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


We found a significant association between low TSH levels and higher incidence of depression, particularly among women. Restricting the analysis to euthyroid participants, similar results were found—that is, low-normal TSH levels were associated with incident depression for all participants and for women. Changing the reference to the first quintile, we found an inverse association between high TSH levels and depression, remaining significant for women only. A sensitivity analysis with TSH levels ≥10 mIU/L did not show significant results, probably due to the small sample. Analysing thyroid dysfunction as categories, we found no association between subclinical hyperthyroidism or hypothyroidism and incident depression.

Evidence for the association between TSH levels and depression is conflicting, and even though it is complicated to compare our study to other cross-sectional analysis, it is possible that our findings expand the results of three previous studies on subjects with subclinical thyroid dysfunction and euthyroidism. A study by Kvetny et al.,[5] which analysed a sample of 14,787 men and women from the Danish General Suburban Population Study (GESUS), found that low TSH levels were associated with higher prevalence of depression, when assessed using the Major Depression Inventory. Panicker et al.,[6] with a sample of 33,234 from The Nord-Trøndelag Health Study (HUNT study), found that higher TSH levels were associated with lower risk of depression. A smaller study using data from the 2014 Korean National Health and Nutritional Examination on 1,651 adults, aged 20 years and over, found similar results to ours for women, with an inverse association between high TSH levels and depression. However, they found opposite results for men, where higher TSH levels were associated with higher prevalence of depression.[7] Another study, using data on 6,869 participants, aged 17–39 years, from the National Health and Nutrition Examination Survey (NHANES III), which assessed depression as per the Diagnostic and Statistical Manual of Mental Disorders (DSM-III), also found different results for men and women. Low TSH levels were associated with depression for men, but not for women.[4] The difference in age range could have influenced our different results for men. Another study, with a sample of 1,125 participants in the Nijmegen Biomedical Study, evaluated depression using BDI and found no association between TSH levels and depression, which was probably due to the small number of subjects outside the normal range in this study.[8]

Some studies prospectively analysed the association of TSH levels and depression. Almeida et al.,[10] with a community sample of 3,932 men aged 69–87 years in Western Australia, found no association between TSH levels and incident depression in men. Furthermore, a large prospective study consisting of 92,206 participants aged over 18 years found no association between high-normal TSH levels and depression after a mean follow-up period of 2 years.[11] Our analysis, focusing on high-normal TSH levels (reference in 1st quintile), did not show an association either, again suggesting an association towards low TSH levels only. However, Kim et al.,[9] in a retrospective cohort study using data from 13,017 participants aged 17–84 years in South Korea (mean follow-up period of 3.2 years), reported that high TSH levels were associated with increased risk of depression as assessed by the Beck Depression Inventory (BDI) in women. Sample characteristics, particularly for the younger age range, included study design while the instrument used to evaluate depression might have influenced the difference in our results.

In Brazil, few studies have analysed the association between TSH levels and depression. Guimarães et al.,[23] using a sample of 1,249 women, found an association between TSH levels >10 IU/ml and depression in a representative sample of the general population. They included participants with overt hypothyroidism, while we only had 36 participants with TSH levels higher than 10, which might have influenced the difference in results. Chueire et al.,[24] also reported a positive association between high TSH levels and depression in elderly participants. Almeida et al.[25] reported a positive association between subclinical hypothyroidism and depression in a small sample. Chueire et al. and Almeida et al. included hospital samples, unlike ours which was selected from apparently healthy civil servants. Using cross-sectional analysis performed with the ELSA-Brasil baseline data, no association between subclinical thyroid dysfunction and prevalent depression was reported.[26]

Our results for subclinical hyperthyroidism and hypothyroidism as categorical variables showed no association with incident depression, prospectively confirming our abovementioned baseline results. These results tally with two prospective cohort studies that reported no association between subclinical hypothyroidism and depression.[11,27] However, Blum et al found an association of subclinical hyperthyroidism and an increased risk of depression after 3 years of follow-up.[27] Our findings for subclinical hyperthyroidism are consistent with our main findings of a positive association of low TSH levels and depression, but the small number of participants with both subclinical hyperthyroidism and depression in our sample might explain the lack of statistical significance.

One could argue that the U-shaped curve of association may not appear due to the small number of participants in the lower TSH quintile compared with the highest. This is not the case here, since we found that low TSH levels were associated with higher incidence of depression, and high TSH levels were associated with lower incidence of depression. Our study includes a relevant proportion of men in the sample (5,800; 48.8%); therefore, our findings may reflect characteristics of this specific sample and not exclusively a lack of power to analyse for this in men. Our results for women were positive even considering the relatively low prevalence of depression in the sample at baseline (4.2%) and after a 4-year follow-up (4.4%), and also when restricting the analysis to TSH levels within the normal range, indicating an association between low-normal TSH levels and incident depression.

This study has some strengths, such as the large sample size, standardized protocols and its prospective design. Frequency of previous cardiovascular disease or cancer is low in the sample and no participants were in current treatment for cancer during data collection. Some limitations are that a single measurement of TSH levels was performed at baseline; however, ELSA-Brasil is a cohort of healthy workers and no thyroid measurement was performed in patients with acute clinical disease. Even though CIS-R levels at baseline were higher in participants who will develop depression at follow-up as expected, there was no difference in TSH levels at baseline between the groups (without depression vs. with depression, at baseline ) suggesting no reverse causality related to TSH levels as the cause of depression at follow-up. Also, participants with subclinical thyroid disease receiving thyroid medication for any reason were excluded from the analysis, which may have influenced our results.

In conclusion, our study showed that low TSH levels seem to be positively associated with incident depression, even considering low-normal TSH levels. These associations were stronger among women. In contrast, high TSH levels were inversely associated with incident depression, also among women.