Context: Many observational studies have reported on the association between educational attainment (EA) and thyroid function, but the causal relationship remains unclear.
Objective: We aimed to obtain causal effects of EA on thyroid function and to quantify the mediating effects of modifiable risk factors.
Methods: Two-sample mendelian randomization (MR) was performed by using summary statistics from large genome-wide association studies (GWAS) to assess the effect of EA on thyroid function, including hypothyroidism, hyperthyroidism, thyrotropin (TSH), and free thyroxine (FT4). A multivariable analysis was conducted to assess the mediating role of smoking and help to explain the association between EA and thyroid function. Similar analysis was further performed using data from the National Health and Nutrition Examination Survey (NHANES) 1999 to 2002.
Results: In MR analysis, EA was causally associated with TSH (β = .046; 95% CI, 0.015-0.077; P = 4.00 × 10-3), rather than hypothyroidism, hyperthyroidism, and FT4. Importantly, smoking could serve as a mediator in the association between EA and TSH, in which the mediating proportion was estimated to be 10.38%. After adjusting for smoking in the multivariable MR analysis, the β value of EA on TSH was attenuated to 0.030 (95% CI, 0.016-0.045; P = 9.32 × 10-3). Multivariable logistic regression model in NHANES suggested a dose-response relationship between TSH (quartile [Q]4 vs Q1: odds ratio = 1.33; 95% CI, 1.05-1.68; P for trend = .023) and EA. Smoking, systolic blood pressure, and body mass index partially mediated the association between EA and TSH, with the proportion of the mediation effects being 43.82%, 12.28%, and 6.81%, respectively.
Conclusion: There is a potentially causal association between EA and TSH, which could be mediated by several risk factors, such as smoking.
Keywords: educational attainment; mediation analysis; mendelian randomization; smoking; thyroid function.
© The Author(s) 2023. Published by Oxford University Press on behalf of the Endocrine Society. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.