Development and Prospective Validation of the Levothyroxine Dose Prediction Model in Primary Hypothyroidism

Abstract

AbstractMultiple visits are needed to achieve euthyroidism on levothyroxine in newly detected primary hypothyroidism. We aimed to develop a levothyroxine dose estimation algorithm for primary hypothyroidism. Adults with newly diagnosed hypothyroidism were enrolled prospectively, first in the training cohort, followed by the validation cohort separated by time and person. We developed a predictive algorithm from Training Cohort and validated the model in Validation Cohort. Training Cohort: In this cohort, 358 subjects (259 women and 99 men) were enrolled. The median duration needed to achieve euthyroidism was 4±0.5 months. The mean levothyroxine daily dose was 60.5±34.1 μg. Data of euthyroid subjects within 6 months of treatment initiation and age range 18–65 years were used for algorithm development. In the multivariable linear regression algorithm, pretreatment serum thyrotropin level, and sex formed the best-fit predictive model (adjusted R2 0.73, p-value <0.001). Validation Cohort: Eighty-four subjects (61 women and 23 men) were enrolled and started on an estimated levothyroxine dose derived from the developed prediction model. On the first follow-up on treatment, 34/50 participants achieved euthyroidism (68%) at 1.5 months. In conclusion, the proposed prediction model for levothyroxine dose estimation effectively achieves early euthyroidism in two-third subjects in the age range of 18–65 years.