To investigate the accuracy and precision of the portable, battery-powered StatPal II and the laboratory-based NOVA StatProfile 1 blood gas and pH analyzers for use in analysis of equine blood. Patient sample comparison and whole blood tonometry. Patient sample comparison: 125 arterial or venous blood samples from 49 healthy, awake, or anesthetized horses or ponies. Tonometry: venous blood samples from 11 healthy Thoroughbred horses. Arterial and venous blood taken from awake and anesthetized equine patients was placed in an ice-water bath, then analyzed within 30 minutes of collection. Bias and limits of agreement between analyzers in measurement of pH, P(CO2), and P(O2) were calculated according to the method of Bland and Altman. Tonometry, using analyzed gases with a range of P(O2) of 28 to 286 mm Hg and P(CO2) of 21 to 85 mm Hg, was performed on equine whole blood or blood with abnormally high (55%) or low (20%) hematocrit. Samples were introduced directly from the tonometer into the analyzers. Inaccuracy (% of target value) and imprecision (coefficient of variation) were determined for each instrument. In addition, results of analysis of blood samples introduced into the analyzers at 36 degrees C, 0 to 3 degrees C, and 22 degrees C were compared. In the patient sample comparisons, bias between analyzers (StatPal-NOVA) for measurement of P(O2) less than 60 mm Hg was -0.33+/-6.2 mm Hg (x +/-2 SD) and for P(O2) between 60 and 110 mm Hg bias was -1.48+/-9.2 mm Hg. Bias was 46.5+/-67 mm Hg (significantly different from bias at the lower P(O2) levels) for measurement of P(O2) values of 111 to 505 mm Hg, and at P(O2) values greater than 110 mm Hg, bias increased with increasing P(O2). During the course of the study, a significant shift in bias between instruments occurred for P(CO2) and pH measurement, coincident with a change of P(CO2) and pH electrodes in the NOVA and use of a new lot of StatPal sensors. Bias (StatPal-NOVA) for P(CO2) before and after the electrode change was -3.74+/-4.2 and -0.88+/-6.8 mm Hg, and bias for pH before and after the electrode change was 0.026+/-0.034 and -0.024+/-0.038. The change in bias was significant (P<.05). In the whole blood tonometry trials, mean recovered values of P(CO2) and P(O2) from blood with a normal hematocrit ranged from 94% to 109% of target values for StatPal and from 98% to 107% for NOVA. Imprecision ranged from 3.3% to 5.3% for StatPal and from 2.2% to 4.3% for NOVA. With extremes of hematocrit (55% and 20%), StatPal's mean recovered P(CO2) values were 115% and 112% of the target value of 21 mm Hg, whereas NOVA's recovered P(CO2) values were similar to those recovered from samples with normal hematocrit. Introduction of cold blood samples (0 to 3 degrees C) into StatPal resulted in P(CO2) readings that were approximately 2 mm Hg lower than those of 22 degrees C and 36 degrees C samples (P<.05). No other effects of sample temperature were found for either instrument. StatPal and NOVA are of similar accuracy and demonstrate acceptable precision for measurement of P(CO2) and P(O2) in equine blood with values in the normal arterial and venous range. Mean recovered values during tonometry differed by as much as 10% between instruments, indicating that they should not be used interchangeably for a single patient or for a group of subjects in a research setting. The StatPal is a portable blood gas analyzer of acceptable accuracy and precision, for clinical or investigational work in horses.
Read full abstract