Data-driven age partitioning for pediatric serum copper: implications for reference interval interpretation

Objectives To establish age-specific reference intervals for serum copper in healthy Vietnamese children and adolescents using a data-driven age-partitioning approach and to assess its implications for result interpretation. Methods Apparently healthy participants from birth to <19 years were recruited. Serum copper was measured by graphite furnace atomic absorption spectrophotometry. Candidate age thresholds were explored using recursive partitioning analysis and subsequently rounded to clinically interpretable groups. Final age partitions were determined based on recursive partitioning results, between-group statistical comparisons, effect size, clinical interpretability, and sample size. Differences between adjacent age groups and between sexes were assessed using the Mann–Whitney U test, Cliff's delta, and the Harris–Boyd method. Outliers were identified using the Horn algorithm. Reference intervals were estimated by the nonparametric percentile method in accordance with CLSI EP28-A3c. Results Of 1157 eligible participants, 36 outliers were excluded, leaving 1121 for analysis. Final age groups were 0–<6 days, 6 days to <12 months, 12 months to <7 years, and 7 to <19 years. Significant differences were observed between adjacent age groups, supporting age partitioning. Sex-related differences were small and did not support sex-specific partitions. The derived reference intervals were 1.9–15.4 μmol/L, 6.5–26.9 μmol/L, 9.3–35.8 μmol/L, and 7.9–22.6 μmol/L for the respective age groups. Conclusions Serum copper concentrations vary substantially with age, particularly in early life, while sex differences are limited. Age-specific, but not sex-specific, reference intervals are recommended. These findings support clinically appropriate age partitioning to improve interpretation of pediatric copper results and reduce potential misclassification associated with overly broad age groups.