Placement of finger oximeter on the ear: comparison with oxygen saturation values taken from the finger

Pulse oximetry is an indirect, non-invasive, accurate and safe method of measuring oxygen saturations (SpO₂). It is widely used in a range of outpatient and inpatient settings to record clinical observations, which includes calculation of the patient's National Early Warning Score 2 (NEWS 2) (Royal College of Physicians, 2017). A number of treatment protocols are guided by the results of oxygen saturation measurements, for example during assessment for, and response to, interventions such as oxygen therapy (O'Driscoll et al, 2017) and non-invasive ventilation (NIV) (Davidson et al, 2016). More recently, pulse oximetry has been used to monitor the condition of patients with COVID-19 both in hospital (Shenoy et al, 2020) and home settings (Luks, 2020).

Pulse oximeters are designed to record SpO₂ by measuring the absorption of specific wavelengths of light by oxygenated haemoglobin (HbO₂) versus that in deoxygenated haemoglobin (Hb). Oximeter probes contain light-emitting diodes (LEDs) that project light of two wavelengths—red (660 nm) and infrared (940 nm)—from one side of the probe towards a photodetector on the opposite side. Pulsatile arterial blood during systole delivers oxyhaemoglobin (HbO₂) to the tissue, which results in the absorption of more infrared light and so less light will reach the photodetector. The level of oxygen saturation of the blood therefore determines the degree of light absorption. The result is processed and a digital readout of the oxygen saturation results is shown on the oximeter screen, represented as SpO₂.

The accuracy of a pulse oximeter is determined by the differences between SpO₂ measured on oximetry and oxygen saturation measured via arterial blood gas (ABG) sampling (SaO₂) performed at the same time. Manufacturers claim an accuracy of 2% with a standard deviation (SD) of the differences of 2-3% (Nitzan et al, 2014; Jubran, 2015), but there is evidence that measurement errors are generally closer to 3-4% (Nitzan et al, 2014). High reliance is also placed on the continuing reliability of the wavelength of light emitted, however functional changes, such as LED general wear and tear or damage to due to cleaning, can change the rate of light absorption, which will affect accuracy of SpO₂ estimates. This can result in up to 30% of pulse oximeters failing to perform according to the manufacturer's specifications (Milner and Mathews, 2012). In addition, the algorithms developed to convert light absorption ratios into arterial saturation estimates for use with pulse oximeters were developed using healthy volunteers. Consequently, the mean difference between SaO₂ and SpO₂ becomes greater when SpO₂ falls to below 80% as a result of inaccuracies in calibration and co-oximetry.

The accuracy of oximetry readings are affected by the strength of arterial pulse, body movements, colour interferences, venous pulsations and several physical factors. Measured saturations also fluctuate significantly with ventilation changes related to coughing, talking, breath-holding and physical activity. Although easy to perform, pulse oximetry requires clinical skills training to ensure that accurate readings are obtained (Fluck et al, 2003). Evidence suggests that clinicians are not always aware of the limitations of pulse oximetry and that motion artefact and poor signal quality can result in inaccurate readings (Jubran, 2015). Due to these factors, it is essential that saturation measurements are observed for several minutes to identify the most frequently measured value, rather than relying on the first value provided (Luks, 2020).

Because a pulse oximeter measures the amount of light transmitted through the tissue, any bright light shining directly on the sensor has the potential to produce inaccurate readings. Where a sensor is applied incorrectly or to a tissue site not intended by the manufacturers, for example using a finger probe on the ear, optical shunting can occur, with light reaching the detector without passing through the blood-perfused tissues (Mannheimer, 2007). The impact on the measured SpO₂ will depend on which of the light wavelengths are subject to optical shunting, as well as external light sources, which may consequently increase or decrease the true value recorded. For this reason, it is imperative that oximetry probes are used only on the anatomical site they were designed for (Clayton et al, 1991).

