Evaluation of thyroid function in dogs by hormone analysis: effects of data on biological variation

The purpose of the present study was to investigate commercially available ELISA methods designed for the determination of total and unbound thyroxine (TT(4) and FT(4)) and total triiodothyronine (TT(3)) in human serum for their usefulness in evaluating thyroid function in dogs when data describing the biological variation were included in the characterization of the assays. The TT(3) analysis was evaluated with intraassay coefficients of variation (CV%) ranging between 12% and 20%, and interassay CV% ranging between 5 and 17% at naturally occurring TT(3) concentrations. At concentrations around the limit of detection (0.27 nmol/l) CV% was considerably higher (99%). The analysis exhibited a satisfying accuracy since the recovery of added TT(3) was not different from unity and since parallelism between the dose-response curve and plasma dilutions could be verified. Determination of TT(4), TT(3) and FT(4) in eight normal dogs during 4 weeks resulted in a significant variation between dogs and between weeks in the individual animals (p < 0.01 in all cases). From the inter- and intraindividual CV%, quality goals for the maximally allowed analytical variation could be computed to be 8.4, 10.0, and 10.1% for individual testing of animals, and 12.0, 12.9, and 15.8% for screening for diseased animals in healthy populations for TT(4), TT(3) and FT(4), respectively. A comparison between quality goals derived from the inter- and intraindividual CV% and the measured analytical CV% (4.0, 17.3, and 6.7%, respectively) evidenced that TT(4) and FT(4) analyses fulfilled the requirements for analytical precision, whereas the TT(3) analysis could not be accepted as an effective tool for the evaluation of thyroid function in dogs due to too high analytical variation.