Racial bias in pulse oximetry: more statistical detail may help address the issue
Dear Editor,
We were glad to see the BMJ news article(1) highlighting the article by Sjoding and colleagues(2) which identifies a racial bias in pulse oximetry, with hypoxaemia identified less frequently in black people compared to white. The research is valuable, identifying a highly important issue, however further details would aid clinical application of their findings and potential ways it could be addressed.
Useful additional (commonly used) metrics absent from their paper include i) bias (mean difference, SpO2-SaO2); ii) precision (standard deviation of the differences); iii) limits of agreement (bias ± 1.96 x standard deviation); and iv) Accuracy value, used in Food and Drug Administration approvals of pulse oximeters, where accuracy=√(bias^2)+ (precision^2) (also referred to as ARMS). A comparison of these metrics between groups would assist interpretation of the research findings and help identification of potential solutions.
The BMJ news article(1) rightly highlights controversies regarding "race adjustments" in relation to renal function measurement. However, regarding pulse oximetry, some form of adjustment for individual patients’ skin colour maybe appropriate given how pulse oximetry works (transmittance or reflectance of light). Such adjustments could potentially be built into future pulse oximeters, however a bedside adjustment may be the least bad approach to reducing the impact of this vitally important issue in the immediate short term until a concrete solution is found. Additionally, if indeed the issue could be rectified with a correction based upon the individual patient’s skin colour, there is no obvious reason why oximeters should not be calibrated to darker skin and the bedside adjustment made for readings taken from white people.
Sjoding and colleagues(2) also note the relevance of their findings to COVID-19 pandemic. Indeed, we also identified sub-optimal accuracy of pulse-oximetry in people with COVID-19 when looking at people stepping down from ICU(3). In these patients, skin colour (67% of our patients were BAME) is likely to be one of multiple potential factors that could negatively impact the accuracy of pulse oximetry.
In summary, further statistical detail regarding how the accuracy of pulse oximetry is impacted by skin colour would have been useful to address, not just highlight, this very important issue.
References
1. Dyer O. Pulse oximetry may underestimate hypoxaemia in black patients, study finds. Bmj 2020;2020;371:m4926 doi: 10.1136/bmj.m4926
2. Sjoding MW, Dickson RP, Iwashyna TJ, et al. Racial Bias in Pulse Oximetry Measurement. The New England journal of medicine 2020;383(25):2477-78. doi: 10.1056/NEJMc2029240 [published Online First: 2020/12/17]
3. Philip KEJ, Bennett B, Fuller S, et al. Working accuracy of pulse oximetry in COVID-19 patients stepping down from intensive care: a clinical evaluation. BMJ open respiratory research 2020 doi: doi:10.1136/bmjresp-2020-000778
Competing interests:
No competing interests
19 January 2021
Keir EJ Philip
Respiratory SpR
Robert Tidswell (Bloomsbury Institute for Intensive Care Medicine, University College London); Charles McFadyen (Bloomsbury Institute for Intensive Care Medicine, University College London)
NHLI, Imperial College London, Royal Brompton Campus, Fulham Rd, London SW3 6HP
Rapid Response:
Racial bias in pulse oximetry: more statistical detail may help address the issue
Dear Editor,
We were glad to see the BMJ news article(1) highlighting the article by Sjoding and colleagues(2) which identifies a racial bias in pulse oximetry, with hypoxaemia identified less frequently in black people compared to white. The research is valuable, identifying a highly important issue, however further details would aid clinical application of their findings and potential ways it could be addressed.
Useful additional (commonly used) metrics absent from their paper include i) bias (mean difference, SpO2-SaO2); ii) precision (standard deviation of the differences); iii) limits of agreement (bias ± 1.96 x standard deviation); and iv) Accuracy value, used in Food and Drug Administration approvals of pulse oximeters, where accuracy=√(bias^2)+ (precision^2) (also referred to as ARMS). A comparison of these metrics between groups would assist interpretation of the research findings and help identification of potential solutions.
The BMJ news article(1) rightly highlights controversies regarding "race adjustments" in relation to renal function measurement. However, regarding pulse oximetry, some form of adjustment for individual patients’ skin colour maybe appropriate given how pulse oximetry works (transmittance or reflectance of light). Such adjustments could potentially be built into future pulse oximeters, however a bedside adjustment may be the least bad approach to reducing the impact of this vitally important issue in the immediate short term until a concrete solution is found. Additionally, if indeed the issue could be rectified with a correction based upon the individual patient’s skin colour, there is no obvious reason why oximeters should not be calibrated to darker skin and the bedside adjustment made for readings taken from white people.
Sjoding and colleagues(2) also note the relevance of their findings to COVID-19 pandemic. Indeed, we also identified sub-optimal accuracy of pulse-oximetry in people with COVID-19 when looking at people stepping down from ICU(3). In these patients, skin colour (67% of our patients were BAME) is likely to be one of multiple potential factors that could negatively impact the accuracy of pulse oximetry.
In summary, further statistical detail regarding how the accuracy of pulse oximetry is impacted by skin colour would have been useful to address, not just highlight, this very important issue.
References
1. Dyer O. Pulse oximetry may underestimate hypoxaemia in black patients, study finds. Bmj 2020;2020;371:m4926 doi: 10.1136/bmj.m4926
2. Sjoding MW, Dickson RP, Iwashyna TJ, et al. Racial Bias in Pulse Oximetry Measurement. The New England journal of medicine 2020;383(25):2477-78. doi: 10.1056/NEJMc2029240 [published Online First: 2020/12/17]
3. Philip KEJ, Bennett B, Fuller S, et al. Working accuracy of pulse oximetry in COVID-19 patients stepping down from intensive care: a clinical evaluation. BMJ open respiratory research 2020 doi: doi:10.1136/bmjresp-2020-000778
Competing interests: No competing interests