Trivariate evaluation of a thyroid hormone panel in clinical practice compared with multiple univariate evaluation.

The results from 510 consecutive routine determinations of free thyroxine index, free triiodothyronine index and thyrotropin were evaluated using both a trivariate reference region and the combined three univariate reference intervals. The results from 109 patients were discordant when evaluated by both the trivariate and the triple univariate reference regions. In 108 of these subjects the hormone results were found to be abnormal by the triple univariate evaluation method but normal when the trivariate reference region was used. The latter evaluation was in accordance with the clinical findings of the patients, who were euthyroid as evaluated from the 105 medical records we could trace. In one subject, clinically euthyroid, trivariate evaluation misclassified the patient as abnormal in contrast to the classical univariate evaluation. We conclude that the trivariate evaluation method was in better agreement with the clinical diagnosis of the patients and should be used in the routine evaluation of trivariate data in order to diminish the number of false abnormal results.