Blog
A number on a screen isn’t a measurement until the probe is in the right place

A pulse oximeter will give you a number from almost anywhere you clip it. That is exactly the problem. The screen does not tell you whether the sensor is on a site it was designed for, whether the tissue underneath is well perfused, or whether the number is 3% or 8% out. This article looks at what published studies show happens when probes are misplaced or used on poorly perfused sites, why it matters for oxygen decisions, and what the evidence cannot yet tell us.
What a pulse oximeter actually measures
A pulse oximeter estimates the oxygen saturation of arterial blood (SpO2) by passing red light (around 660 nm) and infrared light (around 940 nm) through tissue and measuring how much is absorbed. It relies on a pulsing arterial signal. Those absorption patterns are different when the light travels through a fingertip, an earlobe or a toe, and a sensor built for one site is calibrated and shaped for that site.
A reading also says nothing about some things clinicians sometimes assume it covers. It gives no information on acid-base balance or ventilation, so a normal SpO2 can sometimes give a falsely optimistic picture of oxygen delivery to the tissues.
Using a sensor on the wrong site
A finger clip on the ear
A 2007 study in Respiratory Care tested something clinicians do in real life: when no reading can be obtained from a finger, some move the finger-clip sensor to the ear and trust the result if the pulse rate matches. The study put a finger-clip sensor on both the finger and the ear in 30 adult patients undergoing pulmonary function testing, and compared both with arterial blood gas results taken at the same visit.
- Only 1 of 30 finger readings (3.3%) differed from the true saturation by more than 3%.
- 24 of 30 ear readings (80%) did, and every one overestimated the true value. In one patient the ear reading was 8% too high.
- Agreement with the blood gas value was within about plus or minus 2.35% for the finger, but ranged from about 7.2% too high to nearly zero for the ear.
The key point is that these ear readings passed the usual sense checks. Pulse rate matched the finger and the waveform was stable and acceptable. The author attributes the error to the different way light is absorbed by an earlobe, and possibly to light bypassing the tissue because the clip fits an ear poorly. The study has limits: one device model was used, and all patients had saturations of 90% or above, so it says little about patients who are genuinely hypoxic. The author notes that overestimating saturation in hypoxic patients carries risk from non-treatment, and cites the American Association for Respiratory Care guideline that a probe should be attached only to its intended site.
Malpositioned sensors during real hypoxia
A 1993 study in Anesthesiology examined what happens when sensors are misplaced while oxygen levels are actually falling. Twelve healthy volunteers had an arterial line placed and wore eight oximeters, five of them with deliberately misplaced sensors, while breathing gas mixtures that lowered saturation gradually from 100% to 70%. Results varied by model: one oximeter underestimated at every level, while three others underestimated at high saturations and overestimated at low ones. The authors concluded that, at low saturations, misplacement could make a profoundly hypoxic patient appear only mildly hypoxic, and recommended checking sensor position frequently.
Peripheral sites are slower to show a problem
Even when the probe is on a recognised site, the further it sits from the core, the longer it can take to see a fall in oxygen. Several studies measured this directly:
- Toe versus ear and hand. In healthy volunteers exposed to rapid-onset hypoxaemia, the toe detected a fall below 90% on average 57 seconds after the hand and 63 seconds after the ear. The ear and hand differed by only 6 seconds.
- Finger versus earlobe. In 104 patients undergoing bronchoscopy, a finger probe flagged a fall to 90% on average 36 seconds after an earlobe reflectance probe. Compared with arterial blood gases, the bias was 0.24 for the earlobe and 2.31 for the finger.
- During breath-holding. In seven men holding their breath during light exercise, ear probes fell to a mean of 78% while finger probes still read 94.6%. With one of the two oximeters tested, finger readings were on average 6.0% higher than ear readings at their lowest points, and the finger low point came about 15 seconds later. The authors link this to blood vessels in the hands and feet narrowing when oxygen falls.
These were small studies, several in healthy volunteers, but they point the same way: a peripheral reading can look reassuring while the picture is worsening.
Poor perfusion: when the finger itself is the wrong place
In shock, cold or low-flow states, even the standard finger site may struggle. In a study of 32 intensive care patients receiving high-dose vasopressors, forehead reflectance sensors were compared with finger sensors against arterial saturation measured by co-oximetry. The finger sensor failed to give a value in four cases and the forehead sensor never failed. Outlying readings occurred in 15% of forehead measurements and 32% of finger measurements.
The evidence is not one-sided, though. A review of probe sites under poor perfusion found that bias ranged from 0.2 to 1.7 for finger probes and from 0.1 to 8.1 for ear, nose and forehead probes. In patients undergoing cardiac surgery with mildly hypothermic bypass, the forehead sensor tested showed no clear advantage over the best finger combination. And in 18 anaesthetised patients, forehead and finger probes performed similarly under normal perfusion, but when blood flow to the arm was restricted, the forehead reading fell sooner during hypoxia. The sensible reading of this body of work is that site, sensor design and the patient’s circulation all interact. A sensor should be used where it is designed to be used, and a reading from a poorly perfused site deserves less trust.
The blood pressure cuff on the same arm
An inflating cuff can switch the oximeter off. In volunteers wearing a finger sensor with a blood pressure cuff on the same arm, all four oximeters tested lost their signal during cuff inflation, though the time lost varied between models. The authors point out that important information may be missed if the signal drops out at the moment a haemodynamically unstable patient is being assessed. Other published work states plainly that routine placement of the oximeter and cuff on the same arm is not recommended.
