FDA panel: Pulse oximeters show clear ‘disparate performance’ for darker skin pigmentation
Click Here to Manage Email Alerts
At a Medical Devices Advisory Committee meeting held Nov. 1, panelists recommended improvements to the FDA regarding pulse oximeters and their accuracy among patients with darker skin pigmentation.
“The panel believes that clearly there is a disparate performance in patients with darker skin pigment, and this does increase these patients’ risk for whatever disease they may be having,” Steven Nathan, MD, chair of the Anesthesiology and Respiratory Therapy Devices Panel of the FDA’s Medical Devices Advisory Committee, said during the meeting.
In February 2021, the FDA released a safety communication about limitations and possible inaccuracies of pulse oximeters “under certain conditions,” according to data provided by the FDA head of the meeting.
Additionally, the FDA noted that pulse oximeters have been granted a general indication for use as a tool for noninvasive measurement of blood oxygen saturation, with prescription devices undergoing clinical assessment to show accuracy; however, over-the-counter pulse oximetry devices have not been cleared for medical use.
The panel addressed four main questions posed by the FDA that addressed: accuracy of pulse oximeters among patients with darker skin pigmentation, tools to assess pigmentation, how accurate the FDA should require pulse oximeters to be and labeling of prescription and OTC pulse oximeter devices.
Accuracy in darker skin pigmentation
The panel reviewed several studies that demonstrated potential bias in the real-world use of pulse oximeters among patients with darker skin pigmentation.
One such study involved a meta-analysis of 32 studies published between 1985 to 2021. Shi and colleagues found that 15 studies that used skin pigmentation levels and 22 studies that used race/ethnicity demonstrated potential bias.
Despite the moderate to low certainty of the findings, researchers found that in individuals with high levels of skin pigmentation (pooled mean bias, 1.11%; 95% CI, 0.29%-1.93%) and those described as Black/African American (1.52%; 95% CI, 0.95%-2.09%), oxygen saturation was likely overestimated by pulse oximetry.
“Such a small overestimation may be crucial for some patients: particularly at the threshold that informs clinical decision-making,” Shi and colleagues wrote.
Specifically, given the fact that the devices may overestimate oxygenation among individuals with darker skin pigmentation, critical life-saving therapies and oxygen may be inappropriately withheld from this population, panelists said.
When addressing all the clinical evidence that they heard, the panel agreed that evidence suggests disparate performance of pulse oximetry in patients with darker skin tones.
“During the opening sessions, some of the makers of these devices pointed out imperfections in real-world clinical studies,” Jason Connor, PhD, president and lead statistical scientist at ConfluenceStat LLC, said. “We shouldn’t let some of the imperfections of those studies, which are really just real-world experience, negate from what is a clear signal with skin pigmentation leading to poor performance.”
However, a few panelists said that the accuracy issues found in the presented studies may be amplified by low perfusion, obesity and motion, and they heard data showing that dyshemoglobinemias, severe anemia, low perfusion, dyes, nail polish, ambient light and motion artifacts all may influence pulse oximeter accuracy in addition to skin pigmentation.
“At least the way I read the data, it’s not just the difference in skin pigmentation, but combined with a low perfusion index,” panelist Hugh A. Cassiere, MD, FCCP, FACP, director of critical care services at Sandra Atlas Bass Heart Hospital, said. “In other words, patients who are critically ill or sick in the hospital, by definition, have perfusion issues. This is amplified by the fact that there are also racial skin pigmentation issues that exacerbate it.”
Although the data presented focused on prescription-use devices, Nathan said such findings likely have “ramifications for outpatient OTC devices as well.”
“We would encourage the FDA to consider these other issues as well that can be addressed perhaps in some fashion at the same time,” he said.
Assessing pigmentation
The panelists heard evidence that self-identification of race in studies evaluating pulse oximeters should be considered but may not be the best tool on its own for assessing skin pigmentation.
“I don't like the idea of using self-identification of race,” panelist Richard D. Branson, MSc, RRT, professor of surgery and director of clinical research at the University of Cincinnati College of Medicine, said, adding that two individuals with the same self-identification can have significantly different skin colors.
The panel reviewed evidence on tools to assess skin pigmentation such as skin color scales, including the Fitzpatrick scale and von Luschan color scale, melanosome volume fraction, colorimetry and spectrophotometer.
When deliberating the FDA questions, the panel recommended that both objective and subjective measures be utilized when looking at skin pigmentation in relation to pulse oximeters.
