Population-specific reference intervals for 22 biochemical analytes: three indirect methods application in a...

Reference intervals (RIs) vary according to demographic characteristics; however, non-local, fixed RIs remain widely applied in multi-ethnic regions. This study aimed to establish population-adapted RIs for routine biochemical analytes in Uyghur and Han adults from Aksu, Xinjiang.

A total of 1,825,257 laboratory test results obtained from routine health check-ups were retrospectively analyzed. Variance components analysis was used to quantify the relative contributions of sex, age, and ethnicity, with standard deviation ratios guiding decisions on subgroup partitioning. RIs for 22 biochemical analytes were derived using three indirect algorithms (kosmic, refineR, and Hoffmann). Continuous age-dependent RIs were constructed using a refineR-generalized additive model for location, scale, and shape (GAMLSS) framework and evaluated using an independent validation dataset.

Sex was the predominant source of variability for most analytes. Ethnic effects were selective: apolipoprotein A-I (ApoA-I) was the only analyte exceeding the IFCC threshold for ethnic partitioning (SDR_ethnic = 0.39), showing consistently lower reference limits in Uyghur compared with Han individuals, while high-density lipoprotein cholesterol (HDL) exhibited a borderline ethnic signal. Continuous modelling identified age-related declines in albumin and increases in urea, glucose, and lactate dehydrogenase, particularly among adults aged ≥60 years. Overall, 79.23% of RI sets were successfully validated, and most differed from manufacturer-provided reference limits.

In this multi-ethnic population, sex and age were the primary determinants of biochemical variability, whereas ethnicity exerted a selective but clinically relevant effect on ApoA-I. Continuous age-dependent RIs more accurately characterize age-related changes in older adults. Indirect, data-driven approaches provide a practical and scalable strategy for establishing population-adapted RIs in demographically diverse settings.