Read on Twitter
1/ Our research letter on racial bias in pulse oximetry measurement, out today in NEJM https://www.nejm.org/doi/full/10.1056/NEJMc2029240
2/ The finding that pulse oximetry is biased due to skin pigmentation isn’t new: https://pubmed.ncbi.nlm.nih.gov/18048893/
https://pubmed.ncbi.nlm.nih.gov/15791098/
As a pulm/crit trained physician, I didn’t know this literature, and thats a problem
https://pubmed.ncbi.nlm.nih.gov/15791098/
As a pulm/crit trained physician, I didn’t know this literature, and thats a problem
3/ The mechanism is likely that skin pigmentation (melanin) interferes with pulse oximeter light transmission
4/ Compared to prior studies, our population was 10x larger, and focused on a clinical range of saturations (88% to 96%) where small differences in accuracy might make a big difference in treatment
5/ We first looked at Michigan Medicine data from 2020, and confirmed the finding in the eICU database from 2014-2015 https://eicu-crd.mit.edu/
6/ We also 1) adjusted for cardiac SOFA score (low cardiac perfusion can interfere with accuracy), 2) excluded patients with diabetes, 3) excluded patients with higher carboxy hemoglobin…the finding continued to hold
7/ @iwashyna continued to run the analysis on any data he got his hand on … the finding continued to hold
8/ So what to do ? 1) Recognize that pulse ox is imperfect, 2) Recognize that it is racially biased, 3) Don’t always feel reassured when sat is low normal, 4) Consider ABG when it really matters, 5) Keep fighting to remove structural racism from health care
9/ FDA requires new devices to be analyzed for safety and effectiveness in important demographic groups, which often means ensuring its tested in diverse populations. But device effectiveness may not always be compared across groups. https://pubmed.ncbi.nlm.nih.gov/30203600/
