SpO2 is the most underrated vital
June 2025 , Sam Moreland
In my career I’ve looked at thousands of hours of vital signs. Not just the numeric metric that you get out, such as heart rate or BP, but the raw underlying waveforms from which we get these measures. I spent a lot of time at Current Health manually reviewing and researching anomalies and un-expected vital signs. What has always amazed me is the underutilisation and misunderstandings around SpO2.
While working at Current Health we monitored very sick people in and out of hospital for long periods of time, during which we unfortunately saw a lot of deteriorations and deaths. Those sick patients gave us insights into what vitals look like on the very sick patients, where it matters to understand these things deeply. I would often find that SpO2 would be the first vital or only vital that would be anomalous in these patients.
It also became obvious that the way that traditional SpO2 works and is being utilised hid extremely important information. This is a problem both with how the SpO2 devices work and how spot check measurements are utilised with patients. This mismatch has allowed potentially significant early warning signs for deterioration to get missed.
What is the point of a vital sign?
Why do we have vital signs and why do we use the ones we have. In short, vital signs are indicators of performance of different aspects of the body.
Blood pressure - Indicates arterial stiffness and the ability for blood (oxygen) to be able to travel to all parts of the body.
Heart rate - Indicates that the “power house” of the body is working in an expected way.
Respiratory rate - Indicates how effectively oxygen is getting into the body (and CO2 out).
SpO2 - Indicates how effectively oxygen is getting transported around the body.
Each of these vitals gives a different view of how well the body is working. A lot of the times these vitals are correlated, for example when you stand up your blood pressure, heart rate and respiratory rate will increase to make sure oxygen being supplied to the brain does not decrease. When the body doesn’t do this, people can faint which is known as postural (orthostatic) hypotension.
But sometimes these vitals can change in an uncorrelated way, and that's a very important signal to know. The mechanism that controls this is known as the baroreflex system, whose whole purpose is to make sure oxygen (and other things) are getting to where they need to get to. If a vital sign is not varying “correctly” this can be a serious sign of deeper issues with the patients autonomous nervous system.
What people don’t know about SpO2
All of this commentary is based on what I have seen, these observations have not come from randomised control trials. These findings have been taken from doing retrospectives on extremely ill patients, where vitals have been abnormal but correct i.e. not caused by motion or faulty hardware.
Fast Drops
Your SpO2 can vary rapidly. I have seen SpO2 levels go from 100 % to below 70 % in around 10 s. This is real and happens a lot. Often this would be followed by a slow rise in SpO2 over the course of several minutes, but I have seen it take 15 mins to hours to stabilize.
The prime cause of the drops were postural changes in patients, usually overnight. I assume this is fluid in the lung being moved and covering the open alveoli, and then the other alveoli opening to compensate. But I have seen this happen independently of posture.
These drops usually happened at night with some patients having multiple drops.
Inter beat differences
Your SpO2 can vary between blood pulses. On the sickest patients we monitored I saw this on about 4-5 patients. SpO2 varied between 100 % and 80 %, and this would alternate between beats. We did a lot of work analysing this, and it was not a hardware error or an error to do with motion noise.
The problem with SpO2 monitors
Long term averaging
All medical devices contain errors. This is because of the environments that they are used in, such as high motion, which can cause errors in the system which produce incorrect measurements. These measurement errors are almost always at levels which indicate there is something wrong with the patient. Heart rates of 250 bpm, respiratory rate of 60 brpm etc. For SpO2 it is usually 82-84 % because of how it's calculated. These errors are usually short in duration, but can be prolonged say if someone is walking for a while.
Because of these errors, devices average data over different time scales to try and “smooth” or remove erroneous readings. Most monitors will average over 1-2 minutes or more. This has an impact on short time scale SpO2 drops. SpO2 drops to 70 % or 80 % would be smoothed over making them seem smaller i.e. a drop to 80% could appear like a drop to 92 %.
Inter-beat differences are completely smoothed out and there is no ability to to see this phenomena.
