Statistical Analysis
Statistical analysis was conducted using the Statistical Package for Social Sciences (IBM SPSS, version 25.0). We first compared the distribution of baseline characteristics and we then compared outcomes between the exposed and the non-exposed groups. Means and standard deviations (SDs) or median and interquartile ranges (IQRs) were used for continuously distributed variables, and t-test or median Mann-Whitney U test were used for statistical significance as appropriate. Percentage was used for categorical variables and χ2 test was used for statistical significance.
Adjusted relative risk (RR) and 95% CI were estimated by log-binomial model to assess the independent association of mirtazapine use with risk of developing hypothyroxinemia (with no mirtazapine use as the reference). Multiple linear regression model was used to assess the association of mirtazapine use with change in free T4 level [Δ free T4(2–1): free T4 value after exposure (2)– free T4 value at baseline (1)]. Potential confounding variables included in the models were age, gender, prior drug exposure (no prior use of psychiatric medications eg. antipsychotics, mood stabilizers and antidepressants for a minimum 4 months before admission, were defined as drug free. No drug free versus drug free), baseline free T4 level, duration of treatment, the use of other kinds of antidepressants, and concomitant non-antidepressant medications.