Thyroid-Stimulating Hormone Levels and Incident Depression

Results From the ELSA-Brasil Study

Ana C. Varella; Isabela M. Benseñor; Carolina C.P.S. Janovsky; Alessandra C. Goulart; Marina G. Birck; Itamar S. Santos; Andre R. Brunoni; Paulo A. Lotufo


Clin Endocrinol. 2021;94(5):858-865. 

In This Article


The final sample of 11,986 participants provided available information for both thyroid function levels and depression at baseline and follow-up. Mean age was 51.5 years, and 51.2% (n = 6096) were women; 1.4% (n = 169) had subclinical hyperthyroidism; 5.7% (n = 677) had subclinical hypothyroidism; and 92.9% (n = 11,050) were euthyroid individuals.

Table 1 shows the population characteristics according to quintiles of TSH. Those with higher TSH (Q5) levels were more frequently white-skinned, less frequently current smokers had slightly higher BMI and were older.

Table 2 shows the Poisson regression models. Overall, we found a significant association between low TSH levels and higher incidence of depression after 4 years of follow-up comparing the 3rd and 1st TSH quintiles. Results remained significant after adjustment for sex, age, race, education, BMI, smoking, alcohol consumption, use of antidepressants/benzodiazepines, kidney function and comorbidities (adjusted RR = 1.36, 95% CI 1.02–1.81). This association remained significant for women (adjusted RR = 1.64, 95% CI 1.15–2.33), but not for men (adjusted RR=0.94, 95% CI 0.57–1.56) in stratified analyses according to sex. When restricting the analysis to euthyroid participants (Table 3) we found similar results: a significant association between low TSH levels (within the normal range) and incident depression comparing the 3rd with 1st TSH quintiles (adjusted RR=1.46, 95% CI 1.08–1.99). It remained significant for women (adjusted RR = 1.63, 95% CI 1.12–2.38), but not for men.

In our supplementary analysis (not shown), changing the reference category to the 1st TSH quintile, we found a significant inverse association between high TSH levels and incident depression, even adjusting for covariates (adjusted RR = 0.71, 95% CI 0.53–0.96). It remained significant for women (adjusted RR = 0.67, 95% CI 0.47–0.95), but not for men. Conducting the same analysis among euthyroid participants, no association between normal TSH levels and incident depression was found for any participants and either sex. On analysis of TSH levels ≥10 mIU/L (not shown), no association was found for any participants or for women. But only 36 participants had TSH levels ≥10 mIU/L and no men with high TSH levels had depression at follow-up.

No association was found between thyroid dysfunction as a categorical variable and incident depression (not shown), for either subclinical hyperthyroidism (adjusted RR = 1.02, 95% CI 0.47–2.21) or subclinical hypothyroidism (adjusted RR = 0.86, 95% CI 0.56–1.34), even stratifying the analysis by sex.