A Comparison of Finger and Forehead Pulse Oximeters in Heart Failure Patients during Maximal Exercise

Abstract

Background

Pulse oximeters, clinically used to measure oxygen saturation (SpO₂), rely on adequate perfusion of the tissues over which they are placed. Heart failure (HF) patients can have impaired peripheral perfusion which may compromise the accuracy of a peripherally placed pulse oximeter. This decrease in peripheral perfusion may be especially apparent during exercise. The objective of this study was to determine if pulse oximeter accuracy is dependent on location in heart failure patients during peak exercise.

Methods

20 participants with HF (7F, age 64.±11yr) and 9 participants with coronary artery disease as controls (CAD: 3F, age 66±5yr) performed a maximal exertion treadmill exercise stress test while wearing both finger and forehead pulse oximeters.

Results

At peak exercise, the two pulse oximeters measurements of SpO₂ differed from each other by 3.8±3.3% in the HF group (p<0.01) and 2.0±1.4% in the CAD group (p=0.065). The difference between the pulse rate from the pulse oximeters and the heart rate from the 12-lead ECG in the HF group was 12±20 BPM (p<0.01) for the finger pulse oximeter, and 2±3 BPM (p=0.162) for the forehead pulse oximeter.

Conclusions

Forehead pulse oximeters may be more reliable compared to finger pulse oximeters in obtaining SpO2 measurements in HF patients during a treadmill maximal exercise test.

Introduction

Pulse oximeters are routinely used in the noninvasive determination of oxygen saturation in patients, both at rest and during exercise tests¹. These devices are typically placed in areas with high vascular density, such as a finger, ear lobe, or on the forehead. However, these devices are particularly reliant on adequate perfusion to the tissue in which they are placed, and low perfusion may result in inaccurate readings or even complete failure. Finger pulse oximeters rely on perfusion from the radial arteries to the digital arteries, whereas forehead pulse oximeters rely on the supraorbital arteries for SpO2 measurements. Forehead vasculature has limited capacity for vasoconstriction compared to the finger vasculature², and therefore in conditions of high sympathetic output and low peripheral perfusion, such as heart failure (HF)³, finger-placed pulse oximeters may be inferior to forehead-placed pulse oximeters. Further, poor perfusion conditions may be exacerbated by exercise, where increased oxygen demand may result in even further reduced peripheral blood flow and movement artifact can further complicate measurements^{1, 4, 5}. In addition, forehead-placed pulse oximeters may perform better in cases of vasoconstriction, peripheral or otherwise, and have faster response times due to their central location⁶. For these reasons, the forehead-placed pulse oximeter may report more accurate measurements in patients with poor peripheral perfusion during a cardiopulmonary stress test.

Pulse oximeters determine oxygen saturation by emitting several wavelengths of light that are preferentially absorbed by oxyhemoglobin or deoxyhemoglobin. The absorbance is either determined by transmittance, which detects the light after passing through the tissue of interest, or reflectance, which detects the light scattered back towards the sensor. Finger pulse oximeters typically use transmittance, and forehead pulse oximeters use reflectance to determine oxygen saturation. Both pulse oximeter types have previously been shown to perform well under a variety of conditions and patient groups⁷. Pulse oximeters measure heart rate and oxygen saturation by first detecting an arterial waveform and processing this waveform to exclude non-arterial blood readings. From this, arterial oxygen saturation can be measured independent of venous oxygen saturation⁴. Due to the reliance on detecting heart rate, it follows that if the heart rates reported from an ECG and pulse oximeter are discordant, then the reported oxygen saturation from the pulse oximeter is likely to be less accurate due to a failure to reliably detect the arterial waveforms.

Coronary artery disease (CAD) may be managed by similar medications as HF patients, including beta-blockers, ACE inhibitors and aspirin. Additionally, stress tests are used to provide diagnostic and prognostic information for CAD patients⁸. As such, these patients may be more appropriate than healthy individuals as a control population for HF studies.

