Agreement between hematocrit values determined by the Cobas b121 blood gas analyzer and the microhematocrit method in dogs, cats, and horses.

Abstract

Packed cell volume (PCV) is important for assessing a patient's health status. Some blood gas analyzers measure hematocrit, and the agreement with PCV varies among different analyzers. We aimed to determine the agreement between PCV measured by microcentrifugation and hematocrit measured by the Cobas b121 blood gas analyzer in dogs, cats, and horses. Whole blood samples for PCV and blood gas analysis were collected in lithium-heparin syringes and analyzed within 10 min of collection. Agreement and association between the PCV and Cobas b121 generated hematocrit were assessed by the Bland-Altman method, Pearson's correlation, Deming regression analysis, and paired t tests. A total allowable error of 10% was used for the analysis. This study included 45 dogs, 45 cats, and 33 horses. The respective mean ± SD (minimum-maximum) of PCVs and hematocrits were: dogs, 34.9 ± 9.9% (9.0-55.0) and 32.5 ± 8.8% (10.4-50.6); cats, 29.0 ± 9.6% (11.0-51.0) and 26.9 ± 9.3% (10.2-50.9); horses, 34.2 ± 6.5% (24.0-47.0) and 34.1 ± 6.0% (22.5-46.1). There were no significant differences between the methods. The bias ± SD was: dogs, -2.4 ± 2.6%; cats, -2.2 ± 2.3%; horses, -0.1 ± 2.4%. Pearson's correlation coefficients were > 0.90 for all species (P < 0.0001). In 60%, 49%, and 85% of the samples for dogs, cats, and horses, respectively, the percentage differences between the methods were within 10%. The Cobas b121 blood gas analyzer provided accurate estimates of PCVs in horses. However, in dogs and cats, there was a large frequency of unacceptable differences between the methods.