Thyroid-stimulating hormone reference intervals in pregnancy: comparison of refineR and KOSMIC algorithms

Abstract Objectives The aim of this study was to establish reference intervals (RIs) for serum thyroid-stimulating hormone (TSH) levels obtained in pregnant women in conjunction with first- and second trimester Down syndrome screening using the refineR and KOSMIC algorithms. Methods TSH test results of the pregnant women (16–49 years) who underwent first (n=1,679) and second-trimester (n=1,217) screening tests between April 2021 and April 2025 were retrieved from the Laboratory Information System. RIs were estimated using both the refineR and KOSMIC algorithms. The bias ratio matrix (BR) was used to quantify the differences between the lower (LRL) and upper limits (URL) of RIs. Results In pregnant women at 11–14 and 16–22 weeks of gestation, the RIs for TSH obtained using the refineR algorithm and the KOSMIC algorithm differed from both the American Thyroid Association (ATA) guideline limits and the manufacturer-provided RIs. The LRLs of TSH were comparable between algorithms and manufacturer provided RIs; whereas, the URLs showed significant deviations (BR ≥0.375). In the 11–14 week gestational cohort, elevated TSH was detected in 0.7 % women using ATA RIs, compared with 4.1% and 3.2 % women exceeding the upper reference limit when applying the refineR and KOSMIC approaches, respectively. Conclusions The KOSMIC and refineR algorithms provide comparable RIs estimates from routine laboratory data. The minor differences between the estimated limits emphasize the importance for laboratories to explicitly state the algorithm used to determine reference intervals.