Objective: The use of age-specific reference ranges for TSH is advocated, but the impact of this on laboratory diagnosis of thyroid dysfunction is unclear. Our aims were to determine age-specific TSH reference ranges and to examine interassay differences in performance.
Design: We analysed TSH results from 223,045 consecutive samples assayed over 1 year by a single pathology provider using the Siemens Centaur assay. We excluded patients with evidence of thyroid disease to derive a reference population of 148,938 individuals and analysed results in the 5-year age bands. We reassayed 120 samples using three other methods (Architect, Roche and Immulite) to assess precision and bias.
Results: The 2·5th percentile for TSH was consistent across age groups (approximately 0·5 mU/l), whereas the 97·5th percentile increased from age 40 upwards, with the reference range upper limit being 3·75 mU/l at age 40 and 5·0 mU/l at age 90. In most age bands, the use of age-specific upper limits reclassified only 0·1-1·9% of participants as normal or abnormal compared with a common cut-off of 4·0 mU/l; in participants aged 85 years or more, reclassification rates were higher (2·1-4·7%). The four TSH assays showed good agreement at low-normal TSH concentrations (<2 mU/l), but at concentrations of 4·0 mU/l, there were intermethod differences of approximately 1 mU/l.
Conclusion: The use of age-specific reference ranges for TSH has only minor effects on thyroid status, except in the very old. At high-normal TSH concentrations, between-method differences in performance have a comparable impact to that of age and may affect clinical decision-making.
© 2012 Blackwell Publishing Ltd.