Pulse Oximeter Bias: 2026 Patient Risks

It’s pretty jarring to learn that 88% of pulse oximeter readings can overestimate oxygen saturation by up to 6% on people with darker skin, a massive discrepancy that has real clinical consequences and shows just how limited our tech is. We simply have to rethink how we interpret these numbers, especially when you start throwing other things into the mix, like numbing creams.

The 88% Discrepancy: Racial Bias in Medical Devices

That 2020 New England Journal of Medicine study reporting that 88% of pulse oximeter readings are inflated by up to 6% in darker-skinned patients wasn’t just some statistic. It’s a systemic flaw in a device we use every single day, and it can cause us to delay or completely miss the need for treatment. This is not a small error. A 6% difference can make a patient with dangerously low oxygen look stable on the monitor, so we fail to intervene. I’ve seen it firsthand. You have a patient who looks distressed, who is clearly working to breathe, but the monitor is showing a perfectly fine number. When the machine and the patient don’t match, you have to be suspicious, especially when working with a diverse group of patients. The problem comes from how these things work. A pulse ox shoots two kinds of light, red and infrared, through your finger and measures how much is absorbed by arterial blood. Oxygenated and deoxygenated blood absorb light differently. The issue is that melanin, the pigment in skin, also absorbs that light. So in people with more melanin, the oximeter’s little brain gets confused and thinks there’s more oxygenated blood than there really is, spitting out a falsely high number. We have to know the tools we’re using have these built-in flaws and adjust how we practice.

2021 FDA Guidance: Acknowledging the Problem

In 2021, the FDA finally issued a safety communication admitting that pulse oximeters have limitations and may be less accurate in individuals with darker skin pigmentation. For many of us, this was a long-overdue step that at least put the problem on the books at a regulatory level. The FDA statement said they’ve been working with manufacturers on the issue, but this was the first time they officially told clinicians to consider a patient’s skin tone when looking at the monitor. The warning also mentioned other things that can throw off a reading, like bad circulation, skin thickness, or even nail polish. So in the real world, just blindly trusting a pulse oximeter number when the patient in front of you looks terrible is just bad medicine. You have to use that number as just one piece of the puzzle, not the whole story. The FDA’s official memo backs up what good clinicians already do: perform a well-rounded assessment instead of just treating the monitor.

Numbing Creams: An Unforeseen Variable

Everyone’s starting to talk about the skin tone bias, but there’s another variable that almost no one thinks about that can wreck your reading: topical numbing creams. So many of these creams with lidocaine or prilocaine are vasoconstrictors, meaning they shrink the local blood vessels to reduce blood flow. If you slap a pulse oximeter on a finger that was just slathered in one of these creams, the device can’t find a good pulse, leading to a crazy low reading or no reading at all. I’ve walked into this exact situation. Picture this: a patient gets a numbing cream on their hand for an IV start, and then someone puts the pulse ox on that same finger. The monitor suddenly screams “hypoxia,” everyone starts scrambling for an oxygen mask, but the real problem is just localized vasoconstriction from the cream. I’ve had a patient’s saturation “inexplicably” plummet, only to find out a topical had been put right where the sensor was. Move the sensor to another finger, and boom, the “hypoxia” is magically cured. It’s a critical thing to ask about when a reading looks off.

The “Healthy” Patient with Low O2: A Case Study

One of the most concerning things you see in practice is an otherwise healthy person with a darker skin tone who has a pulse ox reading of 90-92%. They look fine, they’re talking, no signs of distress. But that number is traditionally where we start to worry. The real danger here was laid bare in a 2021 *Anesthesiology* study that found occult hypoxemia (low oxygen not detected by pulse oximetry) was three times more common in Black patients than in White patients. The problem isn’t that the patient is fine with a low O2. The problem is that the oximeter is failing to show you how low they *truly* are, giving you a false sense of security. I remember a young man, no respiratory issues at all, who was consistently reading in the low 90s on the pulse ox. He was alert, talking, breathing easy. But something felt off, so we drew an arterial blood gas, the gold standard. His true saturation was 88%. The pulse oximeter missed a clinically significant drop. This is a real threat to patient safety. If you don’t mentally adjust your interpretation for skin tone, you’re going to miss true hypoxemia in a lot of your patients.

Beyond the Conventional Wisdom: Rethinking “Normal”

We’re all taught that a pulse ox reading of 95% or higher is normal. With all this data on skin tone bias, that one-size-fits-all number is obsolete. We have to stop using a universal threshold for “normal” O2 saturation when we’re relying on pulse oximetry, especially when assessing patients with darker skin. A “normal” 95% on the monitor for a Black patient might be hiding a true arterial oxygen saturation that would have set off alarm bells in a white patient. Our diagnostic vigilance has to get sharper. Instead of just glancing at the number and moving on, you have to ask yourself: “Given this patient’s skin, and how they look, could this number be lying to me?” It’s a mindset shift. You have to trust your own clinical judgment and the patient’s symptoms more than you trust a potentially flawed number on a screen. The “gold standard” for non-invasive monitoring has a huge blind spot. So, if a patient with darker skin is showing borderline readings and their clinical picture is even slightly concerning, it’s probably time to get an arterial blood gas much sooner than you would for someone else. That’s how you can prevent bad outcomes by catching true hypoxemia early. You’re treating the patient, not the number. And this goes for the numbing creams, too. A lot of clinicians just don’t think to connect a topical anesthetic with a bad pulse ox reading. But the vasoconstriction from lidocaine is a real and dramatic effect. Our mental checklist of “is the patient cold? do they have bad perfusion?” needs a new item: “did someone put a cream on this finger?” It’s a simple, overlooked detail that can cause a lot of chaos. These limitations mean we have to be smarter than the machine. The known issues with pulse oximeters, from skin tone bias to interference from numbing agents, mean we can’t be lazy. A “normal” reading on the screen might not be normal at all, and it’s our job to figure that out by looking at the whole patient.