Other things that make a number unreliable
- Carbon monoxide. Standard pulse oximeters cannot tell oxygen-carrying haemoglobin from haemoglobin carrying carbon monoxide. A meta-analysis found pulse oximetry had 77% sensitivity and 83% specificity for identifying carbon monoxide poisoning, with a false negative rate of 23%.
- Skin tone. The Independent Review of Equity in Medical Devices, published in March 2024, found extensive evidence that pulse oximeters overestimate true oxygen levels in people with darker skin tones, which could delay treatment if dangerously low oxygen is missed. Its recommendations included updated instructions for use that state the device’s limitations and differences in accuracy by skin pigmentation, stronger approval standards, and exploring multi-wavelength oximeters. The review stressed that it was not advising curtailment of oximeter use.
- Nail polish. The evidence is mixed. A 2023 systematic review of 21 studies did not recommend routine removal of nail polish before a reading, although some EMS guidance advises removing dark colours or changing probe site if in doubt.
- Cold extremities and movement. Both are recognised causes of unreliable finger readings.
Why a wrong number matters
Oxygen is given to a target, so a wrong reading can push treatment in either direction. A falsely high number can hide hypoxaemia, as the ear and malpositioning studies suggest. A falsely low one can lead to oxygen being given unnecessarily. The evidence on targets matters here: a meta-analysis found that liberal oxygen therapy (target SpO2 above 96%) in acutely ill adults was associated with higher mortality than a more conservative approach, although the British Thoracic Society committee felt the review did not settle the optimal range. For patients at risk of hypercapnic respiratory failure, a lower target of 88 to 92% is appropriate. Neither strategy works if the number guiding it is not trustworthy.
What the evidence suggests in practice
- Use the sensor on the site it was designed for, and follow the manufacturer’s instructions. Moving a clip to another site because the finger gives no reading can produce a convincing but wrong number.
- Do not treat a matching pulse rate or a tidy waveform as proof that the saturation is accurate.
- Check the sensor position regularly, and avoid sites you cannot see or reach.
- Avoid placing the sensor on the arm with the blood pressure cuff.
- When the patient is cold, shocked or poorly perfused, or carbon monoxide exposure is possible, treat the reading as less reliable and weigh it against the clinical picture.
- Treat the oximeter as one input. UK practice places SpO2 within a full ABCDE assessment and early warning scores, not in place of them.
Echelon’s training is built around UK pre-hospital frameworks, where observations like SpO2 are interpreted alongside a full clinical assessment.
What these studies cannot tell us
- Several studies are older and tested specific device models, so modern oximeters may perform differently.
- Many were in healthy volunteers or in hospital patients, not in field conditions with cold hands, gloves, vehicles and movement.
- The ear study included only patients with saturations of 90% or above, and the breath-hold study involved seven men.
- Findings on forehead sensors depend on the sensor design and the placement used.
Key takeaways
- A finger-clip sensor placed on the ear overestimated saturation by more than 3% in 80% of patients in one study, even when pulse rate and waveform looked fine.
- The further the probe is from the core, the longer it can take to show a fall in oxygen. Toe probes lagged ear and hand probes by about a minute in healthy volunteers.
- In shock and low perfusion, finger readings were more often outliers than forehead readings in one ICU study, but results across studies are not uniform.
- Carbon monoxide, skin tone, cold, movement and a nearby blood pressure cuff can all make a reading unreliable.
Over to you: Cold hands, gloves, shock. When the number and the patient disagree, which do you trust?
Sources
- Haynes JM. The Ear as an Alternative Site for a Pulse Oximeter Finger Clip Sensor. Respiratory Care 2007;52(6):727–729
- Barker SJ et al. The effect of sensor malpositioning on pulse oximeter accuracy during hypoxemia. Anesthesiology 1993;79(2):248–254 (abstract)
- Delays in the detection of hypoxemia due to site of pulse oximetry probe placement (abstract)
- Significant delay in the detection of desaturation between finger transmittance and earlobe reflectance oximetry probes during fiberoptic bronchoscopy: analysis of 104 cases
- Pulse oximetry to detect hypoxemia during apnea: comparison of finger and ear probes
- Pulse oximetry and high-dose vasopressors: a comparison between forehead reflectance and finger transmission sensors. Intensive Care Medicine, 2012
- Pulse oximetry review, Critical Care 1999 (probe bias under poor perfusion)
- Yamaura K et al. Evaluation of finger and forehead pulse oximeters during mild hypothermic cardiopulmonary bypass. J Clin Monit Comput 2007 (abstract hosted by a manufacturer)
- Forehead versus finger pulse oximetry in anaesthetised patients (abstract)
- Kawagishi T et al. Failure times of pulse oximeters during blood pressure cuff-induced hypoperfusion in volunteers. Anesth Analg 2004 (abstract hosted by a manufacturer)
- Balaguer J et al. Effect of intermittent non-invasive blood pressure monitoring on pulse oximetry and SpHb monitoring (conference abstract)
- Equity in medical devices: independent review (DHSC, March 2024)
- Independent review recommendations summary, Patient Safety Learning
- Pulse oximetry and carbon monoxide poisoning: meta-analysis coverage (European Emergency Medicine Congress)
- Reliability of pulse oximetry and the factors affecting it, Paramedic Practice
- The value and limitations of pulse oximetry, Acute Care Testing
- Hatchett R. Pulse oximetry: exploring its role, limitations and challenges in clinical practice. Nursing Standard, 2026
- Target oxygen saturation in acutely ill adults, BMJ Best Practice
Interested in pre-hospital medical training for your team? Contact operations@echelonmedicalsolutions.com.
