Interpretation of TSH and T4 for diagnosing minor alterations in thyroid function: a comparative analysis of two separate longitudinal cohorts

Abstract

Abstract Background Minor alterations in thyroid function are frequent, and interpretation of thyroid function tests in the individual patient can be challenging. Furthermore, the choice of thyroid function test is debatable. To inform the debate, we performed a comparative evaluation of the variation in thyrotropin (TSH) and thyroxine (T4) in two different cohorts to illustrate the precision of TSH and T4 in the diagnosis and monitoring of thyroid dysfunction. Methods A comparative analysis of two separate longitudinal studies previously surveyed with monthly blood sampling for one year among 35 subjects. Participants were included based on T4 within the reference range and TSH either within (euthyroid; n = 15) or above (subclinical hypothyroidism; n = 20) the laboratory reference range on two independent blood samplings before inclusion. Exclusion criteria were known thyroid disease or use of thyroid interfering medication. TSH and T4 in individual samples were measured in a single batch to prevent between-batch variation. The distributions TSH and T4 were compared among euthyroid and subclinical hypothyroid individuals, and bootstrap estimates were used to calculate area under the curve (AUC). Results Collection of twelve, monthly blood samples in the 35 participants provided 420 samples, and data completeness was 100%. The mean TSH was 1.27/7.19 mIU/L and the mean total T4 was 106/85 nmol/L in euthyroid/subclinical hypothyroid participants. The subclinical hypothyroidism state deviated from the euthyroid by 20% for total T4 and by 466% for TSH. The overlap between the euthyroid and subclinical hypothyroid groups was 92.6% (389/420) for total T4 and 9.0% (38/420) of test results for TSH. The estimated AUC was 0.999 (95%-CI: 0.995; 1.00) for TSH and 0.853 (0.736; 0.935) for total T4. There was no confidence interval overlap between participant groups for TSH while there was a considerable overlap for total T4 (p < 0.001). Conclusion The distributions of thyroid function tests illustrated how TSH outperforms T4 for detecting delicate differences in thyroid function in an individual. Thus, TSH was markedly better than T4 to discriminate between the subtle differences in thyroid function corroborating that TSH is the more sensitive and accurate index of thyroid function status in the individual patient.