Abstract: Background Point-of-care, cortisol quantitation can facilitate case management, but independent analytic performance verification is warranted prior to widespread use.
Hypothesis/objectives To evaluate the precision of a point-of-care cortisol assay (TCA; Truforma cortisol assay, Zomedica) for canine serum cortisol quantitation, as well as agreement and bias compared to a validated chemiluminescence assay (CLA; Immulite 2000XPi, Siemens Healthcare Diagnostics Products).
Animals Residual frozen (-20ºC) serum from client-owned dogs (n=92) submitted for cortisol quantitation.
Methods Precision of the TCA was assessed using pooled serum samples with low (0.5-2 µg/dL), medium (5-10 µg/dL), and high (18-30 µg/dL) cortisol concentrations. For method comparison, 20 samples each with low (0.5-6 µg/dL), medium (7-17 µg/dL), high (18-30 µg/dL) recorded cortisol concentrations were used. Samples were analyzed in duplicate by CLA and TCA. Spearman rank correlation, Bland-Altman analysis, Passing-Bablok regression, and a clinical decision error grid were used to compare results.
Results Within-laboratory coefficients of variation for pooled low, medium, and high cortisol concentrations were 15.3%, 12.2%, and 11.3%, respectively. Correlation between TCA and CLA was strong (rho=.92) for the method comparison sample set. Constant bias (y-intercept=-.043, 95%CI -.070 to .069) was not identified but Passing-Bablok analysis revealed proportional bias (slope=1.26, 95%CI 1.11-1.39). Disagreement sufficient to cause misdiagnosis occurred for 6 and 5 samples with low and high cortisol concentrations, respectively.
Conclusions and clinical importance Overall, the TCA and CLA assays had strong agreement. Coefficients of variation were above recommended standards for low concentrations, which could have contributed to clinical disagreements in that sample subset.
