Evaluating Pulse Oximetry Bias in Children With Darker Skin Pigmentation

Prospective Clinical Study to Evaluate the Accuracy of Pulse Oximeters in Children With Darker Skin Pigmentation

In this prospective study, the investigators will enroll 154 children with arterial lines to determine the accuracy of pulse oximeters in children with darker skin pigmentation. Studies in adults suggest pulse oximeters may overestimate the true level of oxygenation in the blood as measured directly by co-oximetry. However, pediatric data are relatively limited. This study, which is funded by the FDA through the Stanford-UCSF (University of California San Francisco) Clinical Excellence in Regulatory Science and Innovation (CERSI) Program, will determine if the error/bias is associated with skin pigmentation and whether the error falls outside FDA standards. The broader purpose of the study is to work toward eliminating health disparities.

As a conventional patient vital sign, pulse oximetry is used widely to determine whether a patient is adequately oxygenated. However, studies suggest that peripheral pulse oximetry (Sp02) systematically overestimates the true arterial oxygen saturation (Sa02) in patients with darker skin pigmentation. This error or bias places patients with darker skin pigmentation at considerable risk by failing to detect important levels of hypoxemia that drive critical treatment decisions like medication usage for Severe Acute Respiratory Syndrome (SARS)-COV-2 (COVID) infection, hospital admission, ICU transfer and intubation. Except for the original description of the problem published by Bickler and colleagues in 2005, most studies (including one pediatric study) have been limited to retrospective studies where the racial/ethnic category is used as a proxy for skin pigmentation, so-called "paired" oximetry measurements may be separated by several minutes where the oxygen saturation can readily change, and important technical factors such a perfusion quality, motion artifact, and light transmittance are not reliably documented. This information is critical because emerging studies suggest that perfusion effects, as measured by the perfusion index (PI), may account for much of the error observed in patients with darker pigmentation. The limitations of prior studies can be readily overcome with a prospective study where skin pigmentation is measured objectively and at the location of the oximeter sensor, SpO2/SaO2 (arterial oxygen saturation) measurements are collected nearly simultaneously while at steady state, and technical factors such as perfusion status, transmittance, and temperature can be accurately recorded. This study will enroll 154 children at Stanford's Lucile Packard Children's Hospital to determine whether the mean bias (error) in FDA-cleared pulse oximeters increases with darker skin pigmentation, and whether this error falls outside of the FDA standard of 3%. The primary study hypothesis is that in children ≤21 years of age managed in a real-world hospital setting, the mean bias (error) in FDA-cleared pulse oximeters increases with increased skin pigmentation as measured by colorimetry and standard pigmentation scales (Fitzpatrick and von Luschan pigmentation scales). The secondary hypothesis is that the mean bias is mitigated when technical factors like perfusion index are accounted for in adjusted models. Data generated from this study will help to identify the impact of darker skin pigmentation on medical device performance in the pediatric population and ultimately help eliminate health disparities.

Inclusion Criteria:

1. Age ≤21 years of age
2. Requires arterial vascular access as part of routine clinical care
3. Patient or legally authorized representative (LAR) willing to provide written informed consent

Exclusion Criteria:

1. Anemia defined as a Hgb <8 g/dL
2. Methemoglobinemia or carbon monoxide poisoning where the SpO2 is known to be inaccurate
3. Skin condition, such as epidermolysis bullosa, where pulse oximetry not expected to be accurate, or application of skin probes is medically inadvisable
4. Non-pulsatile patients (e.g., left ventricular assist device (LVAD), extracorporeal membrane oxygenation (ECMO)
5. Lack of informed consent