A prospective study was performed to encourage a rational thyro-id-stimulating hormone (TSH) test request behavior of physicians, using prediction instruments. The latter give direct feedback about request adequacy on the basis of pretest probabilities of hyperthyroidism using patients' signs and symptoms. For instrument design, stepwise logistic regression was used on diagnostic data acquired through questionnaires, answered by 80 physicians and 668 patients for whom physicians requested a TSH test. Instruments were designed for clinical and subclinical hyper-thyroidism and for clinical hyperthyroidism alone. Use of the instrument for clinical or subclinical hyperthyroidism on the selected group, at a 5% probability threshold, can result in a 37% reduction of unnecessary TSH test requests. With the instrument for clinical hyperthyroidism at a 5% probability threshold, the number of unnecessary test requests can be reduced by 57%. Therefore, it can be concluded that the instruments can determine TSH test request adequacy and encourage a rational TSH test request behavior of physicians at low pretest probability thresholds.
Print ISSN: 0026-1270
Volume: 38, 03/1999
Pages: 31 - 36