TSH continuous reference intervals by indirect methods: A Comparisons to Partitioned Reference Intervals

Objective: To establish continuous reference intervals for TSH by data mining, using quantile regression with restricted cubic splines model. Design and method: TSH results (n = 13,333) were collected for a four years period (between March 2015 and February 2020). After an exclusion step, TSH results (n = 8838) were used to derive continuous reference intervals (i.e. 2.5th and 97.5th percentiles) spanning 18-90 years of age, using quantile regression with restricted cubic splines (RCS) model, then compared to age-partitioned reference intervals generated by Bhattacharya analysis. Results: Despite similar reference intervals to the Bhattacharya analysis, continuous reference intervals appeared to give a more accurate and consistent estimation of the upper reference limits (i.e.97.5th percentiles) with complex age-related variations in serum TSH concentrations. Conclusion: Our results suggest that quantile regression with RCS model appears to be a very useful tool available for clinical laboratories to establish local continuous TSH reference intervals.