Educational aims
The aims of this article are to:
- •
Recognise the role of accurate oxygen saturation readings in guiding oxygen therapy.
- •
Identify the limitations of the use of pulse oximetry in measuring oxygen saturation.
- •
Understand the basic principles of the mechanism by which pulse oximeters obtain saturation readings.
- •
Have an understanding of the studies investigating the degree of bias of skin colour on pulse oximetry readings.
- •
Interpret pulse oximetry readings in the clinical setting with the knowledge that readings may be less reliable in patients with pigmented skin.
Abstract
Pulse oximetry is widely used to non-invasively estimate the oxygen saturation of haemoglobin in arterial blood (SpO 2 ). It is used widely throughout healthcare and was used extensively during the Covid-19 pandemic to detect and treat hypoxic patients. Research has suggested that pulse oximetry is less accurate in patients with darker skin. This led the US Food and Drug Administration agency (FDA) to issue a safety statement warning that pulse oximeters may be inaccurate when patients have pigmented skin.
Evidence suggests that the oxygen saturation of arterial blood (SaO 2 ) may be being overestimated by measuring SpO 2 in those with pigmented skin. The degree of overestimation increases as SaO 2 decreases especially when SpO 2 reads below 80%. We review how pulse oximetry works and consider the implications for a patient’s health when interpreting SpO 2 in individuals with pigmented skin.
Introduction
The ability of pulse oximetry to continuously and non-invasively monitor the oxygen saturation of haemoglobin in arterial blood (SpO 2 ) has revolutionised modern medical practice . It is a vital tool used within inpatient, ambulatory and outpatient settings and contributes to the assessment of a patient.
During the COVID-19 pandemic, the need to monitor SpO 2 was never greater due to the SARS-CoV-2 virus reducing oxygen-carrying capacity because of inflammation and subsequent ventilation-perfusion mismatch, leading to hypoxaemia . Low arterial oxygen levels are a risk factor for poor outcome and ‘silent hypoxia’ can be easily missed without measuring SpO 2 . NHS England published guidance for the triage of patients in primary care or the community with suspected COVID-19 which along with other observations used pulse oximetry to determine whether a patient required hospital assessment . Community use of small device pulse oximeters and the creation of ‘virtual wards’ to monitor patients remotely helped reduce the burden on hospitals .
In December 2020 Sjoding et al. highlighted the inaccuracy of pulse oximetry in patients with dark skin; demonstrating a three-time higher level of occult hypoxaemia . This led to the US Food and Drug Administration agency (FDA) issuing a safety statement in February 2021 warning that pulse oximeter accuracy may be affected by the following situations: poor circulation, skin pigmentation, skin thickness, skin temperature, tobacco use and fingernail polish . In November 2022 the FDA stated that they are reviewing recommendations due to ongoing concerns that pulse oximeters may be less accurate in individuals with darker skin pigmentations . If pulse oximetry readings can differ between those of different skin colours, SpO 2 cut-offs used to stratify patients have clear safety concerns. We therefore set out to explore the importance of pulse oximetry, what is already known about racial bias in pulse oximetry, why this happens and what we can do about it.
Importance of pulse oximetry
Pulse oximetry has been widely accepted in clinical medicine since the 1980s; utilising a small portable device which can monitor SpO 2 virtually anywhere. Traditional signs of hypoxaemia such as circumoral cyanosis are subjective and typically late indicators of reduced oxygen levels. In contrast, pulse oximetry provides an objective measure of oxygenation and when used appropriately, offers a simple, accurate and reliable tool to assess the arterial oxygen status of patients . The accuracy and reliability of pulse oximetry has been further increased by developments in photoelectric sensors, light-emitting diodes, and microprocessor technology . Furthermore, the pulse oximeter is a non-invasive method of assessing oxygen levels compared to arterial blood gas, enabling easier and faster arterial oxygen monitoring. Although pulse oximetry should not be used as a replacement for arterial blood gas testing, it does serve as a screening tool when decreased oxygenation is clinically suspected, enabling faster recognition of critically unwell patients hence rapid management and stabilisation . In children, this is of particular importance to avoid painful arterial blood gases. However, the inherent limitations of non-invasive technology require a cautious interpretation of results alongside the clinical picture.
As well as early detection of clinical deterioration, pulse oximetry allows clinicians to easily monitor and titrate oxygen delivery. This can ensure that sufficient oxygen is delivered whilst avoiding hyperoxia. There are several clinical scenarios where over-oxygenation should be avoided. In adult patients with acute coronary syndrome (ACS) there is emerging evidence that excessive oxygen administration causing hyperoxia increases myocardial injury . Therefore, newer guidelines do not recommend oxygen therapy unless SpO 2 are less than 90 % . Similarly, in premature infants, hyperoxia is associated with retinopathy of prematurity (ROP) and infants require careful titration of oxygen to optimise growth and development whilst minimising complications from ROP . Children with neurological conditions such as brain injuries or brain malformations, neuromuscular disease, severe deformities of the spine or chest wall, or are on medications that may reduce the drive to breathe may hypoventilate causing retention of carbon dioxide . These children often require positive pressure ventilation and careful oxygen titration. Pulse oximetry is used to maintain tight SpO 2 control to prevent the suppression of the hypoxic drive. If pulse oximetry underestimates the oxygen saturation of arterial blood (SaO 2 ), this could prompt more aggressive oxygen administration potentially increasing risk of respiratory failure. Highlighting another need for accurately monitoring of pulse oximetry.
Principles of pulse oximetry
Whilst pulse oximetry offers clear advantages in clinical practice, they are only providing an estimate for SaO 2 . It is important to understand the clinimetric properties of the pulse oximeter being used, and the limitations associated with this measurement.
