When Recalibration Reproduces Inequity: The Pulse Oximetry Paradox and Why Device Standards Must Account for Melanin

Pulse oximetry has saved countless lives since its introduction four decades ago. The technology is elegant in its simplicity—two wavelengths of light penetrate tissue, hemoglobin absorption patterns are measured, and oxygen saturation is calculated in real time. In intensive care units, operating theatres, and primary health clinics across Georgia and the world, clinicians rely on this device to make life-or-death decisions about supplemental oxygen, intubation, and patient disposition. Which is precisely why systematic measurement error in pulse oximetry constitutes not merely a technical problem, but a patient safety crisis with profound equity dimensions.

The recent discovery that pulse oximeters at a major medical centre shifted their systematic error—moving from overestimating oxygen saturation in darker-skinned patients to underestimating it following device recalibration—deserves urgent scrutiny. This is not a story of simple device failure. It is a cautionary tale about how well-intentioned engineering corrections can inadvertently entrench rather than eliminate racial disparities in clinical measurement, and how our international standards infrastructure remains dangerously inadequate to prevent such harm.

The Fundamental Problem: Melanin and Light Absorption

The physics underlying pulse oximetry bias is well-established. Melanin in darker skin absorbs light across the wavelength spectrum, particularly in the infrared range that conventional pulse oximeters use. This altered light transmission creates systematic measurement error. For decades, this manifested primarily as overestimation of oxygen saturation in patients with darker skin—a bias that masked hypoxemia and delayed recognition of respiratory compromise. The clinical consequence was measurable: Black and multiethnic patients in high-income settings received less supplemental oxygen, were intubated less frequently, and experienced higher mortality from acute respiratory illness.

But here lies the paradox: when manufacturers recalibrated their algorithms, presumably to correct this known bias, some devices began to underestimate oxygen saturation in the same populations. The direction of error reversed, yet the inequity remained. Patients with darker skin were still receiving systematically inaccurate information about their oxygenation status—only now in the opposite direction, potentially leading to unnecessary interventions or false reassurance depending on the clinical threshold being applied.

Why This Exposes a Standards Failure

Current international standards for pulse oximetry validation—including ISO 80601-2-61 and FDA guidance—typically require manufacturers to validate devices across a limited range of skin pigmentation, often with insufficient representation of individuals with darker skin. Validation cohorts frequently skew toward lighter-skinned populations, particularly in high-income country studies. Even when darker-skinned individuals are included, the statistical power to detect differential bias across racial groups may be inadequate.

More troublingly, recalibration and algorithm refinement often occur after regulatory approval and broad deployment. When a manufacturer adjusts their calibration curve to address recognised bias, there is frequently no requirement to re-validate across the full spectrum of human skin pigmentation before the updated device reaches clinical practice. The device that leaves the factory may perform differently than the one that was originally tested.

The Global Equity Dimension

This problem is not confined to high-income settings where it was first documented. Pulse oximeters manufactured in and exported from wealthy nations are deployed across Africa, South Asia, and other resource-limited regions where they may be among the only available tools for continuous oxygen saturation monitoring. If these devices systematically misrepresent oxygenation in the populations they serve most intensively, the cascading failures in clinical decision-making are profound and preventable.

What Must Change

We require international regulatory harmonisation mandating that pulse oximetry devices be validated across representative cohorts stratified by skin pigmentation—not as a voluntary best practice, but as a non-negotiable condition of approval. Recalibration and algorithm updates must trigger re-validation before clinical release. Manufacturers must publish disaggregated accuracy data by skin tone in the scientific literature, not merely in regulatory filings.

Additionally, clinicians must understand that no pulse oximeter is equally accurate across all populations. Until these devices demonstrate equivalent performance across the human spectrum of melanin content, we must interpret them with appropriate scepticism and maintain lower thresholds for confirmatory blood gas analysis in patients for whom bias is documented.

Patient safety and health equity are not competing principles—they are prerequisites for one another. A measurement device that is systematically inaccurate for any demographic group has failed both.