LAUSR

Abstract Background: An accurate measurement of serum thyroid stimulating hormone (TSH) is crucial for a thyroid disorder diagnosis and management. We compared the values of TSH between three automated immunoassays with aims to provide insights into variations in TSH levels that are important for a clinical decision: 2.50 mIU/L, 4.00 mIU/L and 10.00 mIU/L. Methods: We measured TSH with three different fully automated immunoassays: Abbott (Architect ci8200), Siemens (ADVIA Centaur XP) and Roche (Cobas e411). Serum was collected from 110 patients between August 2018 and January 2019. The results were compared using the Passing–Bablok regression method. Additionally, linear regression coefficients were calculated after logarithmic transformation of data. Results: Although all three regression coefficients were high (r2 > 0.98), the slopes from Passing–Bablok plots for the correlation of Abbott with either Roche or Siemens were merely 0.66 and 0.73, respectively. The slope for the correlation of Roche and Siemens was 1.11. Consecutively, the results obtained by the Roche and Siemens methods were proportionally higher than those obtained by the Abbott method (38% and 52%, respectively) at all measured ranges. Conclusions: Although immunoassays correlated among themselves, they cannot achieve the same values for clinical decisions for hypothyroidism (clinical requirements). Clinicians should be conscious of these limitations. A harmonisation of the methods is needed to meet clinical requirements and to enable appropriate clinical decisions in cases of hypothyroidism. Likewise, we suggest the introduction of borderline and high risk values of TSH for hypothyroidism depending on immunoassays to avoid misdiagnosis.