Validation
Validation for SpO2 happens in a very controlled setting. Around 12 people of mixed genders and skin colours will sit as still as possible and be de-saturated from 100% to 70 - 60 % SpO2. This is a very difficult procedure and as such mainly fit and younger people are used.
Validation by this method gives you absolute best case performance. It would be impossible to do while the patient is moving, as blood draws are taken to calculate the true blood oxygen level. In the real world patients move, devices are not applied correctly, and this means practically that these monitors can be unreliable in moderate motion environments.
We also found at Current Health there is a huge difference in different skin colours. The validation protocol rightly enforces the need for different skin tones including dark skin tones. But we found that within darker skinned individuals there are huge differences in absorption criteria. We found our cohorts from Uganda absorbed 10-100 times more light than African Americans, when their skin tones appear to have less variance than between white and dark skins. Because SpO2 is measured using light, it means that if you design your device purely around American cohorts, your device's sensors may not be powerful enough to penetrate the skin of other populations. I believe this may be the cause of the inaccuracies of SpO2 measurements in the general population.
Make sure your devices are calibrated!
A study evaluated 847 pulse oximeters in use across 29 NHS hospitals, using a portable spectrometer to assess the emission spectra of their sensors. The findings revealed that 10.5% of sensors had functional electrical errors, potentially causing inaccurate readings. Additionally, 22.3% of the remaining sensors had emission spectra deviations from manufacturer specifications, which could result in oxygen saturation inaccuracies greater than 4% within the clinically relevant range (70–100% SpO₂). This highlights that a significant proportion of pulse oximeters in the NHS may not provide reliable readings.
The problems with how SpO2 is measured
Not all SpO2 monitors are the same.
Unless it has been FDA cleared, do not trust the monitor. You can go on amazon and buy many SpO2 monitors which have not been validated or validated poorly. A study did just this and found huge variances in accuracy between different low cost widely available SpO2 monitors. I would love to see a larger scale study on wrist mounted SpO2s.
It is crucial that you do not go with cost, and fully evaluate the accuracies of the devices before selecting them for clinical use.
The act of measuring will change the SpO2 value
At Current Health we would see desaturating patients quite often. During development work and in early trials, we would have spot check measurements taken by nurses to validate our measurements. We found that during the act of reading, the patients with low SpO2 readings would increase to normal. This is because the nurse or assistant would often wake or arouse the patient either before measuring or during the course of measurement. Once the patient becomes awake, their SpO2 comes back to normal. This means that normal spot check measurements can give misleading readings.
As such an SpO2 readings that are taken by patients in the home setting will give a false idea about how sick they are. That's because they will be in an aroused state when doing so.
Its on the finger
If there is one worse place than the wrist to place a device, it's the finger. Easy to access but terrible for motion noise and light isolation. One of the bigger problems is due to centralization of circulation. This is when the body stops blood flow from the periphery (fingers, toes) to focus the blood on the vital organs, usually when patients are critically ill or cold. Because there is no blood flow in the finger, finger pulse oximeters cannot then detect blood oxygen levels.
There were a few times where I saw this in hospitals. Nurses would try to perfuse the patients hands by warming them up and they were unable to get a reading. But because the Current Health device measured SpO2 on the upper arm it correctly identified deteriorated patients.
Takeways
There are a few serious issues that combined lead to incorrect SpO2 measurements and the underutilisation of it as a vital to detect serious patient issues. Here are a few suggestions to improve patients' lives.
Device manufacturers should use smaller averaging windows or offer different averaging options. They should also make averaging very clear to clinicians when being trained with them (I don’t know how much this is done).
Device manufacturers should include dark skin African people in their validation cohorts.
If possible, when taking SpO2 spot measurements, try not to change the patient's state i.e. don't wake them up.
On sick patients, use long term high sample rate SpO2 devices. There are unfortunately not many of these.
On critically ill patients use reflectance sensors on centrally located sites.
If you're a medical company wanting to improve your monitoring device, or a medical facility needing advice on good device selection, please feel free to reach out.
