[Importance of thyroid hormone determination in the diagnosis of hyperthyroidism--significant improvement in medical assessment using decision theory results].

A "schematic" classification of the results of serum thyroid hormone measurements on the basis of mathematical decision theory proves to be superior to the physicians' conventional evaluation. That has been shown by comparison with the evaluations made by three experienced doctors. When using no control regions (but the results of serum T4, FT4 and T3 simultaneously) the physicians' rates of misrecognition (0-2% false positives, 20-27% false negatives) are only a little lower than when classifying schematically on the basis of only one thyroid hormone. The construction of a suitable control region, however, results when T3 alone is used, in only 7% false negative and 0% false positive results, while these rates remain nearly unchanged for the physicians' evaluations. The optimum application of mathematical decision theory is only possible with the knowledge of the statistical distributions of the thyroid hormone data, both for a representative collective of normals and that of patients with proven hyperthyroidism. Thus, quantitative criteria for the effectiveness of each set of thyroid hormones can be constructed and a multivariate analysis can be performed, in our example - when applying suitable control regions - even with vanishing rates of misrecognition.