Effect of fasting and postprandial states on thyroid function test parameters and diagnostic interpretation: A systematic review and meta-analysis

Thyroid function tests are among the most frequently requested investigations; several pre-analytical factors may influence them. Reference ranges are not standardized for fasting versus postprandial sampling. We aimed to synthesize evidence comparing fasting and postprandial thyroid function test and to assess the effect of these on diagnostic classification.

This systematic review and meta-analysis followed PRISMA 2020 for reporting and was prospectively registered (PROSPERO CRD420261344874). MEDLINE (through PubMed), Embase, Scopus and Web of Science were searched using a PECO framework for observational & RCT studies (publication cut off as March 24th, 2026) reporting paired, same-day fasting and postprandial TSH, FT4 or FT3 in the same individuals. Data extraction and risk-of-bias assessment were performed in duplicate. Comparisons were analysed as single-group pre-post data using random-effects models with the mean difference calculated as postprandial minus fasting levels. The analysis was conducted at study level, comparator sub-group level and a three-level model as alternatives. Heterogeneity was quantified with I² and certainty was rated using GRADE.

Fourteen studies were included. For TSH (14 studies with 24 subgroup comparisons; n = 1257), the pooled mean difference (MD) was - 0.802 mIU/L (95% CI - 1.073 to - 0.531; I² = 99.7%), indicating significantly lower postprandial TSH (low certainty of evidence); and robust across subgroups and sensitivity analyses, with greatest in overt (- 2.518 mIU/L) and subclinical hypothyroidism (- 2.173 mIU/L). No significant fasting-postprandial difference was found for FT4 or FT3.

Caloric intake might significantly lower postprandial TSH, but with low certainty. This has clinical relevance in borderline thyroid dysfunction, where it may drive misclassification towards euthyroid state. Standardizing and recording fasting status may improve interpretation of borderline thyroid results, although current evidence does not establish superiority of either sampling condition.