The purpose of this study was to determine if a forehead-placed pulse oximeter was more accurate when compared to a finger-placed pulse oximeter in HF patients during an incremental exercise test. To do this, 20 HF patients and 9 CAD patients undergoing a clinically indicated cardiopulmonary stress test on a treadmill were instrumented with a forehead pulse oximeter and finger pulse oximeter. It was hypothesized that the forehead pulse oximeter would be more reliable than the finger pulse oximeter in HF patients. Additionally, we expected that the two pulse oximeters would be comparable in CAD patients.

Methods

A total of 34 participants completed the study. Any patient receiving a stress test with the indication of HF or CAD at Mayo Clinic in Rochester, MN was considered for eligibility. Participants were excluded from analysis if they completed a bike stress test (n=3) or if one or more of the devices failed to collect data (n=2). Therefore, 29 patients were included in data analysis. Of the 20 HF patients, six (30%), eight (40%) and six (30%) were NYHA class I, II, and III, respectively. HF patients were included with any co-morbidity, and CAD patients were included if they were on either beta-blockers or ACE-inhibitors. Participants were excluded if they had a BMI >40, or smoked in the 28 days immediately prior to the test. This study was reviewed and approved by the Mayo Clinic institutional review board.

All studies occurred in a dedicated clinical exercise stress testing room. After consent, height and weight, participants were instrumented with a 12-lead ECG and a mask or mouthpiece for respiratory gas analysis. A Nonin 8000R PureLight Sensor (Nonin Medical Inc, Minneapolis, MN) was placed on the left lateral forehead, and a Nonin WristOx₂, Model 3150 (Nonin Medical Inc, Minneapolis, MN) was placed on the right middle finger. Both pulse oximeters sample at a rate of 1 Hz. Participants were asked to keep the hand with the finger pulse oximeter as stationary as possible by placing their hand on a handlebar with minimal grip force in efforts to avoid motion artifacts. After a two minute standing baseline, participants then began walking on a treadmill using the Mayo O2 protocol, which advances in difficulty every two minutes⁹. The stages are: 2mph, 0% grade; 2mph, 7% grade, 2mph, 14% grade; 3mph, 12.5% grade; and 3mph, 17.5%, and so on. Once the participant can go no further, the treadmill returns to 1.7mph, 0% grade for 3 minutes of recovery.

Statistical analysis

Student’s t-tests and Chi-squared tests were used to determine differences in baseline characteristics and demographics between CAD patients and HF patients. Mixed model ANOVAs were performed to determine differences in between SpO₂ and heart rate as reported by two and three different devices, respectively, in CAD and HF patients independently across all stages of exercise. A one-way ANOVA was used to compare the pulse rates and heart rate as reported by pulse oximeters and 12-lead ECG. Bland-Altman plots were used to compare individual SpO₂ and heart rate as reported by different devices. All values are reported as mean and standard deviation for continuous variables, or count and percentage for categorical variables. All statistical analyses were performed using JMP Pro (version 13.0.0, SAS Institute, Cary, NC).

Bland-Altman plots visually represent data in which two methods or techniques are used to measure the same variable with the goal of comparing the measurements. Briefly, bias is the amount in which the two methods differ on average, and the agreement range is ±1.96 standard deviations away from the bias.¹⁰

Results

Participant Characteristics

Participant characteristics for the 20 (7F) HF NYHA class I-III patients and 9 (3F) CAD patients who participated in this study are provided in Table 1. Briefly, HF patients had an average ejection fraction of 38±18%, were well distributed between NYHA classes I-III, and 60% of them were of non-ischemic etiology. All participants completed the standardized protocol to point of complete exhaustion. HF patients exercised for 420±105 seconds, while CAD patients exercised for 537±90 seconds (p<0.05). Peak VO2, and exercise time were all significantly lower in the HF group (p<0.05). Aspirin use was significantly lower and ACE-I/ARB use was significantly higher in the HF group (p<0.05). The test was terminated when the participants reached a rate of perceived exertion of 18 or greater, a standard used to determine the patient’s current exertion and peak VO2.

Table 1.

Baseline clinical and demographic characteristics of the sample.

