Pulse oximeters and the colour of your skin

For several years now, you have been able to go to your local drugstore and buy a device called a pulse oximeter for less than $50. It's a little thing you clip on your finger, and in a few seconds it displays two numbers. One is your pulse rate, and the other is supposed to be the percent of maximum capacity of oxygen that your blood is carrying. Most healthy people show a blood-oxygen percentage of around 98%, but anything considerably less than that means you're not getting enough oxygen to your tissues.

Hospitals and doctors have more sophisticated versions of these devices, but apparently they all share the same flaw these days:  they can give falsely reassuring readings on people whose skin has significant melanin content. Black people, in other words. So for decades, anyone in that category whose blood oxygen has been monitored with a pulse oximeter has been in danger of going untreated for low blood oxygen, compared to a person whose skin was lighter. 

This is not news. The problem has been known for decades, but received added publicity during the COVID-19 pandemic. Studies have shown that people of colour receive less supplemental oxygen than average during medical treatment, and bad pulse-oximeter readings only exacerbate this problem. 

Innovation for inclusion

Fortunately, some engineers at Brown University are trying to address the problem. In a report carried by the health-information site Statnews, Kimani Toussaint, a Black professor of engineering, is reported to be working with students on a patentable idea that will lead to pulse oximeters that give the correct reading no matter who is being tested, and what colour their skin is. 

I wish them well, and hope that they can make a significant difference in what has to be one of the most embarrassing deficiencies in healthcare technology to come to light in years. 

It didn't have to turn out this way.  The same article cites a report in Wired on one of the first oximeters to hit the market way back in the 1970s, developed by what was then a medical branch of the instrumentation company Hewlett-Packard.  In their typically thorough way, H-P included 248 people of colour in their volunteer pool of testing subjects, and made sure the readings were as good for them as for the other volunteers. 

Of course, the H-P device was a little fancier than the ones you get at Walgreen's.  It examined eight wavelengths of light, not one or two like the current ones do, and was about the size of a small beer cooler. I'm sure it sold for more than fifty bucks, too. But it got the oximetry ball rolling, and from that point on it was a question of how cheaply the device could be made, and whether inaccurate readings on a minority of the FDA-required sample population could be disregarded in the approval process, which they apparently were. 

I don't think anybody in the healthcare industry deliberately intended to make devices that discriminated in a purely technological way against people of colour.  But beyond a certain point, ignorance was no longer an excuse, as studies were published describing the problem and cautioning clinicians not to trust readings of pulse oximeters with darker-skinned patients. 

But this is not a good solution. The right fix, as Toussaint and his colleagues recognise, is to make pulse oximeters that work right for everybody, not just for white folks. Supposing the Brown academics succeed (which seems pretty well guaranteed at some level, as H-P got it right in the 1970s with vastly inferior technology), what happens then? 

Counting the cost

Like any industry, the healthcare-technology industry wants to make money and serve its customers as well as it can. Compared to consumer products, devices sold for healthcare purposes are highly regulated and licensed, and jumping through the regulatory hoops is a cost that makes up a significant fraction of the price.

Unless the FDA insists on changing its rules so that pulse oximeters have to read equally accurately for all colours of patients, the industry doesn't have much of an incentive to adopt a newer technology that does that, whether it's patented by Brown or developed on their own.

For one thing, it means a whole new round of proof-testing and regulatory approval.  And for another thing, the market for pulse oximeters is probably not that big, and making a substantial investment in it for a benefit that will show up in only a minority of patients is a hard marketing sell.

I'm reluctant to use the phrase "systemic racism," but it might well apply in this case. As I said, I don't think any individual manager or pulse-oximeter company set out to discriminate against people of colour in developing devices that don't work quite as well for that group. But somewhere along the long road of development between H-P's giant 1970s device and the $50 versions of today, somebody compromised some things and created the problem. It would require a huge effort of investigative journalism and probably subpoenas to find out exactly how it happened, but the outcome is clear.

Sometimes, adverse publicity by itself will make an industry clean up its act. Maybe if enough people of colour ask questions of their clinicians about pulse oximeters, it will have an effect back up the supply chain and the companies will go to the trouble and expense of dealing with the issue. But it's not going to happen automatically.

In the old days, a letter-writing campaign might have had some effect. These days, social media is the obvious channel to use in letting people know there's a problem. It's a pretty blunt instrument, though, a little like putting out a cigarette with a fire hose, and it can backfire on the user as well.

But as the Statnews article quoted Toussaint as saying, this problem is a poster child for increasing diversity in science. If it's not a problem to you or people you know, you simply tend to ignore it. Now that we know it's a problem -- all of us engineers -- I think it's time somebody should do something about it. 

This article has been republished from the author’s blog, Engineering Ethics, with permission.


Join Mercator today for free and get our latest news and analysis

Buck internet censorship and get the news you may not get anywhere else, delivered right to your inbox. It's free and your info is safe with us, we will never share or sell your personal data.

Be the first to comment

Please check your e-mail for a link to activate your account.