Continuous versus discrete reference intervals for pediatric thyroid hormones: Clinical implications from a GAMLSS-based reclassification analysis.

BACKGROUND: Discrete reference intervals (dRIs) may complicate the interpretation of biomarkers that change continuously during childhood. We aimed to establish age-specific continuous reference intervals (CRIs) for thyroid-stimulating hormone (TSH) and free thyroxine (fT4), and to evaluate classification performance of CRIs and dRIs for clinical impact compared with dRIs. METHODS: A total of 1680 TSH and 1614 fT4 measurements from healthy children aged 28 days to 18 years were retrospectively analyzed using an indirect sampling approach from a hospital-based cohort. Age-related CRIs were constructed using the generalized additive models for location, scale, and shape (GAMLSS). Reference limits were defined as the 2.5th and 97.5th percentiles. Agreement and reclassification analyses were performed to quantify classification differences between CRI- and dRI-based interpretations. RESULTS: GAMLSS analysis revealed significant age-dependent fluctuations. CRI was 0.87-4.33 mIU/L at age 12, 0.71-4.23 mIU/L at age 18, whereas dRIs were 0.70 to 4.21 mIU/L at 12-18 years for TSH. CRI upper limits at ages 18 was 1.58 ng/dL for fT4. Conversely, dRI upper limit of 1.50 ng/dL. Agreement was almost perfect for TSH and fT4 (Kappa = 0.85 and 0.82 respectively), with reclassification rates of 1.55% and 1.92%, respectively. The highest reclassification was observed in adolescences for fT4 (2.59%), where some dRI "within range" results were reclassified as "below" or "above range" by CRIs. CONCLUSION: By enabling age-continuous interpretation of pediatric thyroid function tests, CRIs may provide a more age-appropriate framework than broadly partitioned dRIs, particularly for results near the reference limits.