Association of Thyroid Function with Suicidal Behavior: A Systematic Review and Meta-Analysis

Thyroid disease is a very common condition that influences the entire human body, including cognitive function and mental health. As a result, thyroid disease has been associated with multiple neuropsychiatric conditions. However, the relationship between thyroid dysfunction and suicide is still controversial. We conducted a systematic review and meta-analysis to describe the association of thyroid function with suicidal behavior in adults. We searched four data bases (MEDLINE, EMBASE, PsycINFO, and Scopus) from their inception to 20 July 2018. Studies that reported mean values and standard deviation (SD) of thyroid hormone levels [Thyroid-stimulating hormone (TSH), free thyroxine (FT4), free triiodothyronine (FT3), total thyroxine (TT4), and total triiodothyronine (TT3)] in patients with suicidal behavior compared with controls were included in this meta-analysis. The abstracts and papers retrieved with our search strategies were reviewed independently and in duplicate by four reviewers for assessment of inclusion criteria and data extraction, as well as for evaluation of risk of bias. Random-effects models were used in this meta-analysis to establish the mean difference on thyroid function tests between groups. Overall, 2278 articles were identified, and 13 studies met the inclusion criteria. These studies involved 2807 participants, including 826 participants identified with suicidal behavior. We found that patients with suicide behavior had lower levels of FT3 (−0.20 pg/mL; p = 0.02) and TT4 (−0.23 µg/dL; p = 0.045) compared to controls. We found no differences in either TSH, FT4, or TT3 levels among groups. With our search strategy, we did not identify studies with a comparison of overt/subclinical thyroid disease prevalence between patients with and without suicide behavior. The studies included in this meta-analysis had a low-to-moderate risk of bias. In the available literature, the evidence regarding the association of thyroid disorders and suicidal behavior is limited. We found that patients with suicidal behavior have significantly lower mean FT3 and TT4 levels when compared to patients without suicidal behavior. The clinical implications and pathophysiologic mechanisms of these differences remain unknown and further research is needed.


Introduction
Suicide is one of the main causes of mortality around the globe, accounting for about 800,000 deaths every year [1]. Deaths by suicide are more frequent in individuals with psychiatric disorders, such as major depression and alcohol use disorder. By far the single most important risk factor for suicide is a prior suicide attempt [2]. Therefore, suicide and suicide attempt/ideation are serious public health problems, as these disorders deeply affect families and communities and have long-lasting effects on the people left behind.
Thyroid dysfunction is the most common endocrine disorder [3]. Most current estimates showed that around 12% of Americans will develop a thyroid condition during their lifetime and that around 20 million have a thyroid disorder [4]. Additionally, it has been estimated that one in eight women will develop a thyroid disorder during their lifetime [5]. Some endocrine disorders have been correlated with psychiatric conditions, such as major depression, anxiety, etc. [6][7][8][9]. This is not the case for suicidal behavior, as its relationship with thyroid dysfunction is controversial, and mostly based on observational studies.
A nationwide Danish register-based study, including 111,565 participants, reported that mortality due to suicide was increased (0.10% vs. 0.07%, p < 0.001) in patients with Hashimoto's thyroiditis compared to matched controls, suggesting a possible role of Hashimoto's thyroiditis in the pathophysiologic mechanisms of suicidal behavior [10]. Similarly, a study including 1718 patients with a history of major depressive disorder showed that those with history of suicide attempt had higher levels of thyroid-stimulating hormones (TSH) and thyroid autoantibodies compared with those without history of suicide attempt, suggesting a potential role of alterations in thyroid function tests in the risk of suicide attempts [11]. In contrast, in a large prospective cohort study of women with thyroid dysfunction that explored the risk of cause-specific mortality after nearly 30 years of follow-up, neither hyperthyroidism nor hypothyroidism were associated with higher rates of suicide [12].
Due to the scarcity of evidence and the conflicting findings among observational studies, we conducted, to our knowledge, the first systematic review and meta-analysis exploring the association between thyroid dysfunction with suicidal behavior in adults.

Search Strategy
We conducted a comprehensive search in four data bases (Ovid MEDLINE Epub Ahead of Print, Ovid Medline In-Process & Other Non-Indexed Citations, Ovid MEDLINE, Ovid EMBASE, Ovid, PsycINFO, and Scopus) from their inception to 20 July 2018, with no language restrictions. Our search strategy was designed and conducted by an experienced librarian (P.E.J.) with input from the principal investigator (S.M.) (File S1).
A systematic review software (DistillerSR, Ottawa, ON, Canada) was used to upload the results of the search strategies, for title and abstracts screening, and for full-text review. Each paper was reviewed by two reviewers working in an independent manner. First, we screened abstracts for eligibility, and those that could be potentially included in our metaanalysis underwent an in-depth review of the full-text paper. The chance-adjusted interreviewer agreement among reviewers was high (kappa statistic = 0.82). Any disagreement during the reviewing process was resolved by the principal investigator (S.M.).

