RESUMEN
Clinical decisions are influenced by hematocrit values. Centrifugation (reference standard), conductivity, optical and impedance methods are often used interchangeably to measure hematocrit. The effects of albumin, which are known to affect conductivity methods, have not been evaluated for limits of agreement (LOA) between hematocrit assays in small animals. Canine venous blood was collected from 74 clinical cases and measured by centrifugation (n = 72), conductivity (n = 73), impedance (n = 24) and optical (n = 50) methods. Bland-Altman analysis determined bias (± SD) and 95% LOA between methods. There was a statistically significant difference between centrifugation hematocrit values and values obtained via conductivity (p < 0.0001), optical (p < 0.0001), and impedance (p = 0.0082) methods. The conductivity method underestimated hematocrit by 2.1 ± 2.9% (95% LOA -3.54 to 7.88), the optical method by 3.1 ± 3.6% (95% LOA -4.0 to 10.2), and the impedance method by 2.3 ± 3.7% (95% LOA -5 to 9.6) when compared to centrifuged hematocrit values. The hematocrit difference between conductivity and centrifugation methods was statistically different for low (4%, 0-5%), within reference limits (3%, -5 to 8%), and high (2%, -2 to 5%) albumin values, respectively (p = 0.02), with post-hoc analysis demonstrating that the difference occurred between the low and high albumin groups. This study confirms that albumin values outside reference limits can affect the conductivity method and that hematocrit values obtained via conductivity, optical and impedance methods underestimate values obtained via centrifugation. Therefore, the hematocrit methods cannot be used interchangeably. The wide limits of agreement also demonstrates that care must be taken when making clinical decisions with different hematocrit methodologies.