Table 1.
Pertinent studies that compared the accuracy of pulse oximeters in groups of different race or skin pigmentation (available to access).
| Study | No. subjects | Setting | Method for stratifying skin pigmentation | Statistical information | Findings |
|---|---|---|---|---|---|
| Wang et al (1985) (Wang and Poh 1985) | 31 ‘pigmented patients’ | Not available | Not available | Not available | Biox III oximeter gave readings that closely approximated oxygen saturation measurements from arterial blood, confirming the usefulness of non-invasive oximetry even in pigmented patients. |
| Gabrielczyk and Buist (1988) | 21 (4 ‘racially pigmented’) | Post-cardiac surgery | Unclear | Bias ± SD (Overall): +0.6 ± 1.6% a | Statistically insignificant difference in accuracy of SpO2 between skin pigmentations. |
| Cecil et al (1988) | 152 (136 White, 1 Asian, 15 Black) | Not stated | Scale of 1 to 3—Light, medium, or dark pigment | Bias (Overall): Nellcor, +0.59%. Ohmeda, −0.897% a | Both oximeters displayed a statistically significant, but clinically insignificant bias when compared with arterial blood oxyhaemoglobin. Greater accuracy in SpO2 measurements taken from Black patients was observed by the Ohmeda oximeter. |
| Ries et al (1989) | 187 Distribution of racial groups not stated | Laboratory | Munsell colour system | Bias (Overall): Ohmeda, +1.4%. Hewlett Packard, −0.6% a | Readings were slightly less accurate in patient groups with darker skin, suggesting that dark skin colour may. affect the performance and accuracy of ear oximeters. |
| Jubran and Tobin (1990) | 54 (29 Black, 25 White) | Intensive Care | Visual | Bias ± SD (White): 2.2 ± 1.8% a | SpO2 was less accurate and less precise in Black patients. |
| Bias ± SD (Black): 3.3 ± 2.7% a | |||||
| Cahan et al (1990) | 28 (22 White, 6 Black) | Hypoxia Laboratory | Unclear | Bias ± SD (White): 1.9 ± 5.1% a | Pulse oximetry values can be higher in Black subjects than in White subjects, especially at saturations below 80%. |
| Bias ± SD (Black): 5.1 ± 4.6% a | |||||
| Lee et al (1993) | 33 (22 Chinese, 6 Malaysian, 5 Indian) | Intensive Care | Race | Bias ± SD (Overall): +0.82 ± 2.6% a | SpO2 was most accurate in Chinese patients, and less accurate (greater overestimation) in Malaysian and Indian patients. It was also less accurate at low saturations and in patients with elevated bilirubin levels. |
| Range: −4.9% − 10.5% | |||||
| Bothma et al (1996) | 100 darkly pigmented adults | Intensive Care | Reflectance spectrophotometer | Bias (Overall): −1.0% − 1.2% a | The accuracy of pulse oximetry is not adversely affected by skin pigmentation, and it remains a useful oxygenation-monitoring device in darkly pigmented patients. |
| LOA: −6.6% − 6.6% across all pulse oximeters | |||||
| SD: 1.9%–2.4% across all pulse oximeters | |||||
| Adler et al (1998) | 278 (34 Dark, 101 Intermediate, 143 Light) | Emergency Department | Munsell colour system | Bias ± SD (Light): +2.5 ± 4.6% a | The accuracy and precision of SpO2 was not affected by skin pigmentation. Signal quality was poorer in a great proportion of patients with dark skin pigmentation than intermediate or light. |
| Bias ± SD (Intermediate): +2.8 ± 5.2% a | |||||
| Bias ± SD (Dark): +2.2 ± 3.7% a | |||||
| Bickler et al (2005) | 21 (11 Dark, 10 Light) | Hypoxia Laboratory | Race | Bias ± SD (Light): +0.37 ± 3.20% a | SpO2 was less accurate (greater overestimation) in Darkly pigmented subjects. This bias was greater at lower saturations, and differed between pulse oximeters. |
| Bias ± SD (Dark): +3.56 ± 2.45% a | |||||
| Feiner et al (2007) | 36 (17 Dark, 7 Intermediate, 12 Light) | Hypoxia Laboratory | Race | Bias range for intermediate and dark skin: +4.5%–+4.9% (SaO2 = 60% − 70%) a | SpO2 was less accurate (overestimation) in subjects with Dark and Intermediate pigmentation at lower saturations for five out of six combinations of pulse oximeters and probe types (the exception being the Masimo Radical with adhesive probe). |
| +2.4%–+3.6% (SaO2 = 70% − 80%) | |||||
| Witting and Scharf (2008) | 837 (577 African American, 260 White) | Emergency department | Self-report | Not available | African American group without hypoxaemia was associated with a 0.8-unit increase in SpO2 values, and without a change in precision. African American females and White females had an average SpO2 higher than African American men and White men, respectively. Clinicians should regard these findings with particular significance. |
