Bias in feline plasma biochemistry results between three in-house analysers and a commercial laboratory analyser: results should not be directly compared.

Abstract

In-house analysers are commonplace in small animal practices but cannot be calibrated by the operator; therefore, any bias in the generated plasma analyte values cannot be corrected. Guidelines such as grading of renal disease and published reference intervals (RIs) in veterinary textbooks assume plasma biochemistry values generated by different analysers are equivalent. This study evaluated the degree of bias, as well as if bias was constant or proportional, for feline plasma biochemical analytes assessed by three in-house biochemistry analysers compared with a commercial laboratory analyser. Blood samples were collected on 101 occasions from 94 cats and, after centrifugation, plasma was divided into four aliquots. One aliquot was sent to the commercial laboratory and the remaining three were tested using the in-house biochemistry analysers. Results from each analyser were compared with the commercial laboratory results by difference plots and analyses, and by comparing percentages of results within provided RIs. Substantial bias was evident relative to the results of the commercial analyser for at least half of the analytes tested for each machine. In most cases, bias was proportional, meaning that the difference between the methods varied with the concentration of the analyte. The results demonstrate that values obtained from these analysers should not be directly compared and that RIs are not transferable between these analysers. Potential effects of bias on clinical decision-making may be overcome by use of appropriately generated RIs, or reference change values which, for most biochemistry analytes, are more appropriate than subject-based RIs.