Occult hypoxemia in the ICU: Clinical risk factors, mortality and implications for nursing practice

Pulse oximetry (SpO₂) is the primary tool used by Intensive Care Unit (ICU) nurses to monitor oxygenation. Occult hypoxemia - a reassuring SpO₂ despite low arterial saturation (SaO₂) - is a critical patient-safety concern.

To measure SpO₂-SaO₂ discordance in ICU patients, identify situations in which SpO₂ cannot be trusted, and assess its impact on ICU mortality.

Retrospective single-centre cohort study (September 2019 - January 2026), reported according to STROBE. SpO₂ was charted hourly; each arterial blood gas was paired with the most recent preceding entry. Analysis was restricted to pairs within 10 min, using Bland-Altman analysis for repeated measures. Predictors of occult hypoxemia (SpO₂ ≥ 92% with SaO₂ < 88%) and its association with ICU mortality were examined using generalized estimating equations, with adjustment restricted to early-stay covariates.

The primary set comprised 39,554 paired measurements from 7082 ICU stays in 6348 patients. Mean bias was negligible (+0.1%) but limits of agreement were wide (-7.6% to +7.8%); 18.6% of pairs differed by ≥4%. Among 1219 pairs with SaO₂ < 88%, SpO₂ remained ≥92% in 702 (57.6%). Acidosis was most robustly associated with occult hypoxemia (adjusted OR 2.38, 95% CI 1.97-2.86 for pH < 7.35); invasive mechanical ventilation carried lower risk (OR 0.50, 0.42-0.60). Early occult hypoxemia (within 24 h) was independently associated with ICU mortality (adjusted OR 1.55, 1.17-2.05; 24-h landmark OR 1.45, 1.07-1.97). Point estimates were similar when exposure was ascertained only from pairs obtained within 10 or 5 min (1.44 and 1.49), with confidence intervals including unity because restriction reduced the number of exposed deaths by about 70%.

Among sampled measurements with true arterial hypoxemia, SpO₂ frequently remained ≥92%, particularly in acidotic and patients not receiving invasive mechanical ventilation. Early occult hypoxemia was independently associated with ICU mortality. Because inclusion required a blood gas revealing the discordance, the treating team could act on it; the association reflects a severity signal rather than a causal effect of the discordance itself.

In acidotic ICU patients and those on non-invasive oxygen support, SpO₂ frequently disagrees with concurrent SaO₂ and should not be relied on alone. Arterial blood gas confirmation may be considered when SpO₂ appears inconsistent with the clinical picture.