Studies (Haynes, 2007; Johnson et al, 2012) reported that ear placement of a probe designed for use on a finger overestimates SaO₂ by ≥3%. In populations of healthy normal and normoxic outpatients (with SpO₂≥90%) this error did not affect clinical management, but it did demonstrate inappropriate probe placement affected treatment decisions. Malhotra et al (2018) found that ear placement of a finger probe led to the overestimation of oxygen saturations ranging from 0.1% to 12% in patients attending for oxygen assessments. Furthermore, a Patient Safety Alert issued by the NHS Improvement (2018) reported the potential risk of harm arising from inappropriate placement of pulse oximeter probes. The alert raised concerns about the possibility of false reassurances about a patient's clinical condition, whereas in reality the patient was rapidly deteriorating.

Despite this evidence, the authors have identified that clinicians continue to position finger probes on patients' ears (pinna), when readings from the finger are not possible or when values from the finger probe appear unjustifiably low. Consequently, they recognised differences between clinical observations recorded in NEWS 2 and the subsequent measurements obtained from either finger oximetry or blood gas sampling in patients treated with NIV. The clinical impact of overestimating SaO₂ in a hypoxic patient group may have a greater impact, leading to immediate or long-term risk from non-treatment and/or misinterpretation of clinical deterioration.

The authors evaluated the likely consequences for patient management following SpO₂ measurements using a finger probe placed on the pinna of the ear in a cohort of patients receiving NIV for acute hypercapnic respiratory failure (AHRF).

Methods

This was a prospective, single-centre, cohort study performed at University Hospitals Coventry and Warwickshire (UHCW) NHS Trust, a large teaching hospital serving a population of approximately 1 million. The study protocol was reviewed by the hospitals research and development department; ethical approval was obtained (Study Ref: GF0291).

From September 2019 all patients admitted with AHRF and subsequently treated with NIV as per British Thoracic Society (BTS) guidelines (Davidson et al, 2016) were selected in a consecutive series. Verbal consent for SpO₂ measurement on the ear using a finger probe was obtained for each patient prior to testing. Finger probe SpO₂ measurement (Connex Spot, Welch Allyn) and arterial blood (ABG) testing (ABL800, Radiometer) were performed concurrently and as part of routine clinical care. Measurements were taken according to a department-specific clinical operating procedure for managing patients on acute NIV and were undertaken when assessment for therapy or assessment of effectiveness of therapy was required. To ensure consistency of measurements these were performed by staff who had been assessed for competency and undergone good clinical practice (GCP) training.

Patients were excluded from the study if there was clinical evidence of recognised limitations and sources of error for oximetry, including poor peripheral circulation, dye/nail polish, hypothermia (temperature <35°C) or hypotension (systolic blood pressure (BP) <100 mmHg).

Quality assurance of the blood gas measurements was ensured by daily two-point calibration and automated quality control. To quality assure the SpO₂ signal, for all measurements, prior to recording the SpO₂ value photoplethysmographic waveforms were visually assessed to ensure an adequate signal was obtained. Measurements of SpO₂ were in the order of, first, the finger and then the pinna of the ear. The performances of ABG and finger probe SpO₂ were in accordance with the Association for Respiratory Technology and Physiology (ARTP) and British Thoracic Society guidelines (ARTP/BTS, 1994). The NEWS 2 value was calculated using observations taken at the time of saturation measurements.

Statistical analysis

All data were tabulated using Microsoft Excel. Statistical analysis was performed using SPSSv26. Data were examined for normality using the Kolmogorov-Smirnov test. Oxygen saturation measurements (SpO₂) obtained via both ear and finger oximetry were compared with oxygen saturation measurements (SaO₂) obtained from ABG.

A subgroup analysis of normoxic and hypoxic (SpO₂<90%) patients was also performed. One sample t-test was used to calculate mean bias of the difference in techniques and significance. Bland-Altman analysis was performed to evaluate the agreement between the methods and linear regression to assess proportional bias.

Results

Forty-six (28 male) consecutive patients admitted with AHRF requiring NIV were included in the data analysis. The majority (83%) were admitted with acute exacerbations of chronic obstructive pulmonary disease (COPD), with the remaining 17% presenting with AHRF due to obesity hypoventilation or chest-wall deformity. The baseline characteristics of the study population are presented in Table 1. ABG for the group demonstrated stable hypercapnic respiratory failure. In total, 16/46 (35%) were prescribed supplemental oxygen at the time of undertaking measurements (FiO₂ 22.5-35%) and 23/46 (50%) demonstrated an SaO₂ of <90%.

