TSH and fT4 during pregnancy: an observational study and a review of the literature

Background: Trimester-specific reference intervals for TSH are recommended to assess thyroid function during pregnancy due to changes in thyroid physiology. - Laboratories should verify reference intervals for their population and assay used. No consistent upper reference limit (URL) for TSH during pregnancy is reported in literature. We investigated the use of non-pregnant reference intervals for TSH, recommended during pregnancy by current Dutch guidelines, by deriving trimester-specific reference intervals in disease-free Dutch pregnant women as these are not available. Methods: Apparently healthy low risk pregnant women were recruited via midwifery practices. Exclusion criteria included current or past history of thyroid or other endocrine disease, multiple pregnancy, use of medication known to influence thyroid function and current pregnancy as a result of hormonal stimulation. Women who were TPO-antibody positive, miscarried, developed hyperemesis gravidarum, hypertension, pre-eclampsia, HELLP, diabetes or other disease, delivered prematurely or had a small for gestational age neonate were excluded. Blood samples were collected at 9-13 weeks (n=99), 27-29 weeks (n96) and 36-39 weeks (n96) of gestation and at 4-13 weeks post-partum (n95). Sixty women had complete data during pregnancy and post-partum. All analyses were performed on a Roche Cobas e601 analyser. Results and conclusions: In contrast to current Dutch guidelines, the 97.5th percentiles of TSH in the first (3.39 mIU/L) and second trimesters (3.38 mIU/L) are well under the non-pregnant URL of 4.0 mIU/L. The higher TSH in the third trimester (97.5th percentile 3.85 mIU/L) is close to the current non-pregnant URL of 4.0 mIU/L. Absolute intra-individual TSH is relatively stable during pregnancy and post-partum as individuals tracked within the tertile assigned in trimester 1. Even small deviations within the population reference interval may indicate subtle thyroid dysfunction.