Comparison of ionised calcium measured using a portable analyser to a reference method in healthy dogs.

Abstract

To compare the ionised calcium measured on a portable analyser (iSTAT, Abbott) to a reference method. Blood samples from 39 apparently healthy dogs were analysed in duplicate using a portable analyser and a reference method (Radiometer ABL800 FLEX). Bland-Altman plots and Passing-Bablok regression were used to assess constant and proportional bias between the two instruments. A within-assay percentage coefficient of variation and total error (TE) was calculated for both analysers. The reference interval was calculated for the portable analyser using the robust method with confidence interval bootstrapping. The Bland-Altman plot showed a -0.036 mmol/L difference between the two instruments (95% confidence limit -0.08 to 0.01 mmol/L; limits of agreement -0.07 to 0.006 mmol/L). Neither the Bland-Altman plot nor the Passing-Bablock regression (slope -0.03; 95% confidence interval -0.08 to 0.19 and intercept 1; 95% confidence interval 0.83 to 1.2) showed significant proportional bias. The coefficient of variation for the portable analyser was 1.08%, compared to 0.78% for the reference method with a total error of 3.5% for the portable analyser. The estimated population-based reference interval for ionised calcium using the portable analyser is 1.23 to 1.42 mmol/L. For the healthy dogs in this study, compared to the reference method, the portable analyser showed no significant bias for measurement of ionised calcium. Further studies including hyper and hypocalcaemic dogs are required to determine clinical impact of the use of this analyser.