“The panel generally believes that yes, there should be standardized skin pigmentation assessment, which should include an objective measure ideally, but can also include one of these more subjective scales,” Nathan said. “It should be easy enough to include both of them, that’s something companies can do to see what the correlation is, and that would move the field forward a little bit.
“Identified race should be included as part of the demographics captured, and there should be equal representation across the spectrum of skin pigmentation,” Nathan added.
Pulse oximetry accuracy
In addition to demonstrating basic safety, the FDA currently recommends the effectiveness of pulse oximeters be evaluated through “Arms,” which is the root mean square of pooled data pairs of SpO2 — the percent estimate of blood oxygenation provided by oximeters — and SaO2, or arterial blood oxygenation, considered the “gold standard” measurement. FDA allows for an Arms of 3%, indicating a roughly 68% probability that the SpO2 value is within 3% of the SaO2 value.
Overall, the panel recommended validating SpO2 measurements across the spectrum, from lower to higher readings, tightening the thresholds for error.
“I think the clinical data suggest strongly that an Arms between 1.5% and 2% for the [SpO2] range of 72% to 92%-94% would be significantly better clinically than an Arms of 3%, which is the current FDA standard,” panelist Michael F. O'Connor, MD, vice chair of academic/faculty affairs and faulty development at the University of Chicago, said. “We need to have data about positive predictive value for each saturation that pulse ox might report and a range of 95% CI.”
Acknowledging that pulse oximetry tends to perform worse in the context of lower oxygen saturations, assessment also should be based on clinical setting, according to Nathan.
“Healthy volunteers are one thing, but in a simulated hospital setting, it becomes very important. There needs to be validation across the SpO2 spectrum from 70% up to 100%,” he said.
Additionally, the FDA currently recommends that studies on pulse oximeters include at least two darkly pigmented individuals, or 15% of the study population.
“The criteria for developing a pulse oximeter should require a sufficiently large population of patients with different pigments, so that the 95% CIs for Caucasians and African Americans or dark-skinned people will be very similar,” O’Connor added.
“All pigmentation types should be represented equally,” Nathan said, adding that reporting on bias between skin pigmentation types alone would be insufficient. “I would guard against the label having bias Arms that become uninterpretable to the average clinician.”
Labeling recommendations
During their deliberations, the panelists discussed several issues revolving around the labeling of pulse oximeters.
“If I go to a pharmacy and buy a pulse oximeter that gives me a digital reading, I'm going to assume that it's accurate, despite what you put on the box and despite what you put on the machine itself,” panelist Sean Hennessy, PharmD, PhD, founding director of the Center for Pharmacoepidemiology Research and Training, said. “I think it sends a mixed message to allow companies to sell something yet to not have confidence in the results ... particularly because we know it's being used for health reasons.”
For prescription pulse oximetry devices, the panel recommended that the label defer to the prescribing physician on appropriate usage and limitations.
For OTC pulse oximetry devices, the panel agreed that labels should be easy to understand, with transparency for any issues surrounding inaccuracy and what factors may influence the accuracy of readings. However, some members thought that because OTC devices are not as regulated by the FDA, they should be clearly labeled as not intended for medical use. Not all panel members agreed with this idea.
“I can’t say I completely agree ... [that OTC devices] should just say, ‘Not for medical use,’ because ... we’ve already said these are being used for medical use; no one is really using these as a toy or a form of entertainment,” panelist Murad Alam, MD, professor of medical social science at Northwestern University, said. “To the extent that they’re not to be used by physicians in lieu of prescription products, that’s certainly true, but that’s not a message for the general public who don’t even really know what ‘medical use’ means. Since these devices are going to be continued to be used as a lay public as an at-home screening device, I think we should give more detailed instructions about what they’re supposed to do with them.”
Alam recommended that the FDA consider implementing benchmarks for accuracy for OTC devices.
This would “allow the FDA ability to investigate these products to give consumers some guidance on how to use them” as well as give the FDA power to “at the very least compel the manufacturers to convey that simplified and expanded labeling to patients who will be in fact using these to try to save their own lives,” he said.
References:
- FDA executive summary: Review of pulse oximeters and factors that can impact their accuracy. https://www.fda.gov/media/162709/download. Accessed Nov. 2, 2022.
- Shi C, et al. BMC Med. 2022;doi:10.1186/s12916-022-02452-8.