Evaluating pulse oximeters in ICU according to the new FDA guidances: skin tone and LED spectra assessment

Objective.To evaluate whether pulse oximeters cleared under older U.S. food and drug administration (FDA) standards remain accurate in intensive care unit (ICU) patients under newer draft guidance, to assess their performance against different regulatory metrics, and to investigate whether skin tone or LED spectral characteristics could be associated with measurement bias.Approach.In this prospective ICU study, three regulatory-cleared pulse oximeters were compared against arterial blood gas (ABG) analysis. Skin pigmentation was characterized using the individual typology angle, and differential bias was estimated following FDA recommendations. Red and infrared LED emissions were measured using a fiber-optic spectrometer, and Gaussian models were used to extract peak wavelengths.Main results.Two low-cost devices demonstrated large mean bias, wide limits of agreement, and poor correlation with ABG, failing both 2013 FDA guidance (accuracy root mean square, ARMS ⩽ 3%) and updated FDA accuracy requirements. The clinical bedside monitor showed lower bias and minimal pigmentation-related differential error, though still above the proposed ARMS threshold. All devices remained inside the 2017 ISO standard (ARMS ⩽ 4%). Spectral analysis revealed that the most accurate device used a longer infrared peak and a slightly shorter red peak, a combination that could influence separation at key hemoglobin absorption wavelengths and potentially reduce melanin-related interference.Significance.These findings indicate that some oximeters cleared under older standards may underperform in ICU patients when assessed against newer accuracy proposals, underscoring the importance of reevaluating device performance as guidance evolves.