Analytical and clinical agreement between the Siemens Immulite 2000 veterinary cortisol and standard cortisol assays in canine serum

Background In 2020, a permanent change in the antibody used in the Siemens Immulite 2000 cortisol assay resulted in a negative bias of approximately 23%. In response, the manufacturer introduced a veterinary-specific cortisol assay. Hypothesis/Objectives To compare the analytical and diagnostic agreement between the Siemens Immulite 2000 standard cortisol (SC) and veterinary cortisol (VC) assays. Animals Residual canine serum samples submitted to a veterinary diagnostic laboratory. Methods Cortisol concentrations were measured using both SC and VC kits on the Siemens Immulite 2000 chemiluminescent immunoassay analyzer. Results A total of 105 dogs contributed 154 cortisol measurements. Median (minimum, maximum) cortisol concentrations for VC and SC assays were 4.19 and 4.67 (<1.00, > 50.00) μg/dL, respectively. Further analysis included only 1 measurement per dog. Cortisol concentrations measured by the 2 assays were highly correlated (Spearman’s rho = 0.981, P < .001). Passing–Bablok regression demonstrated no significant constant bias (intercept −0.035; 95% CI, −0.184 to 0.061) or proportional bias (slope 1.036; 95% CI, 0.997-1.098). Bland–Altman analysis showed no significant mean bias between assays (−0.494 μg/dL; 95% CI, −3.524 μg/dL to 4.513 μg/dL), with limits of agreement of −4.353 μg/dL to 5.342 μg/dL. Diagnostic agreement was excellent (Cohen’s κ = 0.961); 3/105 cases were classified differently, with discordant results occurring near diagnostic decision thresholds. Conclusions and clinical importance The SC and VC assays demonstrate strong analytical and diagnostic agreement in canine serum. Use of either assay is unlikely to alter clinical interpretation in most cases, and both assays appear to share similar analytical characteristics.