Analytical performance of free and total thyroxine assays.

Excessive bias and imprecision are major analytical problems associated with some assays for free and total thyroxine (T4). Bias in free T4 methods is largely proportional to variations in serum T4 binding. In direct methods, this is attributable to requirements for substantial quantities of protein-bound T4 to replace analytical losses of free T4. In some total T4 methods, bias is inversely proportional to the amount of serum T4 binding and is attributable to the incomplete release of serum protein-bound T4. In others, bias is fixed and attributable to inaccurate calibration. Manufacturers should report the bias in their methods. Calibrations should be standardized. Imprecision varies widely among methods, but is generally less for total T4 methods than for free T4 methods. A consensus on quantitative analytical performance goals for free and total T4 methods would be helpful. Here, performance goals are proposed, based in part on the best achievements of current methods.