Clinical question: What is the prevalence of, and what are the risk factors for, occult hypoxemia in ICU patients, and does that impact patient outcomes?
Background: The finding of occult hypoxemia, i.e., the presence of hypoxemia as measured by arterial blood gas compared to the readings of pulse oximeters, is more common in patients identifying as Black. There is limited evidence addressing whether occult hypoxemia is associated with differences in patient outcomes.
Study design: Retrospective cohort analysis of registry data (Blood-gas and Oximetry Linked Dataset [BOLD])
Synopsis: Data were drawn from the BOLD database, a composite of three publicly available registries of de-identified patient information. However, the data used were limited to only the eICU-CRD registry, which is drawn from the Philips Healthcare eICU Telehealth Program. This registry comprises patient information from 208 hospitals and more than 200,000 admissions (from 2014 to 2015). BOLD reports SaO2-SpO2 pairs with a separation of less than five minutes. The rationale for limiting the review to this cohort was to limit technological variation as a cause of error. The authors analyzed 36,280 patient readings. The authors defined occult hypoxemia as follows: 1) PaO2 below 60 or SaO2 below 89%, 2) SpO2 over 88%, 3) discrepancy between SaO2 and SpO2 over 2.99%. Of that group, 8,559 patients were found to have SaO2-SpO2 discrepancies, and 1,719 met the above criteria for occult hypoxemia (4.7% incidence). Results showed that Black patients were more likely to have an SaO2-SpO2 discrepancy (adjusted odds ratio [aOR], 1.35; CI, 1.25-1.47) and more likely to have occult hypoxemia (aOR, 1.22; CI, 1.04-1.44). The authors also reported a novel finding: a higher Charleston Comorbidity Index score in patients with occult hypoxemia (3.9 versus 4.2; aOR, 1.05: CI, 1.03-1.08). Importantly, the unadjusted risk of mortality with an SaO2-SpO2 discrepancy was 18.7%, while the unadjusted risk of mortality for patients with occult hypoxemia was 26.4%. The adjusted multivariate regression analysis gave an adjusted odds ratio for mortality with occult hypoxemia of 1.73 (CI, 1.53-1.94). Though other factors had an elevated adjusted odds ratio, none were as significant as the presence of occult hypoxemia.
Bottom line: Occult hypoxemia is more prevalent in patients identifying as Black and, to a lesser extent, in patients with multiple comorbidities. Occult hypoxemia is independently associated with an increased risk of mortality.
Citation: Saidy S, et al. Pulse oximetry discrepancies and occult hypoxemia in ICU patients: predictors and clinical outcomes. J Intensive Care Med. 2025;40(12):1269-1278. doi:10.1177/08850666251351594.
Dr. Lampman
Dr. Lampman is a hospitalist and associate professor of internal medicine at the University of North Carolina in Chapel Hill, N.C.