Reference change values based on uncertainty models

• Published in: Clinical biochemistry Vol. 80; pp. 31 - 41
• Main Authors: Rigo-Bonnin, Raúl, Muñoz-Provencio, Diego, Canalias, Francesca
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01-06-2020
• Subjects: Analytical phase; Biological variation; Biological Variation, Individual; Blood Chemical Analysis; Humans; Post-analytical phase; Pre-analytical phase; Reference change values; Reference Values; Uncertainty; Uncertainty; Biological variation; Analytical phase; Pre-analytical phase; Reference change values; Post-analytical phase
• ISSN: 0009-9120; 1873-2933
• DOI: 10.1016/j.clinbiochem.2020.03.016

•Uncertainty models are proposed to estimate the reference change values.•Uncertainty sources considered were related to analytical and extra-analytical phases.•Compatibility studies were performed to estimate the reference change values.•The proposed models could be more realistic than classical approaches because more variation sources were considered.
Different reference change values approaches are widely accepted to interpret the change between two consecutive measured values from the same biological quantity in an individual. Nevertheless, we propose two uncertainty-based models to estimate reference change values which will include all possible variation sources.
The models consisted in 1) estimation of the uncertainty related to each measured value, 2) calculation of the change between these values and its uncertainty, and 3) performing a compatibility study to know if the values are of the same degree of equivalence. Also, results obtained using the proposed models and the classical approaches are shown.
The primary uncertainty sources corresponded to the within-subject biological, followed by those related to the analytical or pre-analytical phase, and post-analytical, respectively. We observed higher reference change values when our models were applied than those obtained using the classical approaches.
The estimation of reference change values using our models could be more realistic than the classical approaches because more identified variation sources were considered. We hope that this study will help and stimulate clinical laboratories to perform uncertainty and reference change values studies, and it allows a greater understanding of these concepts and its application in disease monitoring.