included Black race, dark skin pigmentation, diabetes, hypertension, peripheral vascular disease, tobacco use, hypoxemia, perfusion index and hypothermia.
“We did find differences in the way the devices performed across skin pigment,” Hendrickson said during the press conference. “Everybody had negative bias, meaning the pulse oximeters read too low ... which protects against missed hypoxemia. The median bias across all groups was negative, but the bias was smaller in magnitude in the patients with dark skin pigment, meaning they had less protection. So, this relationship between Black and white patients is the same as previously published data.”
The adjusted mean bias for white patients’ oxygen saturation (–2.4%) was more negative compared with Black patients (mean difference, 1%; P < .001), Asian patients (mean difference, 1%; P < .001) and Hispanic patients (mean difference, 0.6%; P < .001), according to the presentation.
“It’s really challenging for bedside ICU clinicians to know when a pulse oximeter may not be accurate,” Hendrickson said. “In this real-world study, critically ill patients had worse pulse oximeter performance than the validation studies that are recommended by the FDA. ... These differences that we observed in the relatively normal ranges of oxygenation are probably amplified in the lower saturations, which could explain the difference between our study and the retrospective studies.”
Reference:
Collapse
Hendrickson CM, et al. Late-breaking clinical science IV. Presented at: American College of Cardiology Scientific Session; March 29-31, 2025; Chicago (hybrid meeting).
