Effects of anemia on pulse oximetry and continuous mixed venous hemoglobin saturation monitoring in dogs

The accuracy of pulse oximetry (for pulse hemoglobin oxygen saturation SpO2]) and mixed venous oximetry (for mixed venous hemoglobin oxygen saturation SvO2]) was assessed during progressive normovolemic anemia in dogs. Splenectomized mongrel dogs under general anesthesia were monitored with a three-wavelength pulmonary artery oximeter catheter (10 dogs) and a pulse oximeter (11 dogs). Data were collected while fractional inspired oxygen concentration (FIO2) was varied from 1.00 to 0.05 in seven steps. The dogs then underwent isovolemic hemodilution, and the FIO2 was again varied. This sequence continued until data no longer could be obtained. The accuracy of each device was assessed by determining the bias (the average difference between the continuous monitor oximeter and the bench oximeter) and the precision (the standard deviation of the difference). For the three-wavelength Oximetrix catheter (for hemoglobin oxygen saturation denoted here SoxO2), the overall bias (SoxO2 - SvO2) and precision were -0.7 +/- 8.6% for the 193 data points. The accuracy as assessed by bias and precision for SoxO2 was similar for hematocrits of 40-15%. (Bias +/- precision was 2.1 +/- 5.7% for hematocrits greater than 40%, and -1.1 +/- 7.5% for hematocrits of 15% to 19%). At hematocrits between 10 and 14%, the precision worsened to 12%, and for hematocrits less than 10% the bias +/- precision was -11.5 +/- 11.8%. The overall SpO2 accuracy was 0.2 +/- 7.6% for 178 points. The pulse oximeter's accuracy was similar, down to hematocrits of 10%. Below 10%, the bias and precision worsened to -5.4 +/- 18.8%.(ABSTRACT TRUNCATED AT 250 WORDS)