A growing body of medical research has demonstrated the decreased accuracy of pulse oximetry in patients with dark skin tones, who may receive an overestimate of their true oxygen saturation through pulse oximetry and may experience worse clinical outcomes because of this. These are among the findings and conclusions of a literature review published in Annals of the American Thoracic Society.
A multidisciplinary team of researchers at the Veterans Affairs Puget Sound Healthcare System, Seattle, Washington, conducted an Evidence-Based Practice literature review of existing data on “the accuracy of pulse oximeters for individuals with dark skin tones as well as the clinical implications of device inaccuracy at both a patient and public health level.” The review, conducted in April and May 2021 in PubMed and the Cumulative Index to Nursing and Allied Health Literature, identified relevant research using search terms that included skin tone/pigmentation, pulse oximetry, SPO2, hypoxia, and racial bias.
The reviewers sought to quantify the degree of inaccuracy in pulse oximetry measurements in those with darker skin tones by measuring bias, which they defined mathematically as the “average difference between pulse oximeter-measured oxygen saturation (SpO2) and the true arterial saturation (SaO2)” in a particular individual. Bias can vary according to device, SaO2 range, and other factors; it may be a positive or negative value, the reviewers noted.
According to review authors, the US Food and Drug Administration (FDA) and other regulatory agencies use a pooled mean bias across all oxygen saturations (ARMS) as the performance benchmark for pulse oximeters. Based on the type of sensor used, an ARMS value of 3.0% to 3.5% or less is required. To date, the FDA does not require that ARMS be calculated separately by oxygen saturation range or skin tone categorization. “Notably, reliance on summary measures is a potential pitfall of current Food and Drug Administration (FDA) requirements for oximeters,” said review authors.
Understanding the effect of skin tone in pulse oximetry presents a challenge, the researchers noted. Published literature lacks a standardized approach to sampling, quantifying bias, defining skin tone, and reporting the results. Some studies do not report mean bias itself, but instead report the differences in mean bias or the rates of occult hypoxemia among racial groups. Some studies provide subgroup analyses based on oxygen saturation range, whereas other studies do not.
Further, studies vary according to clinical setting, pulse oximeter model, sensor type, and patient sample size. Additionally, although it is preferred to refer to the diversity of skin color using the language of dark or light skin tones, the categorization of individuals in the literature varies, with some studies categorizing patients by race, and others referring to dark or light “skin pigmentation,” graded subjectively. Variations in the practices used to categorize skin tones, study methodology, and study endpoints render direct comparison of these data difficult to carry out.
The researchers compiled several tables on: 1) studies of skin-tone-related pulse oximeter accuracy in nonclinical settings; 2) studies of skin-tone-related pulse oximeter accuracy in clinical settings, with calculations of mean bias; 3) studies of skin-tone-related pulse oximeter accuracy, in clinical settings, reporting rates of occult hypoxemia; 4) the hypothetical implications of SaO2 overestimation in patients with dark skin tones.
The researchers noted that no easy algorithm is available for use by practicing clinicians or health care facilities to correct pulse oximetry bias. The manufacturers of these devices and the FDA bear the responsibility of improving the performance of pulse oximeters. Current pulse oximeter approval criteria do not require robust data on individuals with dark skin tones, noted the reviewers.
The researchers concluded that “Inaccurate measurement of oxygen saturation in individuals with dark skin tones is one manifestation of structural racism in healthcare. Review of relevant literature indicates higher frequency of technical difficulties, increased mean bias, and higher rates of occult hypoxemia for subjects with dark skin tones, which in turn may be associated with less aggressive clinical management and increased mortality.”
Jamali H, Castillo LT, Morgan CC, et al. Racial disparity in oxygen saturation measurements by pulse oximetry: evidence and implications. Ann Am Thorac Soc. Published online September 27, 2022. doi:10.1513/AnnalsATS.202203-270CME