Inaccuracy assessment of pulse oximeter in black-skinned adults admitted to ICU in Tikur Anbessa Specialized Hospital
No Thumbnail Available
Date
2024-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Background:The measurement of oxygen saturation via pulse oximeter is a safe and noninvasive method to
approximate arterial oxygen saturation (SpO2), which is an important vital sign in clinical
assessment [1-3]. According to manufacturers, reported values ranging from 70 to 100 percent
are considered accurate within 2 to 3 percent of the true value, as per FDA standards [12,13].
Any variation in oxygen saturation measured by the pulse oximeter (SpO2) compared to the gold
standard measurement by Arterial Blood Gas (ABG) measured oxygen saturation (SaO2) of ≥4%
is considered inaccurate [12,13]. There have been recent reports of systemic racial bias where
pulse oximeter-measured oxygen saturation (Spo2) has been found to overestimate true arterial
oxygen saturation (SaO2) in patients with darkly pigmented skin, leading to concerns about the
clinical accuracy of pulse oximetry [6]. Though this finding is deeply relevant to Africa, there is
no data from Africa including Ethiopia.
Objective: To determine the frequency of Pulse Oximeter inaccuracies in black-skinned adults
admitted to the ICU at TASH in Addis Ababa, Ethiopia between July 2023, and February 2024.
Methodology:We conducted a cross-sectional observational study at a tertiary medical center. While collecting
an arterial blood gas, we utilized a pulse oximeter to measure oxygen saturation in the opposite
arm. Pulse oximeter readings were recorded after 10 seconds of stable measurement. We used
Fitzpatrick skin color tone classification to scale skin color of the patients. Patients with shock,
severe anemia, profound hypoxemia and hypothermia were excluded [3, 5,15].
Statistical Analysis
Quantitative analysis was conducted to compare the oxygen saturation estimates obtained from
the pulse oximeter with the gold standard measurement from the arterial blood gases. The
resulting discrepancies were documented and entered into SPSS for analysis. Descriptive
statistics were then employed to determine the mean discrepancy and the prevalence of pulse
oximeter error.
Result:During our study, we simultaneously measured SpO2 and SaO2 for 72 patients. 37 patients were
excluded due to shock, severe anemia, profound hypoxemia and hypothermia, leaving 35
patients for preliminary result analysis. The skin color of the patients according to Fitzpatric
classification (Table1) were Class III (16 patients), Class IV (12 patients), Class V(6patients) and
Class VI(1patient). The SpO2 readings taken by pulse oximetry showed that 31 patients (88.6%)
had SpO2 ≥ 92%. Among the 4 patients who had SpO2 of ˂92%, 2 (5.7%) had SpO2 between
88-˂ 92%, 1(2.9%) had SpO2 between 80-˂88%, and 1 (2.9%) had SpO2 between 70-˂80%. On
ABG, we found that 17 patients (48.8%) had SaO2 readings of ˂92%. 16 patients (45.7%) had a
pulse oximeter inaccuracy/error. The mean discrepancy of SpO2 to SaO2 was 5.4666 (SD= ±
8.39870, range=-4.8 to 33.4). The prevalence and mean discrepancy percentages of pulse
oximeter error at different ranges of SpO2 sequentially were as follows: SpO2 ≥ 92%
(38.7%,4.95), 88-˂92 % (100%,6.95%), 80-˂88% (100%,5.6%), and 70-˂80% (100%, 29.4%).
Even if the proportion of patients with SpO2 <92% is low, this study indicated that mean
discrepancy might be even more as saturation drops.
Discussions and Conclusion:The results indicate a higher frequency/prevalence of error compared to other
studies conducted in different settings, which reported 11.7% (7) and 17% (9) prevalence of pulse oximeter error in
black patients.
While these preliminary findings suggest that the pulse oximeter overestimates arterial oxygen
saturation relative to ABG-measured arterial oxygen saturation in our sample, it is important to
note that a larger sample size is required to draw a more definitive conclusion.
Recommendations
We should be cautious in interpreting SpO2 in Dark-skinned patients, especially when patients
have low oxygen saturation. Pragmatic studies with larger sample size are needed to observe
differences in Pulse oximeter error across different ranges of SpO2 and skin color classes.
Description
Keywords
ABG, Adult, Black skin, Fitzpatrick classification of skin color tone, ICU, Pulse Oximeter, Pulse oximeter error, SaO2, SpO2.