Automated and manual serum free thyroxine assays evaluated with equilibrium dialysis.

Two fully automated serum FT4 assays (Ciba Corning: ACS-180; Abbott Diagnostics: IMx) and two manual/semiautomated assays (Wallac: Delfia; Diagnostic Products Corp.: Coat-A-Count) were compared with a FT4 reference method based on equilibrium dialysis (ED) using routine clinical samples (n = 105-150). For the full range of FT4 concentrations the correlation with ED was good (r = 0.932-0.959), except for Coat-A-Count (r = 0.852), a single-step analogue-type method. Analytical inaccuracy of the automated methods at low FT4 concentrations was revealed by the differential plot: the IMx assay overestimated concentrations < 10 pmol l-1 and the ACS-180 assay concentrations < 6 pmol l-1. The between-assay CVs for the automated methods were not better than for the non-automated assays indicating problems of calibration curve stability for the automated assays.