Episode #46 Pulse Oximetry Accuracy and Skin Tone

Hello and welcome back to the Anesthesia Patient Safety Podcast.  My name is Alli Bechtel and I am your host. Thank you for joining us for another show. Today, we are going to review an important article that dropped on the APSF website in the Articles Between Issues Segment on March 2, 2021. Let’s see if you can find the article and can you recognize this sound?

[pulse oximeter sounds]

Before we dive into today’s episode, we’d like to recognize Merck, a major corporate supporter of APSF. Merck has generously provided unrestricted support to further our vision that “no one shall be harmed by anesthesia care”. Thank you, Merck – we wouldn’t be able to do all that we do without you!”

If you haven’t found the article yet, head over to APSF.org and click on the Newsletter heading. Second one down is Articles Between Issues. The articles are listed by their date of publication to the website with the newest articles on top. Scroll down until you see March 2021 and our featured article is, “APSF Statement on Pulse Oximetry and Skin Tone.” This statement by the APSF follows the December 2020 New England Journal of Medicine publication by Sjoding and colleagues called “Racial Bias in Pulse Oximetry Measurement.” I will include a link to the article in the show notes. The references include the studies that we will be talking about today as well.

Before we get into the New England Journal publication and the APSF Statement, we are going to hear from one of the APSF Board of Directors members, Dr. Meghan Lane-Fall. I will let her introduce herself.

[Lane-Fall] Hi, my name is Meghan Lane-Fall. I am the vice chair of inclusion, diversity, and equity and the David E Longnecker Associate Professor of anesthesiology and critical care at the University of Pennsylvania and a member of the board of directors of the APSF.

[Bechtel] I reached out to Lane-Fall and asked her what got her interested in this topic. Let’s take a listen to what she had to say.

[Lane-Fall] So, 2020 was a trying year for a lot of reasons. Clearly Covid was a challenge for all of us but also because of the racial reckoning that happened in the United States, especially my experience as a black woman. It was a really, really tough year and I remember reading the New England Journal article and of course the title caught my eye…that there was racial bias in pulse oximetry. I sort of had my antennae up for papers related to racial bias because of my role in inclusion, diversity, and equity in my department. I also noticed this article because it was co-authored by someone who I considered to be a friend, Jack Iwashyna, who is an outstanding researcher at the University of Michigan. I remembered reading the paper and being profoundly disappointed that there appeared to be a measurement bias in this technology that we use every day and that we take for granted.

I think the thing that additionally caught my attention is that they referenced an article from the journal, Anesthesiology, from the 2000s and the idea that this measurement bias was known and I had no idea. Right, you know, I was an anesthesia resident when this article came out. This was never something that we talked about. The idea that the pulse oximeter could be biased in the first place was something that I think was probably more billed as a theoretical concern more than anything else. So, when I read that this foundational research had come out of our own discipline and I didn’t know about it. I think that was the thing that really took me back and caught my attention. And so, I set out in my own institution to work with folks to try to understand if this really was something that was real and we re-demonstrated it as well in our own data. I reached out to Jeff Feldman at the APSF because of his history of working with technology and his deep, deep understanding of the tech that we use and I said Jeff, we need to figure this out. And he let me know that he had been in communication with manufacturers and with contacts at the FDA and people were working on trying to understand more about the measurement bias, but it was clear that there was an opportunity for us to educate the anesthesia community and the perioperative community about this potential measurement bias and really get people thinking about the accuracy of this instrument that we use every day, multiple times per day without really thinking critically about it.