No statistically significant difference was found between ABG analysis (SaO₂) and finger SpO₂ values (P>0.05) (Figure 1). However, there was a statistically significant difference between ear SpO₂ and SaO₂ ABG (P<0.001) (Figure 2) when using a finger probe. There was no statistically significant difference between the NEWS 2 score calculated using saturation from the finger versus that from the ear (P=0.460), which is a concern.

Figure 1. Bland-Altman analysis assessing agreement between oxygen saturation measurements in arterial blood via ABG (SaO2) and pulse oximetry via the finger (SpO2). The solid line represents the mean bias (-0.66%) and the dotted lines represent the limits of agreement (+/- 1.96 standard deviations) Figure 2. Bland-Altman analysis assessing agreement between oxygen saturation measurements in arterial blood via ABG (SaO2) and pulse oximetry via the ear (SpO2). The solid line represents the mean bias (-4.29%) and the dotted lines represent the limits of agreement (+/- 1.96 standard deviation)

Differences between techniques were also evaluated in a subset of patients with evidence of hypoxia on blood gas. In total, 50% of patients demonstrated hypoxia (SaO₂<90%). There were statistically significant differences between ABG SaO₂ and finger and ear SpO₂ (P=<0.05 and P=<0.001 respectively), specifically in patients with SaO₂<90% (Figure 3 and Figure 4). There was also evidence of proportional bias with a linear regression of -0.498 (P<0.05) and -0.422 (P<0.05) for finger and ear placement respectively.

Figure 3. Bland-Altman analysis of the agreement between oxygen saturation measurements in arterial blood via ABG (SaO2) and pulse oximetry via the finger (SpO2) in subjects with an SaO2 of <90%. The solid line represents the mean bias (-1.11%) and the dotted lines represent the limits of agreement (+/- 2 standard deviations). The dashed line represents the regression line (y=29.17-0.35*x) and confidence interval limits are presented as a continuous line Figure 4. Bland-Altman analysis of the agreement between oxygen saturation measurements in arterial blood via ABG (SaO2) and pulse oximetry via the ear (SpO2) in subjects with an SaO2 <90%. The solid line represents the mean bias (-5.72%) and the dotted lines represent the limits of agreement (+/- 1.96 standard deviations). The dashed line represents the regression line (y=47.5-0.59*x) and confidence interval limits are presented as a continuous line

Discussion

The results of the study showed that SaO₂ measured via ABG sampling does not differ significantly from SpO₂ measured via the finger using a finger probe oximeter. The mean difference of -0.66% is within the 2% error described by manufacturers (Nitzan, 2014).

However, the data have demonstrated that it is not appropriate to put a finger probe oximeter on the pinna of the ear to estimate SaO₂. Engaging in this type of clinical practice may lead to a significant overestimation of SpO₂, which, when used in a clinical decision-making tool such as NEWS 2, would result in a failure to escalate treatment. The findings of the study are in keeping with those reported in previous published literature that demonstrated both statistically and clinically meaningful differences between SaO₂ and SpO₂ when measured on the ear with a finger probe (Malhotra et al, 2018).

NEWS 2 values calculated using finger-determined saturation versus pinna-determined saturation did not reach the threshold for statistically significant difference. However, it is important to note that in 10 patients, using the SpO₂ measurements made on the ear led to the overscoring of NEWS 2, suggesting that there was a clinically significant difference. This clinically significant difference relates to saturations measuring >95% on the ear and the impact this has on scoring using NEWS 2, increasing observation frequency and consequently the workload of nursing staff. In addition, six subjects exhibited an ear SpO₂ that was 10% greater than ABG SaO₂. This, when contextualised within clinical practice on an enhanced respiratory ward during the COVID-19 pandemic, may of course lead to an inability to recognise a deteriorating patient who would otherwise benefit from escalation to the critical care ward. These differences most likely relate to optical shunting, caused by external light sources registering on the photodetector due to poor fitting of the finger probe on the patient's ear (Clayton et al, 1991).