| Foglia et al (2017) (Foglia et al 2017) | 35 (14 Dark, 21 Light) | Infants with congenital heart disease | Munsell colour system | Bias ± SD (Overall): Nellcor, +3.9 ± 2.0% Masimo, +0.8 ± 2.4% a | No significant difference in SpO2 accuracy between patients with Dark and Light pigmentation. |
| Ebmeier et al (2018) | 394 Distribution of racial groups not stated | Intensive Care | Fitzpatrick scale | Bias (Light versus Medium): −0.9% a | The accuracy of SpO2 was influenced by skin pigmentation and pulse oximeter model (Masimo underestimating, Philips overestimating). |
| Bias (Light versus Dark): −2.4% a | |||||
| LOA: −4.4%–4.4% | |||||
| Murphy and Omar (2018) | 146 (6 Light, 111 Medium, 29 Dark) | Intensive Care | Massey New I mmigrant Survey skin colour scale | Bias ± SD (Overall): +1.64 ± 0.15 g dL−1a | The degree of skin pigmentation does not appear to influence the magnitude of bias, rather the increasing severity of illness and decreasing lower mean arterial blood pressure. |
| LOA (Overall): −1.05 g dL−1–4.33 g dL−1 | |||||
| Smith and Hofmeyr (2019) | 220 12 Type I, 28 Type II, 69 Type III, 45 Type IV, 28 Type V, and 38 Type VI | Perioperative areas e.g. pre-assessment clinics, recovery rooms, operating theatres and intensive care units | Fitzpatrick scale | Bias (Overall): −0.55% a | Darker skin pigmentation showed no trend to an effect on the accuracy of oxygen saturation measured using portable fingertip pulse oximeter and a conventional bedside pulse oximeter. |
| LOA: −3.25%–2.16% | |||||
| Sjoding et al (2021) | 10,001 (1,326 Black, 8,675 White) | Intensive Care, and Inpatients receiving oxygen | Race | Not available | SpO2 was less accurate (overestimation) in Black than White patients. Hidden hypoxemia (arterial oxygen saturation of <88% and SpO2 of 92%–96%) was nearly three times as common in Black than White patients. |
| Valbuena et al (2021) | 372 (65 Asian, 51 Black, 70 Hispanic, 186 White) | Intensive Care | Race | Incidence of occult hypoxaemia: 10.2% (White), 21.5% (Black), 8.6% (Hispanic), and 9.2% (Asian) | Hidden hypoxemia (blood gas arterial oxygen saturation of <88% and SpO2 of 92%–96%) was more common in Black patients than Asian, Hispanic or White patients. |
| Vesoulis et al (2021) | 294 (124 Black, 170 White) | Neonatal Intensive Care | Race | Bias (White): +0.72% a | SpO2 was less accurate (overestimation) in Black than White patients. Hidden hypoxemia (arterial oxygen saturation of <85% and SpO2 ≥90%) was more common (although not significantly so) in Black than White patients. |
| Bias (Black): +1.73% a | |||||
| Wiles et al (2021) | 194 (34 Asian, 19 Black, 6 Other, 135 White) | Critical Care | Race | Bias (White): + 0.28% b | SpO2 was less accurate (overestimation) in Black, than Asian, than White patients (although no statistical test was used). Correlation between SpO2 and arterial oxygen saturation was lower in Black patients than Asian or White patients. |
| LOA (White): −1.79%–2.35% | |||||
| Bias (Asian): −0.33% b | |||||
| LOA (Asian): −2.47%–1.80% | |||||
| Bias (Black): −0.75% b | |||||
| LOA (Black): −3.47–1.97 | |||||
| Okunlola et al (2022) | 491 (108 Dark, 383 Light) | Hypoxia Laboratory | Unclear | Not available | SpO2 was less accurate (overestimation) in Dark than Light to Medium skin pigmentations. |
| Wong et al (2021) | 87,971 (1,919 Asian, 26,032 Black, 2,397 Hispanic, 57,632 White) | Intensive Care and other Hospital Wards | Race | Proportion of patients with hidden hypoxaemia: | The incidence of hidden hypoxemia (arterial oxygen saturation of <88% and SpO2 ≥88%) was greatest in Black, then Hispanic, Asian and finally White patients. It was associated with greater organ dysfunction 24 h later, and higher in-hospital mortality. |
| White: 4.9% | |||||
| Asian: 4.9% | |||||
| Hispanic: 6.0% | |||||
| Black: 6.9% | |||||
| Bangash et al (2022) | 16,818 (81.2% White, 11.7% Asian, 4.0% Black, 3.2% Other ethnicities) | Hospital | Ethnicity | Relative to White patients: | Pulse oximetry tends to overestimate O2 saturation, and this is more pronounced in patients of Black ethnicity. These differences resulted in 6.1% versus 8.7% of White versus Black patients classified as normoxic on SpO2 who were hypoxic on the gold standard SaO2 reading. |
