EXPRESS: Comparative Optimization of Alkaline Phosphatase Reference Intervals Using Hoffmann and refineR Methods in Age- and Sex-Stratified LIS Data

Alkaline phosphatase (ALP) is a routinely used biochemical marker in clinical practice, and its concentrations are substantially influenced by age, growth, developmental stage, and sex. The use of an overly broad or unified reference interval (RI) may therefore introduce bias into result interpretation. In this study, 345,469 ALP test results were included after data cleaning from a large-scale Laboratory Information System (LIS) dataset. After stratification by age and sex, ALP RIs were estimated using the Hoffmann and refineR methods, and their stability, concordance with external RIs, and effects on abnormal result classification were compared. ALP exhibited an overall right-skewed distribution, with higher levels in children and adolescents than in adults. Supplementary pediatric subgroup analysis further showed substantial heterogeneity within the original 0-18-year stratum, especially across childhood and pubertal age ranges. Males generally had higher ALP levels than females before the age of 50 years, whereas this sex-related difference became less pronounced thereafter. ALP varied nonlinearly with age, reaching a peak during puberty, followed by a gradual decline and stabilization in adulthood. Although the two methods yielded similar lower-limit estimates, the refineR method provided a better fit to the right tail of the distribution, generated higher upper limits, and showed closer agreement with external RIs, while also producing a more balanced pattern of abnormal classification. A small independent set of 50 healthy individuals was used only for preliminary verification, and no out-of-range results were observed. These findings suggest that ALP interpretation should consider age and sex stratification, and that the refineR-derived intervals may provide a practical option for local RI optimization, pending further validation in larger independent cohorts.