BACKGROUND: Thyroid disorders affect millions of people around the world, mainly women. Measurement of hormones is critical in their identification, treatment, and monitoring
Therefore, the methods must be as reliable and valid as possible. METHODS: We designed a cross-sectional study to compare thyroid-stimulating hormone (TSH) and free tetraiodothyronine (fT4) concentrations using ELISA (Enzyme-Linked Immunosorbent Assay) and LFA (Lateral Flow Assay) in human serum samples. We obtained 96 serum samples that were evaluated with a commercial kit for ELISA (Accubind Monobite, Lake Forest, California, US) and LFA (Micropofit Fluorecare, Guangdong, Shenzhen, China). We used the Wilcoxon nonparametric test, Bland-Altman, and Passing-Bablok regression to compare TSH and fT4 concentrations obtained by ELISA and LFA. RESULTS: The median TSH concentrations obtained by ELISA and LFA were 1.92 and 2.11 μIU/mL, and for fT4 were 1.14 and 1.10 ng/dL. On the other hand, the Spearman's rho between ELISA and LFA for TSH was 0.845. The TSH concentrations between ELISA and LFA had significant differences (p <
0.05). About fT4, the Spearman's rho between methods was 0.348. The bias for TSH was -0.315, and for fT4 was -0.013. The Deming regression for TSH (p = 0.309) and fT4 (=0.938) shows that the levels obtained by both methods do not present significant differences, and the Passing-Bablok regression identifies significant bias between both methods, especially for fT4, in the range of concentrations studied. CONCLUSION: TSH measurement by LFA may be a viable alternative to evaluate thyroid diseases, but fT4 by LFA is not precise and presents a high bias.