Racial, ethnic biases in pulse oximetry accuracy may have delayed COVID-19 treatment for some

Racial and ethnic biases in pulse oximetry accuracy may have resulted in delays and unrecognized eligibility for COVID-19 therapies among Black and Hispanic patients, according to results of a cohort study in JAMA Internal Medicine.

“The COVID-19 pandemic shone a spotlight on pulse oximeters as a useful tool for making triage and treatment decisions, which begged the question of whether the previously reported inaccuracy of pulse oximeters in racial and ethnic minorities may have impacted how these patients were treated,” Ashraf Fawzy, MD, MPH, assistant professor of medicine in the division of pulmonary and critical care medicine at Johns Hopkins University School of Medicine, told Healio. “Our study provides new evidence that the inaccuracy of pulse oximeters in patients of color may have led to a delay in those patients receiving important and potentially lifesaving treatments for COVID-19.”

The retrospective cohort study included data from 7,126 patients with COVID-19 (41.7% women) from five referral centers and community hospitals in the Johns Hopkins Health System who had COVID-19 and self-identified as Black (39.3%), Hispanic (17.7%), Asian (5.2%) or white (37.8%). Fawzy and colleagues analyzed SpO₂ measurements and SaO₂ values from arterial blood gases in the patients’ electronic medical records. Occult hypoxemia was defined as an SaO₂ less than 88% with an SpO₂ of 92% to 96%. In patients with concurrent SpO₂ and SaO₂ measurements, the proportion with occult hypoxemia were compared by race and ethnicity.

The primary outcome was the difference between SpO₂ and SaO₂ measurements occurring within 10 minutes of one another. The secondary outcome was the difference in time to recognition of treatment eligibility with oxygen threshold-specific COVID-19 therapies between racial and ethnic underrepresented groups and white patients with COVID-19.

Overall, 1,216 patients had 32,282 concurrently measured SpO₂ and SaO₂ levels.

Occult hypoxemia occurred in 30.2% of Asian patients, 28.5% of Black patients and 29.8% of non-Black Hispanic patients compared with 17.2% of white patients.

According to the researchers, SpO₂ overestimated SaO₂ by 1.7% among Asian patients, 1.2% among Black patients and 1.1% among non-Black Hispanic patients compared with white patients. In addition, compared with white patients, among the 1,903 patients with predicted SaO₂ levels of 94% or less before an SpO₂ level of 94% or less or oxygen treatment initiation, the probability of treatment eligibility recognition was 29% lower among Black patients (HR = 0.71; 95% CI, 0.63-0.8) and 23% lower among non-Black Hispanic patients (HR = 0.77; 95% CI, 0.66-0.89).

Among 451 patients in whom treatment eligibility was not recognized, 54.8% were Black. Among the 1,452 patients in whom treatment eligibility was eventually recognized, Black patients experienced a median delay of 1 hour (P = .01) longer than white patients.

“The inaccuracy of pulse oximeters in persons of color has been reported as far back as 1990 and has been replicated by several recent studies. We also know that Black and Hispanic patients have had poorer outcomes after COVID-19, so it was not particularly surprising that inaccuracies in a tool that we relied on heavily may have led to delays in care,” Fawzy told Healio. “That may be one piece of the puzzle that explains why these patients had poorer outcomes through studies investigating that association are necessary.”

The researchers reported no significant median difference in treatment eligibility delay between individuals of other racial/ethnic underrepresented groups and white patients.

“Future research needs to investigate the underlying reasons why we are finding this inaccuracy in pulse oximetry among racial and ethnic minority patients,” Fawzy said. “Ultimately, the medical community will need to reengineer the pulse oximeters so they work more equitably for all patients.”

For more information:

Ashraf Fawzy, MD, MPH, can be reached at afawzy1@jhmi.edu.