TSH is lower after a meal, with no evidence that either sampling condition is preferable

Samples compared within the same people

The review brings together studies comparing thyroid measurements taken from the same people on the same day, first fasting and then after a meal. This comparison concerns laboratory results and their diagnostic interpretation, not the long-term course of a disease.

For TSH, the synthesis includes 14 studies and 1,257 participants. The authors used a random-effects meta-analysis and assessed the certainty of the findings using the GRADE approach. The available source material contains only the PubMed abstract, not the full text.

An average difference, but highly heterogeneous results

After a meal, TSH was on average 0.802 mIU/L lower than the fasting value. The 95% confidence interval ranged from −1.073 to −0.531 mIU/L for the difference calculated as post-meal minus fasting. This average does not predict the change in any given person.

The difference persisted in sensitivity analyses, but heterogeneity was very high, with an I² of 99.7%. The authors therefore rate the evidence as low-certainty. They report larger differences in subgroups with overt or subclinical hypothyroidism.

No statistically significant difference was detected for FT4 or FT3. This does not demonstrate that their values are strictly identical under the two conditions.

The main issue is results close to a threshold

According to the authors, lower TSH after a meal could lead some borderline results to be classified as consistent with normal thyroid function. This represents a risk of a change in laboratory classification, not evidence that a meal improves thyroid function.

The synthesis suggests that documenting and standardizing fasting status could help with interpreting these borderline results. However, it establishes neither the superiority of fasting sampling nor that of post-meal sampling. It is therefore insufficient to justify a new general sampling rule.

What we can take away

Sampling conditions may matter when interpreting TSH close to a threshold. This meta-analysis does not show that fasting tests should always be preferred.