In addition to other studies, the authors of this study have presented data on a subgroup of hypoxic patients and demonstrated less accuracy when using SpO₂ to estimate a patient's SaO₂ in the presence of hypoxia. The mean difference of -1.11% for finger oximetry is significantly lower than the -5.72% mean difference seen for the ear, however, both were clinically significant and demonstrated a proportional bias, with both overestimating SaO₂ with increasing oxygen saturation. This reiterates an important clinical point of using the appropriate method to take measurements, ie ensuring that blood gas testing is used to obtain a measurement of partial pressure of oxygen (PaO₂) in patients if there is a concern of hypoxia to inform subsequent decisions about treatment.

Limitations

Quality assurance of saturation measurements was achieved by visual inspection of the photoplethysmographic waveform, however the pulse oximeters used in this study were not calibrated before use. There is acknowledgement that as many as 30% of NHS oximeters fail to perform to manufacturer specifications (Milner and Mathews, 2012). Inaccuracy of a pulse oximeter could lead to discrepancies between blood gas-determined and oximeter-determined saturations; however, because the same monitor was used for ear and finger oximetry readings in this study, differences in these measurements are less likely to be due to calibration inaccuracies.

Incorrect application of an oximeter probe can lead to external light reaching the detector without first passing through a patient's tissue, resulting in inaccurate saturation readings. The results of this study suggest that optical shunting plays a significant part in the large differences between ear and finger saturation readings. Patients participating in this study were all on the same ward at the time of testing, but the time of day and patient accommodation were not standardised. Patients in large bays will have potentially been exposed to a greater degree of optical shunting than those in side rooms, where the windows are significantly smaller. In addition, the study recruited patients during the autumn and winter months, at a time when artificial lighting will have been in use for a larger proportion of the day.

The authors acknowledge that a sample size was not formally calculated as part of the study's methodology, but the study population size is comparable with that of previously published studies (Haynes, 2007).

Conclusion

Previous studies have identified that placement of a finger probe on the ear inaccurately represents arterial saturation in stable patients with chronic respiratory disease attending outpatient clinics. This article further contributes to the evidence base demonstrating that placement of finger oximetry probes on the ear in acutely unwell patients has a high likelihood of leading to inaccurate results. This may well hold greater clinical significance when SpO₂ is used to calculate a patient's NEWS 2 value, and when this is subsequently used to make rapid decisions about ongoing patient care. Of course, acutely unwell patients present other limitations for accurate measurements of finger SpO₂, for example during periods of hypotension and/or poor peripheral circulation, and therefore an appropriate probe designed to be used on the patient's earlobe should be selected (Mannheimer, 2007).

Finally, this study has provided evidence to identify that placement of a finger probe on the ear should be viewed as unsafe clinical practice.

Recommendations

Measurements of SpO₂ should always be checked for accuracy by visually assessing the photoplethysmographic waveform on the oximetry device using a probe appropriate to the anatomical site. Reduced peripheral circulation identified by poor waveform strength should be recognised as a measurement limitation that will lead to inaccurate SpO₂ readings. In these circumstances alternative methods for determining oxygen saturation, for example, ear lobe/forehead probes or blood gas measurements, should be employed.

KEY POINTS

• Oxygen saturation measurements are influenced by external factors, including site placement
• Finger probe placement on the ear is used in an attempt to improve signal accuracy. Finger probe placement on the pinna of the ear leads to artificially increased saturation readings, which become increasingly inaccurate in hypoxic patients
• In hypoxic patients, blood gas measurements are more accurate than saturation readings and should be used to inform subsequent treatment decisions

CPD reflective questions

• Reflect on the key learning points of this article and how they may change your future clinical practice
• In addition to the clinical knowledge gained from reading this article, what other information do you need in order to extend your own professional development in the topic area
• Reflect on how the clinical practices discussed in this article relate to the Nursing and Midwifery Code of practice
• Can you relate any of the key learning points of the article to your own areas of clinical practice? If so, how will you use your new knowledge to ensure the delivery of safe patient care?