	CAD (N=9)	HF (N=20)	p value
Clinical Variables
Age, yrs1	66 (5)	64 (11)	0.598
Sex, F2	3 (33%)	7 (35%)	0.930
Height, cm1	171.4 (10.4)	174.5 (9.3)	0.436
Weight, kg1	81.7 (16.9)	88.4 (17.7)	0.353
BMI, kg/m^21	27.7 (4.5)	28.9 (4.3)	0.497
Systolic BP, mmHg1	126 (7)	115 (15)	0.058
Diastolic BP, mmHg1	74 (14)	72 (10)	0.705
LVEF, %1	58.4 (7.5)	38.4 (17.6)	0.003
NYHA Functional Class, %
I	-	6 (30.0%)
II	-	8 (40.0%)	-
III		6 (30.0%)	-
HF Etiology
Ischemic		8 (40.0%)	-
Non-Ischemic	-	12 (60.0%)	-
Peak VO2, mL/kg/min1	24.9 (4.1)	19.0 (6.3)	0.016
Time to Exhaustion, sec1	537 (90)	420 (106)	0.008
Medications, %2
Beta-Blocker	8 (89%)	19 (95%)	0.548
Calcium Channel Blocer	0 (0%)	2 (10%)	0.326
ACE-Inhibitor	2 (22%)	13 (65%)	0.033
Aspirin	9 (100%)	11 (55%)	0.015
Medical History, %2
CABG	4 (44%)	4 (20%)	0.173
Chronic Kidney Disease	0 (0%)	6 (30%)	0.065
COPD	0 (0%)	1 (5%)	0.495
Diabetes	1 (11%)	5 (25%)	0.393
Atrial Fibrillation	1 (11%)	10 (50%)	0.046
Myocardial Infarction	5 (56%)	6 (30%)	0.189
Obstructive Sleep Apnea	4 (44%)	9 (45%)	0.978
Hypertension	7 (78%)	11 (55%)	0.242
Implanted Cardiac Device	1 (11%)	10 (50%)	0.046

LVEF, Left Ventricular Ejection Fraction; NYHA, New York Heart Association; VO2, oxygen uptake; ACE, angiotensin-converting enzyme, HF, heart failure; CABG, coronary artery bypass graft; COPD, chronic obstructive pulmonary disease.

¹Linear Model ANOVA

²Pearson’s Chi-squared test.

Heart Rate and Oxygen Saturation during Exercise Testing

There was a main effect of pulse oximeter location on SpO₂ in both the CAD and HF patients (Figure 1a–b, p<0.05). In both cases, the finger-placed pulse oximeter read lower than the forehead-placed pulse oximeter. In the CAD patients, there was a main effect of percent of peak VO₂ on SpO₂ (Figure 1A, p<0.05). This effect was not seen in HF patients (Figure 1B, NS). Pulse rate values reported by the pulse oximeters did not differ across 12-lead ECG, finger or forehead-placed pulse oximeters in either group (Figure 2A&B, NS). In both groups, heart rate increased as percent of peak VO₂ increased (Figure 2A&B, p<0.001).

Figure 1.

Oxygen saturation across the increasing levels of exertion during the treadmill exercise stress test for coronary artery disease patients (A) and heart failure patients (B). In both groups, there was a main effect of device used (p<0.05), with no interaction with the exercise stage. In coronary artery disease patients but not heart failure patients, there was a main effect of exercise stage (p<0.05). Peak is defined as the average of the last 30 seconds of exercise; recovery is defined by the first 60 seconds of the walking recovery period, baseline is defined as the last 30 seconds of baseline prior to the exercise protocol beginning.

Figure 2.

Heart rate across the increasing levels of exertion during the treadmill exercise stress test for coronary artery disease patients (A) and heart failure patients (B). In both groups, there was a main effect of exercise stage (p<0.0001) but no main effect of device used nor an interaction between the two (p>0.05).

