FDA Advisory Panel Generally Pleased With Pulse Oximetry Clinical Trial Proposal

An FDA advisory committee gave generally high marks to the agency’s latest effort to improve the accuracy of pulse oximeters when used in darker-skinned patients, but raised questions about some of the FDA’s proposed premarket clinical trial guidance for pulse oximeter manufacturers.

“I really applaud the efforts of the FDA to reduce disparate bias in pulse oximetry,” said Julian Goldman, MD, an anesthesiologist at Massachusetts General Hospital in Boston. Goldman, a member of the FDA Medical Device Advisory Committee Anesthesiology and Respiratory Therapy Devices Panel, was speaking at Friday’s panel meeting. “This has clearly been a very heavy lift for a long time, and we need to address it.”

Building on Prior Work

The panel meeting built on previous work the FDA has done on this issue. At a November 2022 meeting of the same advisory panel, members agreed that pulse oximeters are less accurate in patients with darker skin, and that the FDA should notify patients and providers about the issue and recommend that manufacturers correct the discrepancy.

FDA officials are now considering updating their 2013 guidance to pulse oximeter manufacturers and, as part of that effort, they asked the panel to weigh in on the design of a premarket clinical trial that pulse oximeter manufacturers could perform. The trial would be designed “to improve the quality of premarket studies to evaluate the performance of pulse oximeters, taking into consideration a patient’s skin pigmentation, and patient-reported race and ethnicity,” according to an FDA briefing document.

At Friday’s meeting, FDA officials outlined their proposed new guidance. Kumudhini Hendrix, MD, of the FDA Center for Devices and Radiological Health, noted that under the current guidance for premarket clinical testing, the agency recommends that test participants be “healthy volunteers selected from a pool of limited volunteers. Sample size is not large, typically 10 to 20 subjects. Skin pigmentation is generally qualified subjectively, such as ‘light,’ ‘medium,’ or ‘dark,’ and is not specific to any particular anatomical sites — for example, dorsum of the hand or ventral aspect of the forearm … Importantly, [the trial] is not powered to determine significant differences between cohorts — for example, pigmentation levels.”

Bigger Trial Recommended

Under the proposed new guidance, she continued, the FDA is recommending that the trial include “at least 24 healthy subjects with a minimum of 480 data pairs, and we recommend that at least 40% of participants be of each gender.” In terms of determining pigmentation, the FDA is proposing using both the Monk Skin Tone (MST) assessment — a well-known subjective pigmentation scale that includes 10 different skin tones — and the Individual Typology Angle (ITA), which is an objective pigmentation measure taken at the sensor site.

The agency also proposes tightening the acceptable boundaries for variance in the oxygen measurements. Under the proposed approach, the agency recommends that the estimate of the absolute difference in SpO₂ bias — defined as the mean of the difference when SaO₂ is subtracted from SpO₂ — across ITA and MST levels be less than 1.5% when SaO2 is greater than 85%, and less than 3.5% when SaO2 is greater than 70% and less than or equal to 85%.”

That last recommendation was concerning to some members. “The saturation absolute difference just seems too wide for me if it’s less than 3.5% in the lower saturation range,” said panel member Thomas Wiswell, MD, a neonatologist at the Kaiser Permanente Moanalua Medical Center in Honolulu, “because I have patients that are not infrequently in the 70% to 85% range, and sometimes have wide variations of the saturations that I’m measuring with blood samples …. Personally, I would like to see that number go down and be — unless there’s a big technological problem — closer to 1.5%, or maybe a 2% absolute difference.”

There were also some concerns about the proposed size of the trial. “Honestly, I think doubling the sample size would be a benefit — I think 24 is the bare minimum,” said Rachel Brummert, MS, communications lead at the American Society of Pharmacovigilance, in Charlotte, North Carolina and the panel’s consumer advocate. “I think we have a responsibility as a panel to get as much information as we can, and I think we can accomplish that by at least doubling it.”

Wiswell agreed. “The total ‘N’ [number] of 24 patients just seems low to me, keeping in mind that these are healthy volunteers [which is] not who the devices are going to be used on,” he said. “I would like higher numbers, recognizing that underpowered studies can lead you astray.”

William Wilson, MD, MA, pointed out that a study with 24 patients would cost about $250,000 — “about twice what it cost before,” when 10 participants was the required minimum. Wilson, who is executive vice president for clinical operations at Masimo — a pulse oximeter maker in Irvine, California — and the panel’s industry representative, added, however, that “that’s okay because we need to make sure that we have representative samples. The most important factor is ensuring that we have an almost equal number of dark-skinned and light-skinned [participants] and we cover the full spectrum of individuals … Those factors will markedly improve the data.”

Praise for Proposed Skin Tone Measurements

The proposed use of the MST and ITA drew mostly positive comments. During the meeting’s public comment period, Steven Barker, MD, PhD, chief science officer at Masimo, said that “the use of the MST scale for initial assessment of skin pigmentation followed by an objective measure using the individual typology angle ITA is satisfactory. The MST provides good resolution and range of skin tones, from very light to very dark. The ITA provides an objective measure to detect subjective bias in the assessment of skin tone.”

However, he added, the FDA should stratify the MST skin pigments — which range from 1 being the very lightest to 10 being the darkest — into three groups: 1 to 4, 5 to 7, and 8 to 10. “We agree that at least 25% of participants should come from each MST [group],” Barker said. “Balancing these cohorts will help avoid under- or over- weighting of a particular cohort or skin tone range.” The agency also should require at least one participant in the study with MST values of 2 and 9, he said.

Panel member Raymond Lanzafame, MD, MBA, a general surgeon in Rochester, New York, agreed that ” there certainly is an advantage to using both the MST and the ITA approaches, as well as gaining subjective information regarding race and ethnicity.” He added, however, that “inclusion of a greater proportion of individuals at at the darker MST levels is is something that should be seriously considered.”

The meeting did not include any questions that required a panel vote. The FDA isn’t required to follow its advisory committees’ recommendations, but often does.