Comparison of venous hemoglobin saturation measurements obtained by in vivo oximetry and calculated from blood gas analysis in critically ill dogs

Abstract

To compare in vivo central venous hemoglobin saturation measurement (ScvO2 ) using a fiber optic catheter with saturation calculated from blood gas analysis in critically ill dogs. Prospective observational study. University veterinary teaching hospital intensive care unit. A convenience sample of 20 dogs with severe illness. Dogs were instrumented with either a central venous catheter with an integrated fiber optic cable or a conventional catheter with a fiber optic probe inserted through its distal port. Baseline saturation was measured with the fiber optic system (FSO2 ), then monitored continuously. Central venous blood was collected for analysis and FSO2 was recorded by the principal investigator (PI) or nursing staff participating in data collection (staff) at baseline and at 1, 2, 3, and 6 hours. Hemoglobin oxygen saturation (SO2 ) values calculated using human Bohr coefficients were taken directly from the analyzer (GPSO2 ), and were also calculated using temperature-correction and canine Bohr coefficients (RSO2 ). Ninety-seven paired measurements from 20 dogs were analyzed. FSO2 obtained by the PI (n = 41) had better agreement with both GPSO2 (concordance correlation coefficient ρc = 0.926 vs 0.5562) and RSO2 (ρc = 0.75 for PI vs 0.54) than did staff (n = 56). RSO2 values were always smaller than GPSO2 . FSO2 - GPSO2 differences were smaller when measurements were collected by the PI versus Staff (mean difference 0.21 vs -6.6, respectively, P<0.02). Thirty-six of 41 FSO2 values obtained by PI were within 5% of GPSO2 . Concordance between FSO2 and either calculation method was low, but was better when performed by PI. The larger difference between methods when using RSO2 suggests a positive bias by FSO2 . Difficulty obtaining stable measurements may have contributed to the poor concordance between methods within Staff.