Heart Rate at Peak Exercise

At peak exercise, finger pulse oximeter, forehead pulse oximeter and ECG reported similar values for heart or pulse rate in the CAD group (131±13, 126±12, and 134±14 BPM, respectively; Figure 3A, NS). In the HF group, the finger-placed pulse oximeter reported significantly lower pulse rate than either the forehead-placed pulse oximeter or ECG (104±24, 113±25, and 134±14 BPM, respectively; Figure 3B, p<0.05).

Figure 3.

Heart or pulse rates at peak exercise in coronary artery disease (A) and heart failure patients (B) as reported by the forehead pulse oximeter, finger pulse oximeter, and 12-lead ECG. In heart failure patients, the finger pulse oximeter read lower compared to the forehead and 12-lead ECG (p<0.05, one way ANOVA). No statistical differences we found between measurement devices in the coronary artery disease group.

At peak exercise in HF patients, comparisons between the finger-placed pulse oximeter pulse rate and ECG heart rate showed a bias of −12 and agreement range of ±40 BPM (Figure 4C), while the forehead pulse oximeter pulse rate and ECG pulse rate showed a bias of −2 and agreement range of ±12 BPM (Figure 4D). In CAD patients, finger-placed pulse oximeter pulse rate and ECG heart rate showed a bias of 3 and agreement range of ±5 BPM (Figure 4A), while the forehead pulse oximeter pulse rate and ECG pulse rate showed a bias of −7 and agreement range of 22 BPM (Figure 4B).

Figure 4.

Bland Altman plot comparison of ECG heart rates and pulse rates from both finger (A, C) and forehead (B, D) pulse oximeters at peak exercise in CAD (A, B) and heart failure (C, D).

Oxygen Saturation at Peak Exercise

At peak exercise, comparisons between the two pulse oximeters measurements of SpO2 showed a bias of −3.8% and agreement range of ±6.3% in the HF group, and a bias of 2.0% and agreement range of ±2.8% in the CAD group (Figures 5A & B). These relationships are shown in a Bland-Altman plot¹⁰.

Figure 5.

Bland Altman plot comparison of Finger and Forehead SpO₂ at peak exercise in (A) heart failure and (B) coronary artery disease patients.

Discussion

The data presented show that the finger and forehead-placed pulse oximeters are similarly accurate in determining heart rate and SpO₂ in patients with CAD (Figures 1A, 2A, 3A, 4A, 4B, 5A). In HF patients, both the finger and forehead-placed pulse oximeters reported similar values for pulse rate at non-peak exercise (Figure 2B), but the finger pulse oximeter reported lower pulse rates at peak exercise compared to the forehead-placed pulse oximeter and 12-lead ECG (Figures 1B, 3B, 4C). In HF patients, the forehead pulse oximeter pulse rate showed low bias and tight agreement with the 12-lead ECG heart rate at peak exercise (Figure 4D). In HF patients, the SpO₂ reported by the finger-placed pulse oximeter was lower than the SpO₂ reported by the forehead-placed pulse oximeter during both non-peak exercise (Figure 1B) and peak exercise (Figures 1B, 5B).

Overall, these data show that finger-placed pulse oximeters can accurately measure CAD patients’ SpO2 and heart rate during an exercise stress test, but are generally less accurate in HF patients. We hypothesize that this due to reduced peripheral perfusion in HF patients due to low cardiac output. Pulse oximeters determine the oxygen saturation of the arterial blood by first detecting the arterial waveform and filtering out non-arterial blood readings⁴. As such, pulse oximeters do not function well under conditions with low perfusion where these arterial waveforms are diminished. Acral skin, such as the fingertip, is influenced greatly by increased sympathetic tone resulting in more dramatic decreases in perfusion¹¹. As HF patients have increased sympathetic tone, it is likely the primary reason that finger pulse oximeters do not perform well in HF patients. This could be exacerbated during exercise, which may further increase sympathetic activity.