Study Selection
For this systematic review and meta-analysis, studies reporting mean values of thyroid function tests [TSH, free thyroxine (FT4), free triiodothyronine (FT3), total thyroxine (TT4), and total triiodothyronine (TT3)] along with their standard deviation (SD) in patients with suicidal behavior (recent episode or history of suicidal ideation/suicide attempt) compared with patients without suicidal behavior were included. We contacted the authors of those papers in which additional information or clarification was required in order to confirm eligibility. In cases where we were not able to reach them, these studies were excluded from the analyses. Studies with only abstracts available, papers published in languages other than English or Spanish, papers without a quantitative assessment of thyroid function, and papers without a comparison to a control group (group without suicidal behavior) were excluded.

Data Extraction and Quality Assessment
We designed and used a standardized form to retrieve data from each study included in the meta-analysis. The data extracted included: first author name, paper title, publication year, country, publishing journal, study design, type and criteria used to define the group with suicidal behavior and control group, sample size of suicidal behavior group and control group, mean and SD of age by groups, percentage of female by groups, and mean and SD of thyroid function tests (TSH, FT4, TT4, FT3, TT3) by groups. The data from each paper were extracted independently and in duplicate by four reviewers.
For the assessment of methodological quality and risk of bias of the included studies we used the Newcastle-Ottawa risk of bias tool for observational studies [13]. This tool has been extensively validated and is used to assess the comparability of the study groups, their representativeness, and the ascertainment of exposure and outcomes. Reviewers worked independently, assessing the risk of bias of the included studies. Any disagreement during this stage was resolved by consensus.

Statistical Analysis
The main outcomes summarized in this meta-analysis were the mean difference (MD) in thyroid function tests for the suicidal behavior group compared with the control group. We calculated and pooled unstandardized MDs of the included studies using random effects models [14]. In the case of multiple subgroups from one study sharing the same comparison group, these groups were combined following the recommended procedure [14]. We used the I 2 statistic to estimate the percentage of total between-study variation due to heterogeneity rather than chance [14]. Sensitivity analyses, when possible, were performed according to the type of suicidal behavior of the study population (recent suicide attempt, recent suicidal ideation, history of suicidal attempt, history of suicidal ideation). We performed leave-one-out sensitivity analyses by iteratively removing one study at a time and recalculating the summary MD for those outcomes assessed in ≥10 studies. We assessed publication bias with funnel plot for visual detection of asymmetries and Egger's tests for the detection of statistical asymmetry in the funnel plot when ≥10 studies were available for a specific outcome [15,16]. All values are two-tailed, and p < 0.05 was set as the threshold for statistical significance. Review Manager v5.4, OpenMeta[Analyst], and IBM SPSS statistics v26 were used for statistical analyses.

Study Selection and Characteristics
Our search strategies identified 2278 articles, of which 102 were evaluated as full text and 13 studies [17][18][19][20][21][22][23][24][25][26][27][28][29] were finally selected for the meta-analysis ( Figure 1). Table 1 summarizes the characteristics of the studies included in the meta-analysis. These studies involved 2807 participants, of whom 826 participants had suicidal behavior. Sample sizes ranged from 20 to 555 participants; mean ages of the participants ranged from 23 to 50 years. All the included studies reported thyroid function tests as mean and SD for the suicidal behavior group compared with control group. The studies included in this meta-analysis have a low-to-moderate risk of bias (Supplementary Table S1).

Records after duplicates removed 2278
Additional records identified through other sources 0

Records screened 2278
Full-text articles assessed for eligibility 102

Records excluded 2176
Full-text articles excluded, with reasons:

89
No outcome of interest No or inappropriate control group

22
No population of interest

9
Studies included in the meta-analysis

Publication Bias and Leave-One-Out Sensitivity Analysis
We did not find major asymmetries in the funnel plots to suggest high risk of publication bias (Supplementary Figure S1). For outcomes assessed in ≥10 studies we performed Egger's tests, TSH (p = 0.43) and FT4 (p = 0.69), which suggested no significant publication bias existed for these outcomes. We performed a leave-one-out sensitivity for those outcomes assessed in ≥10 studies (TSH and FT4 levels) and the summary MDs remained stable (Supplementary Figure S2). This finding indicates that our results were not driven by any single study and that similar results could be obtained after excluding individual studies.