| Bias (Asian): 0.5pp greater a | |||||
| Bias (Black): 0.8pp greater a | |||||
| Bias (Other): 0.3pp greater a | |||||
| Shi et al (2022) | 6,505 (4,897 adults and 1,608 children) | 27 out of 32 in hospital, none at home | 15 studies measured skin pigmentation and 22 referred only to ethnicity | Bias ± SD (Light): −0.35 ± 1.49% a | Pulse oximetry may overestimate oxygen saturation in people with dark skin and people whose ethnicity is reported as Black/African American, compared with SaO2, although the overestimation may be quite small in hospital settings. The clinical importance of any overestimation will depend on the particular clinical circumstance. |
| Distribution of racial groups not stated | |||||
| LOA (Light): −1.87–4.09 | |||||
| Bias ± SD (Intermediate): −0.58 ± 1.47% a | |||||
| LOA (Intermediate): −3.46–2.30 | |||||
| Bias ± SD (Dark): +1.11 ± 1.52% a | |||||
| LOA (Dark): −3.27–2.58 | |||||
| Baker and Wilson (2022) | 75 (39 Black, 36 White) | Hypoxia Laboratory | Unclear | Bias ± SD (White): −0.05 ± 1.35 a | There was no clinically significant difference in the accuracy or bias between Black and White subjects monitored with Masimo SET pulse oximetry. |
| Arms (White): 1.35 | |||||
| Bias ± SD (Black): −0.2 ± 1.40 a | |||||
| Arms (Black): 1.42 | |||||
| Crooks et al (2022) | 2997 Distribution of ethnic groups not stated | Hospital | Ethnicity | Bias (White): +3.2% a | Pulse oximetry overestimated arterial oxygen saturations compared to blood gas measurement across all ethnicity groups when SpO2 measurements were below 90%, and underestimated these when SpO2 measurements were above 95%. However, individuals with Black, Asian or mixed ethnicity had a higher reading for oxygen saturation as measured by pulse oximetry than blood gas compared to individuals with a White ethnicity. |
| Bias (Asian): +5.1% a | |||||
| Bias (Black): +5.4% a | |||||
| Bias (Mixed): +6.9% a | |||||
| Wiles et al (2022) | 178 (126 White, 30 South Asian, 13 Black, and 9 other ethnic origin) | Intensive Care | Ethnicity | Bias (White): −0.25% b | Bias was greater in patients of non-White ethnic origin. The study also found that pulse oximetry is less accurate in patients diagnosed with COVID-19 and receiving mechanical ventilation. |
| LOA (White): −4.75%–4.23% | |||||
| Bias (South Asian): −0.96% b | |||||
| LOA (South Asian): | |||||
| −5.62%–3.71% | |||||
| Bias (Black): −1.72% b | |||||
| LOA (Black): −6.8%–3.36% | |||||
| Bias (Other): −1.21% b | |||||
| LOA (Other): −5.48%–3.05% | |||||
| Henry et al (2022) | 26,603 (24,493 White, 1,263 Black, 574 Asian, 273 American Indian) | Intensive Care | Self-identification | Incidence of occult hypoxaemia: White (3.6%), Black (6.2%), Asian and American Indian (6.6%) | Occult hypoxemia is more common in Black patients compared with White patients and is associated with increased mortality, suggesting potentially important outcome implications for undetected hypoxemia. It is imperative to validate pulse oximetry with expanded racial inclusion. |
| Burnett et al (2022) | 151,070 (16,011 Black, 21,223 Hispanic, 70,722 White, 8,121 Asian, 34,993 other) | Unclear | Self-identification | Bias ± SD (White): −0.20 ± 6.3% a | Self-reported Black and Hispanic race/ethnicity are associated with a greater prevalence of intraoperative occult hypoxemia in the SpO2 range of 92% to 100% when compared with self-reported White race/ethnicity. |
| Bias ± SD (Hispanic): +0.5 ± 7.9% a | |||||
| Bias ± SD (Asian): +0.2 ± 6.5% a | |||||
| Bias ± SD (Other): +0.1 ± 5.9% a | |||||
| Bias ± SD (Black): + 0.6 ± 9.1% a | |||||
| Fawzy et al (2022) | 1216 (63 Asian, 478 Black, 215 Hispanic, and 460 White) | Referral centres and community hospitals | Self-identification | Bias (Relative to White Patients) b | Pulse oximetry overestimated arterial oxygen saturation among Asian, Black, and Hispanic patients (ethnic minority groups) compared with White patients with COVID-19. This contributes to unrecognised or delayed recognition of eligibility to receive COVID-19 therapies. |
| Asian: −1.73% | |||||
| Hispanic: −1.13% | |||||
| Black: −1.23% |
LOA: Limits of agreement, SD: Standard deviation, pp: Percentage points
a
Bias= SpO2 - SaO2
b
Bias= SpO2 - SaO2.