[Bechtel] Thank you so much to Lane-Fall for her insights and helping to kick off the show. And now, let’s get into the APSF article. The New England Journal article by Sjoding and colleagues revealed the results from a retrospective analysis of pulse oximetry data from two groups of patients that looked at differences between the pulse oximeter readings compared to oxyhemoglobin saturation that was measured by laboratory co-oximetry. So, a comparison between the SpO2 and the oxygen saturation in the blood. The authors defined occult hypoxemia as an oxygen saturation of less than 90% when the paired SpO2 measurements were 92% or greater. The authors examined sub-groups from the patient cohorts that self-identified as Black and White. Now, for the results. The incidence of occult hypoxemia was three times greater in Black patients at 11.7% compared to White patients at 3.6%. The authors point out this is concerning for patient safety if clinical decisions regarding patient care including escalation of care or discharge from the emergency department are made without awareness of this finding. It is also important to note that this is a retrospective, uncontrolled study that does not include objective measurements of skin tone so there are limitations, but this just leaves us with a call to action to determine if there is the potential for pulse oximeter measurements to provide misleading information about oxygen saturation especially for patients with dark skin tones.

As Lane-Fall talked about earlier, this is not the first time that we are hearing about the relationship between skin tone and pulse oximeter measurements. In fact, the literature goes back over 15 years. A closer look at how the pulse oximeter works reveals that the biased measurement occurs due to overlapping absorption of light in the red region, remember that’s 660nm, for both oxyhemoglobin and the skin pigment, melanin. Bickler and colleagues published an article in Anesthesiology in 2005 that included laboratory evidence for this bias. In this study, the authors reported that the pulse oximeter overestimated the oxygen saturation measurements more for patients with dark skin tones and the bias increased as the oxygen saturation decreased. There were additional variations depending on the type of pulse oximeter used. The largest bias of 3.56 +/- 2.45% occurred in test subjects with dark skin with saturations in the 60-70% range, but the bias was lower, 0.93 +/- 1.64% when the saturation was increased to the over 80% range.

Bickler and colleagues completed another study on pulse oximeters in test subjects with dark and light skin tones and found that errors in pulse oximeter measurements may occur due to skin tone, probe type, saturation level, and gender. Their findings suggested that the biased measurements were important clinically when the saturation was less than 80%.

If we look even further in the past, Jubran and Tobin completed a prospective study of ICU patients and evaluated using pulse ox measurements to titrate oxygen concentration in order to maintain the partial pressure of oxygen above 60mmHg. The findings revealed that there was a greater bias in SpO2 measurements for patients with dark skin tones. The authors’ conclusion was that a higher pulse ox measurement of 95% be used for patients with dark skin tones to safely titrate oxygen concentration.  These studies revealed an important measurement bias, but this knowledge failed to make it into major textbooks in the areas of medicine, surgery, and emergency medicine. It is described in anesthesiology textbooks, but how this knowledge has been understood, used, and impacted clinical care is unknown. The newest study buy Sjoding and colleagues has additional importance and concern since the measurement bias was found to occur at higher levels of greater than 90% and not just at levels of less than 80%. Over the past few months, there has been a lot of work done by pulse oximeter manufacturers, the FDA and independent laboratories to evaluate for a measurement bias due to skin tone. So, this is something that we will be on the look-out for in the future.

I mentioned that the FDA is involved in recent evaluations. This is a very important step that occurred after the December 2020 publication. In fact, on January 25, 2021, Senators Warren, Wyden, and Booker requested a review by the FDA to look at the accuracy of pulse oximeter measurements in racially diverse patients and consumers. The next date that is important is February 19, 2021 which is when the FDA published a safety communication response about the pulse oximeter accuracy and the known limitations for patients with dark skin tones. The response stated that “if an FDA-cleared pulse oximeter reads 90% then the true oxygen saturation in the blood is generally between 86 and 94%” and that there is a “need to further evaluate and understand the association between skin pigmentation and oximeter accuracy.” Another important point is that for FDA clearance for a pulse oximeter, there is a requirement that the participant pool includes 15% of patients with dark pigmentation.

With the evidence to date and the new call to action, where do we stand? The evidence reveals that pulse oximeter measurements may overestimate the actual oxyhemoglobin saturation for patients with dark skin tones. There is a difference in the laboratory data which found a significant bias only when the saturation was less than 80% while clinical findings revealed a difference at higher saturations levels in the 90% range. It is vital that healthcare professionals use more than one pulse oximeter measurement to make clinical decisions with confirmation of oxygen saturation with co-oximetry if needed.