Discussion
In this systematic review and meta-analysis, we found that patients with suicidal behavior have lower levels of FT3 and TT4 compared with controls, but we found no difference in TSH, FT4, or TT3 levels. Furthermore, the subgroup of patients with history of suicide attempt had lower FT3 and FT4 levels, and higher TSH levels compared to controls. This study represents the first summary about the relationship between suicidal behavior with thyroid function.
Our results are in line with previously published evidence regarding the risk/ association of thyroid disorders with suicide behavior [10,30,31]. Heiberg-Brix et al. found that patients with Hashimoto's thyroiditis had an increased frequency of death by suicide (HR = 1.31; 95% CI, 1.04-1.65, p = 0.024) compared to euthyroid controls in a register-based Danish study [10]. In contrast, this frequency is not increased in patients with Graves' disease compared to controls without Graves' disease or euthyroid controls from the general population [32,33]. According to these findings, suicide possibly has a stronger association with autoimmune hypothyroidism rather than autoimmune hyperthyroidism. Similar results have been reported for suicidal attempt/ideation and thyroid disorders. A cross-sectional study showed that the prevalence of hypothyroidism in a group of 31 patients with bipolar disorder and suicide attempt was higher (25.8%) compared with a prevalence of 15.9% in a group of 63 patients with bipolar disorder with no suicide attempt [30]. Sanna et al. reported a 6.5% prevalence of thyroid disorders in males with history of suicidal ideation, compared with a 1.9% prevalence in males without history of suicidal ideation [31]. Nevertheless, these results should be analyzed carefully as all these studies are observational with relatively small sample size and analyses are not adjusted for confounders.
The development of suicide, depression, and other affective disorders has been previously associated with autoimmune diseases, including thyroid-specific diseases [10,[34][35][36]. In a meta-analysis published by Siegmann et al., patients with Hashimoto thyroiditis and either subclinical or overt hypothyroidism showed significantly higher scores on standardized depression instruments compared to euthyroid controls [37]. A second meta-analysis published in 2019, found that patients with subclinical hypothyroidism had higher risk of depression than euthyroid controls. However, there was no difference in the mean TSH level between individuals with depression and healthy controls [38]. Although the majority of patients with depression and other mood disorders have completely normal thyroid function tests, several test abnormalities have been described including elevated TT4 levels, low TT3 and FT3 levels, blunted TSH response to thyrotropin-releasing hormone, and positive antithyroid antibodies [36,[39][40][41][42][43][44][45]. Interestingly, thyroid hormone in the form of liothyronine has been used for the treatment of depression, mainly as an augmentation therapy in severe forms of depression [46][47][48][49]. In addition, in the setting of suicidal behavior, there is evidence suggesting that thyroid hormones might have a role in the regulation of the neurotransmitters involved in suicide pathogenesis, such as serotonin and norepinephrine [50][51][52][53]. As demonstrated by our results, where most of the participants had concomitant affective disorders (depression or bipolar disorder) or schizophrenia, patients with suicidal behavior had significantly lower levels of FT3 and TT4, albeit not clinically meaningful, when compared with the control group.
To our knowledge, this is the first systematic review and meta-analysis of available evidence evaluating the relationship between suicide and thyroid function tests. There were some limitations to our study. First, due to the observational nature of all the studies included in this meta-analysis, it was not possible to establish causality. Second, although we used clear and standardized inclusion criteria and comprehensive search strategies, there remains possible sources of bias such as incomplete searching, publication bias, and the influence of confounding factors (current therapies, psychiatric comorbidities, environmental factors, etc.) in our results. Third, there was heterogeneity of the criteria used to define suicide behavior in the included studies (based on clinical records, psychiatric assessment, self-reported, etc.) and the presence of other concomitant psychiatric diseases which can affect the risk of suicidal behavior. Importantly, suicide is a very complex phenomenon related to many mental health diagnoses and with a multitude of biological, psychological, and social variables which should be considered when interpreting our findings.

Conclusions
In conclusion, we found that the available evidence regarding the association of hypothyroidism or hyperthyroidism with suicidal behavior is limited. In our best effort to summarize all the available evidence, we have described that patients with suicidal behavior had significantly lower levels of FT3 and TT4 levels compared with controls. However, the data in this field are scarce and have significant heterogeneity. Future large, well conducted studies are needed to increase our confidence in the findings presented here, especially studies reporting the specific association of hypothyroidism/hyperthyroidism in this population, which can provide a better understanding, evaluation, and follow-up of patients with thyroid dysfunction and suicidal behavior.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/medicina57070714/s1. Figure S1, Funnel plots for publication bias analysis; Figure S2, Leaveone-out sensitivity analysis of the studies included; Table S1, Summary of risk of bias assessment for the included studies. File S1: Search Strategies Used in this Project.