Pulse oximetry fails to accurately detect low levels of arterial hemoglobin oxygen saturation in dogs.

The accuracy of two commercially available pulse oximeters (the Ohmeda Biox 3700, software version “J,” and the Nellcor N-100) in detecting low levels of arterial hemoglobin oxygen saturation (SaO2) was evaluated in 10 dogs in which hypoxia was induced by stopping the fresh gas flow into the anesthesia machine circle system. Measurements made in vivo with the pulse oximeters, with detectors placed on the tongue, were compared with measurements made in vitro using an IL 282 CO-Oximeter as SaO2 decreased toward zero. Measurements from the two oximeters correlated poorly over the range from 0 to 100% SaO2 (r = 0.69). In this range, the correlation between Nellcor N-100 measurements and those of the CO-Oximeter had an r of 0.82, a regression line slope of 0.82, and a y intercept of 14.8; the correlation between the Ohmeda Biox 3700 and the CO-Oximeter had an r of 0.83, a regression line slope of 0.66, and a y intercept of 32.7. The correlation with the CO-Oximeter was similar for both the Ohmeda and the Nellcor pulse oximeters at an SaO2 of 80% or more. However, when SaO2 was less than 80%, measurements by pulse oximetry correlated less well with CO-Oximeter measurements (r = 0.62, slope = 0.64, and y intercept = 21.0 for Nellcor; r = 0.71, slope = 0.67, and y intercept = 32.4 for Ohmeda). When SaO2 was less than 60%, both oximeters inaccurately indicated the co-oximetry values (r = 0.36 and y intercept = 26.1 for the Nellcor; r = 0.48 and y intercept = 33.2 for the Ohmeda). In this animal model, with pulse oximeter measurements obtained from the tongue and with rapidly decreasing SaO2, measurements of SaO2 by pulse oximetry become inaccurate in comparison with co-oximetry measurements at low levels of SaO2.