The APSF statement also includes the acknowledgement that it is safer to continue to use pulse oximetry to estimate oxygen saturation when monitoring patients. There could be additional patient safety threats if this known measurement bias related to skin tone led to decreased use of the pulse oximeter as a monitor at all.

There is a call to action now for manufacturers, regulators, and clinicians to make sure that medical technology is developed and tested in demographically and clinically diverse populations. This appears to be an appropriate time for the FDA to reconsider their requirement for inclusion of patients with dark pigmentation for medical device development in order to ensure that new medical technology minimizes measurement bias in subjects with dark skin tones and there needs to be a requirement for specific objective measurement of skin tone.  Another consideration includes the clinical use of pulse oximeter measurements and education about measurement errors due to factors such as skin tone, perfusion, dyshemoglobinemias, anemia, pulse oximeter brand and patient movement. Remember, clinical decision making is multi-faceted and cannot depend on a single monitor and the result measurement. I will read the final statement by the APSF now:

“APSF supports the renewed attention to the accuracy of the pulse oximeter, which has rightly revolutionized medical care and augmented patient safety. We call on clinicians, manufacturers, and regulators to work together to ensure that this device offers equitable benefits to all the patients we serve.”

[Bechtel] Before we wrap up for today, we are going to hear from Meghan Lane-Fall again. I asked her what she hopes to see going forward. I’m going to play her inspiring response now.

[Lane-Fall] What I hope to see going forward is that we think about the story of the pulse oximeter as a cautionary tale. I think that what it has to teach us is that our technology has real limitations and that it behooves us as anesthesia care professionals to deeply understand the tech that we use and to not assume that it works without interrogating it. Any device that takes a complex physiologic phenomenon and gives you a number whether that’s a pulse oximetry or a  cvp or bis monitor, you know, we really need to know how that works. We need to know under what conditions it works, under what conditions it doesn’t work and we need to understand how those devices are approved. I think the take away for me here is to understand and to teach residents and other trainees about the pulse oximeter and other similar pieces of technology to understand that it really is a proxy for something else that is happening, mainly the saturation of hemoglobin in the blood, that there are gold standards that we need to know about and understand. So, in the clinical setting I think I am going to have a more nuanced understanding of treatment of oxygen saturation. I think from the perspective of groups of the groups like the APSF zooming out and taking a bigger picture of the view of this, I think we have to be really critical of how our devices are approved, how they are tested and approved. You know probably this extends beyond devices to clinical algorithms to machine learning to understand what’s under the hood, how they work so that we can make informed decision about whether to use the tech, so that we can have a conversation with device manufacturers and other groups to make sure that for instance diverse populations are included in the testing populations so that we don’t have surprises like this when the tech gets spread to a larger population. So, I do think that there are a few lessons to take away and my hope is that we’ll be able to learn, not just about a pulse oximeter, but about many of the other types of technology that we use in caring for patients every day.

[Bechtel] Thank you so much to Meghan Lane-Fall for her contributions to the show today and for helping to address this important anesthesia patient safety concern with technology with practical advice for going forward. It is so important to understand that our advanced medical technology has real limitations and we need to know how the technology works to help keep our patients safe.

If you have any questions or comments from today’s show, please email us at [email protected].

Visit APSF.org for detailed information and check out the show notes for links to all the topics we discussed today. Please keep in mind that the information in this show is provided for informational purposes only and does not constitute medical or legal advice. If you have not done so already, we hope that you will rate us and leave a review on iTunes or where ever you get your podcasts and feel free to share this podcast with your friends and colleagues and anyone that you know who is interested in anesthesia patient safety.

Until next time, stay vigilant so that no one shall be harmed by anesthesia care.

© 2021, The Anesthesia Patient Safety Foundation