This is a prospective observational study designed to quantify and understand errors in pulse oximetry in hospitalized patients in relation to their skin pigmentation. It is driven by three recent retrospective studies showing missed diagnosis of hypoxemia in patients across a spectrum of skin pigmentation, defined as blood SaO2 <90% when their pulse oximeter reads 92% or greater.

This prospective study will overcome the limitations of earlier retrospective studies using three important study protocols:

1. precise time-matching paired blood gas samples to stable periods of pulse oximeter readings
2. objective quantification of skin pigmentation rather than using documented or self-reported race
3. collection of high quality pulse oximeter tracing data, and identifying potential mediators of discordance of noninvasive and invasive measurements such as low perfusion states.

The primary aim is to quantify the bias in SpO2 measurements across skin pigmentation groups in a real-world setting against the gold standard of invasive measurements of arterial blood oxygen saturation measurements across categories of skin pigmentation. These data are essential to addressing health equity and patient safety for people of all skin pigmentations.

Inclusion Criteria:

• 18 years of age or older
• hospitalized and at high risk of hypoxemia, defined as saturations < 90%,
• patients with arterial line in place for clinical purposes

Exclusion Criteria:

• Profound anemia, defined as hemoglobin < 8 g/dl will preclude blood draws for research purposes
• Known pregnancy
• In custody of law enforcement officials

Hospitalized patients at risk for hypoxemia with arterial lines placed for clinical purposes.

De-identified data without private health information or individual identifiers will be made available via the Open Oximetry Project. De-identified data without sensitive information or identifiers will be shared as outlined in the IRB protocol with the study sponsor, the Unites States Food and Drug Administration.

Upon study completion, approximately October 2023 and in accordance with any peer-reviewed journal requirements.

Qualified researchers with appropriate training in Human Subjects Research protocols and without conflicts of interest in data analysis or publication.