Association of Levothyroxine Treatment and Thyroid Peroxidase Antibodies with Antidepressant Use: A Danish Population-Based Longitudinal Study.

Background: Subjects receiving levothyroxine (LT4) treatment have increased prevalence of depression, anxiety, and antidepressant use, but whether the underlying mechanism relates to thyroid autoimmunity is still unclarified. Methods: This is a population-based longitudinal study. Baseline biochemical and questionnaire data from the Danish General Suburban Population Study (GESUS) in 2010-2013 were linked with individual-level longitudinal data in national health registries. The aim was to investigate the associations between thyroid peroxidase antibodies (TPOAbs) and LT4 treatment, separately and through interaction, and at least one redeemed prescription for antidepressants. Logistic and Cox regression were used to evaluate initiation of antidepressant use before and after the baseline examination in GESUS, respectively. All exposures and covariates were fixed at the date of baseline examination. Thyroid autoimmunity was defined as serum TPOAbs >60 U/mL. Adjustments included sex, age, education, income, Charlson comorbidity index, smoking, and alcohol. Sensitivity analyses were performed for missing variables, exclusion of lithium use, exclusion of thyroid surgery, and conservative definitions for LT4 treatment and antidepressant use requiring at least two prescriptions. Results: We included 12,894 individuals, of whom 2353 (18%) had "past or current" antidepressant use at baseline, leaving 10,541 individuals at risk for incident antidepressant use after baseline. The median follow-up was 7.8 years during which 783 individuals (7.4% of 10,541 individuals) had incident antidepressant use. TPOAb positivity was not associated with "past or current" (odds ratio [OR] 0.90 [confidence interval, CI 0.78-1.03], p  = 0.13) nor incident antidepressant use (hazard ratio [HR] 1.02 [CI 0.83-1.25], p  = 0.88). LT4 treatment was associated with increased "past or current" antidepressant use (OR 1.33 [CI 1.10-1.62], p  = 0.004) and increased incident antidepressant use (HR 1.38 [CI 1.03-1.85], p  = 0.03). There were no interactions between the effects of TPOAb positivity and LT4 treatment on the use of antidepressants in logistic ( p  = 0.87) or Cox regression models ( p  = 0.82). Sensitivity analyses were robust, except that incident use of at least two redeemed antidepressant prescriptions was not statistically significant. Conclusions: LT4 treatment, but not TPOAb positivity, was associated with increased prevalent or incident antidepressant use with at least one prescription. Our findings do not support that thyroid autoimmunity is an important factor for antidepressant use in patients receiving LT4 treatment.