Maximizing the Laboratory Setting for Testing Devices and Understanding Statistical Output in Pulse Oximetry

Abstract

Maximizing the laboratory setting for testing baseline pulse oximetry accuracy in an arterial desaturation study requires a study design that considers management of several aspects in the physiology of the test subject, special attention to the device under test, and great care in the preanalytical (sample handling) and analytical (Co-oximeter) phases. Statistics used to describe the resulting SpO2 performance include Precision (size of the data cloud), Bias (offset of the data cloud), and A(rms) (accuracy root mean square), which combines the size and offset of the data cloud in one number. The A(rms) is the primary statistic required by regulatory organizations to describe general performance over the entire saturation range. It does not describe any one point, but is a compilation of all points over the range tested. Most pulse oximeters in use today specify an A(rms) of 2%. To meet this specification, two-thirds of the readings will be within 2% of the Co-oximeter reference; however, some individual readings can be as inaccurate as 6% or more. The A(rms) statistic does not have the capacity to represent all pulse oximeter behavior. Saturation pop-ups, drop-downs, frozen readings, and periods of no reading are not portrayed by the A(rms). The next steps in the advancement of regulatory validation testing would be to develop standards that include an expanded analysis of pulse oximeter performance by assessment of pop-ups, dropouts, frozen readings, and periods of no reading through assessment of sensitivity/specificity and possibly a "Performance Index" similar to the approach taken by Barker.