Long term monitoring of serum valproate assays using patient median values show stable test results over a period of more than 20 years

Monitoring assay performance using traditional internal quality control (IQC), and external quality assurance (EQA) methods has limitations, particularly regarding sample commutability. This study investigates the use of patient-derived valproate data, specifically median and quartile values, as a complementary approach to quality control in a clinical laboratory setting. A retrospective analysis was performed on 13,336 routine serum valproate results collected between January 2004 and December 2024 at Akademiska Hospital, Uppsala. Valproate was analyzed using immunoassays on the Architect ci8200 platform until 2020, after which testing was transferred to the Roche Cobas Pro c503. Annual patient medians and quartiles were calculated, and seasonal patterns were evaluated. Method transfer and performance were assessed using internal controls and patient data. The number of annual test requests declined modestly over the study period. Median valproate concentrations remained stable across the 21-year period, with minimal variability in interquartile ranges. The 0.10 and 0.90 percentiles showed slightly higher fluctuation, likely due to sample timing variation rather than analytical drift. Seasonal analysis revealed no clinically relevant variation. The method transfer in 2021 did not introduce observable shifts in patient medians, supporting good method comparability. Longitudinal monitoring of patient median and quartile valproate concentrations provides a robust, clinically meaningful tool for supplementary assay performance surveillance. This approach is particularly suitable for therapeutic drug monitoring, where treatment protocols and target ranges are stable over time. • Patient medians used to monitor valproate assay performance over 21 years. • Valproate medians remained stable despite platform change in 2021. • Traditional QC was supplemented by real-world patient data trends. • No clinically relevant seasonal variation in valproate levels detected. • Patient-derived metrics offer robust support for method comparability.