ĐÁNH GIÁ ĐỘ TƯƠNG ĐỒNG CORTISOL HUYẾT THANH TRÊN HỆ THỐNG COBAS PRO VÀ ATELLICA SOLUTION Ở TRẺ EM

Abstract

Background: Serum cortisol is a crucial marker in diagnosing and monitoring endocrine disorders. However, differences between automated analytical systems may affect the accuracy and consistency of clinical interpretation. Objective: To assess the analytical agreement of serum cortisol measurements between two automated platforms, Cobas Pro and Atellica Solution, in pediatric samples. Methods: A prospective descriptive study was conducted on 305 pediatric serum samples collected at the Vietnam National Children's Hospital from June to December 2024. Samples were analyzed concurrently on Cobas Pro and Atellica Solution. Analytical performance was evaluated according to CLSI EP15-A3 and EP09-A3 guidelines. Statistical analyses included Pearson correlation, Passing-Bablok regression, and Bland-Altman analysis. Results: The two systems demonstrated a strong correlation (r = 0.9652, p < 0.0001). However, Passing-Bablok regression revealed a significant proportional difference (slope 1.2580; 95% CI: 1.2194–1.2973), and the Cusum test indicated non-linearity (p = 0.04). Bland-Altman analysis showed that Atellica Solution reported higher cortisol values than Cobas Pro by an average of 51.59 nmol/L (corresponding to -19.17%). Calibration differences, antibody specificity, and corticosteroid-binding globulin (CBG) variations likely contributed to the discrepancies. Conclusions: Despite the strong correlation between Cobas Pro and Atellica Solution, significant proportional and systematic differences prevent direct interchangeability without adjustment. It is necessary to establish system-specific reference intervals or apply appropriate conversion factors to ensure accurate clinical interpretation.