In concordance with this study, a study of patients with low cardiac index at rest showed that forehead-based pulse oximeters are more accurate at determining oxygen saturation than digit-based pulse oximeters when compared to arterial blood gases¹². Our study expands upon this by testing pulse oximeters during heavy exercise. In addition, a study of surgical and trauma patients that are at risk for poor perfusion in the periphery showed that forehead pulse oximeters were more accurate in measuring oxygen saturation when compared to arterial blood gases¹³, though it has been shown that finger pulse oximeters can be accurate in low perfusion situations when the devices with considerations for poor-perfusion are used¹⁴. Patients in transport, which are subject to movement and cold ambient temperatures, have been shown to have fewer erroneous measurements and failures in measuring SpO₂ when forehead-placed pulse oximeters are used compared to a finger-placed pulse oximeter¹⁵.

These results contrast against findings from a 1991 study investigating 21 different pulse oximeters’ performance against arterial blood oxygen saturation values in cardiac surgery patients, which found that finger pulse oximeters were superior to ear, nose, and forehead pulse oximeters¹⁶. Similarly, a 1988 study found that forehead pulse oximeters are not reliable in critically ill patients when compared to arterial blood oxygenation¹⁷. This discordance may be reconciled considering advances in pulse oximeter technology since 1991, and given the additional consideration of exercise in the current study. Pulse oximeters have advanced significantly, with improvements in signal processing to allow for lower signal to noise ratios, detection of carboxyhemoglobin and methemoglobin, and more advanced waveform analysis¹⁸. Some recent studies have suggested that finger pulse oximeters may be superior to forehead pulse oximeters in several settings. A 2010 study interested in the ability of disposable finger oximetry sensors to take pulse oximetry measurements on the forehead concluded that forehead placement resulted in inaccuracy¹⁹. A 2017 study showed that in intensive care unit patients, finger and earlobe probes were superior to forehead probes when compared to arterial oxygen saturation²⁰. Again, however, the additional variables of exercise and peripheral perfusion may be responsible for discordant findings in HF patients. A 2018 study showed that finger pulse oximeter probes are more accurate than forehead probes in the outpatient setting, though both probes reported significantly different saturation levels when compared to blood gases²¹.

As only one manufacturer of pulse oximeters was used in this study, these findings may not be applicable to all brands and models of pulse oximeters. Additionally, arterial blood samples were not collected for arterial blood gas analysis, thus we cannot be certain which of the two pulse oximeters is reading more accurately. However, due to the concordance between forehead-measured pulse rate and ECG-measured heart rate at peak exercise in HF (Figure 4D) and simultaneous non-concordance between HR measured by finger pulse rate and ECG at peak exercise in HF (Figure 4C), it is inferred that the forehead-placed pulse oximeter is measuring SpO₂ and pulse rate more accurately than the finger pulse oximeter in these HF patients.

Due to the importance of oxygen saturation in HF patients, especially during peak exercise, it may be worth considering a shift from the heavy reliance on the traditional finger pulse oximeter when providing medical care to HF patients. The data presented within this study suggest that forehead pulse oximeters are superior during an exercise stress test in HF patients, which may reflect a difference in anatomy and vasoconstriction in response to the sympathetic nervous system². For this reason, it may be warranted to investigate accuracy of oxygen saturation measurements in HF patients during sleep, decompensated HF, and routine tests such as echocardiography or six-minute walk tests.

Conclusion

These observations expand on the current understanding of pulse oximetry function in HF patients or in those with low cardiac output by testing these devices during exercise. Overall, these data suggest that the finger-placed pulse oximeter systematically reports lower oxygen saturation values than the forehead pulse oximeter, and that the finger pulse oximeter is poor at detecting pulse rate during peak exercise. This, coupled with previous work on these devices, suggests that forehead-placed pulse oximeters are superior to finger-placed pulse oximeters in HF patients for determination of oxygen saturation and that forehead pulse oximeters should be used during exercise.

Highlights.

• Heart failure patients have reduced ability to perfuse tissues with blood, especially during exercise, potentially limiting pulse oximeter performance.

• Heart failure and coronary artery disease patients completed a maximal exertion exercise test while wearing a finger and forehead pulse oximeter.

• The forehead pulse oximeter was found to be more reliable than finger pulse oximeters in heart failure patients during peak exercise.

• Using forehead models in clinical situations requiring pulse oximeters may be beneficial for patients with heart failure.