Precision
The oxygen-haemoglobin dissociation curve shows that there is a sharp drop off in the saturation of haemoglobin once the partial pressure of O 2 decreases below 93 %, highlighting the importance of sensitivity when collecting this data. As the curve shifts left- and right-wards in response to physiological pressures, accurate measurement of the SpO 2 will inform the decision to give oxygen and to what degree this needs to be titrated. A cut-off value taken where hypoxia is important (such as time below 93 %) may thus be a useful measure. A cut-off of 93 % is taken from the point on the oxygen dissociation curve at which small changes in pO 2 result in exponential decreases in % SaO 2 . Extrapolation from the following graph ( Fig. 1 ) suggests that a SaO 2 value of 93 % equates to a pO2 of around 70 mmHg (9 kPa). SpO 2 is optimally accurate when estimating SaO 2 over a range of 90–100 %, with intermediate accuracy at 80–90 % and wider limits of agreement in the range below 80 % .
Certified pulse oximeters used in health care settings have been tested for accuracy on healthy volunteers comparing blood gas results with SpO 2 . The FDA requires pulse oximeter measured SpO 2 to be accurate to within +/- 3 % of arterial blood gas SaO 2 values . Real world accuracy may be less than reported due to differences in patient population. The accuracy of over-the-counter pulse oximeters may be much less as have not had the same level of testing. Pulse oximeter accuracy is greater at higher oxygen saturations .
Many low-cost home pulse oximeters are available that have not undergone effective validation. The consequences of this include highly inaccurate readings as reported by Lipnick et al. , where only two of the devices assessed met the International Organization for Standardization (ISO) criteria for accuracy.
Averaging times
Pulse oximetry data needs to be averaged before being displayed as the SpO 2 value, to account for measurement artefact. The parameter that defines the duration of this filter is referred to as ‘averaging time’. Whilst this smoothing out of data is necessary, it is also clinically important that pulse oximeters have a fast response time to changes in oxygen saturation . Several studies have demonstrated that the measurement of transient changes in saturation is affected by increasing averaging times. For instance, Farré et al. showed that desaturation was underestimated by up to 60 % by increasing the averaging time. Furthermore, long averaging times may also lead to the duration of individual events being overestimated, as a cluster of short events may be interpreted as a single prolonged event .
Pulse oximetry measurements
Fig. 2 shows a visualisation of how pulse oximeter readings are displayed when the monitoring is continuous; as a photoplethysmograph, with the average SpO 2 next to it. These data can also be used to calculate and display the heart rate in beats per minute.
Pulse oximetry requires a strong, regular pulse for accurate readings, whether this be a continuous photoplethysmography or snapshots using a portable observation machine. However, many critically ill patients have weak or absent pulses, which may affect the integrity of pulse oximetry readings. Sickle cell anaemia and vaso-occlusive crises are known causes of falsely normal or elevated SpO 2 . This is problematic given oxygen is recommended when oxygen saturations are below a certain level and risks unnecessary and dangerous delays in oxygen administration, thus prolongation of pain and increasing the risk of morbidity and mortality.
As well as skin colour multiple other factors affect the accuracy of pulse oximetry and increase the risk of unreliable readings. There are four types of haemoglobin: haemoglobin, oxygenated haemoglobin (oxyhaemoglobin), carbon monoxide-bound haemoglobin (COHb), and methaemoglobin (MetHb). As the pulse oximeter can only differentiate haemoglobin and oxyhaemoglobin it does not measure the SaO 2 but the ‘oxygen saturation as measured by a pulse oximeter’ SpO 2 . Therefore, a known issue with accuracy relates to the percentage of COHb and MetHb.
A further issue that presents when measuring oxygen saturations, is motion artefact; the muscle movement near the oximeter probe can give rise to ‘spurious pulses’ which may be processed to produce results containing errors. This can also cause issues in situations such as cardiopulmonary exercise testing (CPET), where the oximeter is relied on to give an indication of how well the patient is performing under stress . Another issue in infants is the proportion of foetal haemoglobin, as pulse oximeters are calibrated by inducing hypoxia in healthy adults. Bilirubin can also affect absorbance.
Incorrect readings can lead to unnecessary testing, for example, additional arterial blood gases, increasing the risk of iatrogenic harm and subjecting patients to unnecessary pain and anxiety. Furthermore, frequent erroneous readings can distract healthcare workers and thus impact the safety of other patients .
Probe placement
Probe placement can also have an impact on the accuracy of reported SpO 2 , and it is often necessary to make trade-offs between accuracy and a stronger, consistent signal, such as between the earlobe and finger of a patient with poor circulation . Overall, it seems that earlobe probes produce the most accurate signal. For exercise testing, forehead sensors have the advantage that they are less affected by motion and can possibly reduce the measurement error .
Why skin colour might differentially affect SpO 2
Pulse oximeters work using absorption spectroscopy; two waves that peak at systolic blood pressure and trough at diastolic are demodulated to calculate the difference in light transmission and thus the estimated oxygen saturation . The amount of light that is absorbed between the light-emitting diode and the detecting probe at two different wavelengths is measured and the values are looked up against a normalised table to produce a saturation. In theory, this mitigates differences caused by different thicknesses and hues of skin, but in practice, problems do remain.
Beer’s law, which describes the attenuation of light travelling through a uniform medium containing an absorbing substance, states that the intensity of light travelling through the medium decreases exponentially with distance. The transmittance of light is the ratio of transmitted light to the incident light, and from this, the absorbance can be calculated. Multiple absorbers, for example, blood, tissue, and bone, can be represented mathematically in a calculation representing their extinction coefficient, concentration, and optical path length ( Fig. 3 ). From these, we can determine the concentrations of different absorbing substances if we measure the absorbance of light at different wavelengths and know the extinction coefficients of the substances ( Fig. 4 ).
