Our findings support the hypothesis that GAD is an independent risk factor for death by suicide after adjusting for comorbid psychiatric disorders, specifically schizophrenia, bipolar disorder, major depressive disorder, OCD, neurodevelopmental disorders, AUD, and SUD. Additionally, we observed an additive effect of major psychiatric comorbidities with GAD, especially schizophrenia, bipolar disorder, major depressive disorder, AUD, and SUD, on suicide. Individuals comorbid with major depressive disorder and GAD exhibited the highest suicide risk (up to seven times higher than that for the non-GAD controls), followed by those comorbid with bipolar disorder, schizophrenia, AUD, SUD, and OCD.

At least one epidemiological study revealed an association between GAD and suicide, regardless of comorbidity with other major psychiatric disorders, such as schizophrenia and major affective disorders [5]. A national study of 887,859 veterans with depression in the United States observed a significant association between GAD and death by suicide (OR: 1.27, 95% CI: [1.09, 1.47]) [19], suggesting that GAD may be a precipitating factor of suicide in patients with depression. However, Bentley et al. suggested that anxiety disorders are associated with suicidal thoughts (OR: 1.49, 95% CI: [1.18, 1.88]) and suicide attempts (OR: 1.64, 95% CI: [1.47–1.83]), but not death by suicide (OR: 1.01, 95% CI: [0.87, 1.18]) [7]. Sareen et al. followed 4,796 adult participants (52.4% of them with anxiety disorders) for 3 years and observed that GAD was associated with suicidal thoughts (OR: 1.96, 95% CI: [1.45, 2.64]) and suicide attempts (OR: 1.33, 95% CI: [0.86, 2.05]) at baseline, independent of major psychiatric disorders (schizophrenia, mood disorders, AUD, and SUD), revealing that GAD independently predicted the occurrence of suicidal thoughts (OR: 2.48, 95% CI: [1.02, 6.06]) and suicide attempts (OR: 2.35, 95% CI: [0.78, 7.10]) during follow-up [5].

Sareen et al. also identified a similar influence of anxiety disorders alone (OR: 3.34, 95% CI: [1.75, 6.40]) and mood disorders alone (OR: 3.46, 95% CI: [1.78, 6.72]) on new-onset suicidal thoughts during the 3-year follow-up [5], consistent with our findings that a diagnosis of GAD alone was associated with death by suicide and that psychiatric comorbidities with GAD, such as schizophrenia, bipolar disorder, major depressive disorder, AUD, and SUD, increased the likelihood of death by suicide. Furthermore, the nonoverlapping CI between the group with GAD only and the group with GAD and psychiatric comorbidities (schizophrenia, bipolar disorder, major depressive disorder, AUD, and SUD) in the present study supports the hypothesis that GAD is an independent risk factor for death by suicide, suggesting that individuals with GAD and its psychiatric comorbidities should be closely monitored for the development of suicidal ideation. However, we observed no additive comorbid effect of OCD, ASD, or ADHD on suicide among patients with GAD, a finding attributable to the small sample sizes of the groups with comorbid ASD and ADHD and the overlapping factor of anxiety between GAD and OCD.

Dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis may be the biological cause of the association between GAD and suicide [20, 21]. In this context, Lenze et al. analyzed the salivary cortisol levels of patients with GAD at six daily time points (waking, waking + 30 min, noon, afternoon, evening, and bedtime) and revealed that patients with GAD exhibited higher levels of both peak and total cortisol than the control group. Additionally, they observed a positive correlation between cortisol levels and GAD severity as measured by the GAD Severity Scale [22]. Furthermore, they suggested that 12-week treatment with escitalopram reduced salivary cortisol levels and improved anxiety symptoms [23]. Moreover, Vreeburg et al. reported that current anxiety disorders were associated with higher cortisol levels on awakening (p = 0.002) and also observed a trend toward higher morning cortisol levels (p = 0.08) in individuals whose anxiety disorders were in remission [24]. Additionally, a meta-analysis of 779 individuals who had attempted suicide and 1,447 individuals who had not attempted suicide uncovered a positive association between cortisol levels and suicide attempts, especially among individuals aged < 40 years [20]. Another study that evaluated the postmortem brains of 24 individuals who died by suicide and 24 normal controls discovered increased mRNA levels of corticotropin-releasing factor in the prefrontal cortex (PFC) and the central nucleus of the amygdala in those who died by suicide compared with the controls [25]. Finally, Pandey et al. demonstrated a substantial decrease in the protein and gene expression of glucocorticoid receptors in the PFC and amygdala of the postmortem brains of individuals who died by suicide [26].

This study has several limitations. First, we may have underestimated the prevalence of GAD in the Taiwanese population because only those who sought medical consultation and help were identified in the NHIRD. However, all GAD diagnoses were made by board-certified psychiatrists, improving the diagnostic validity in those individuals we studied. Second, time-dependent Cox regression models with adjustment for demographic data, psychiatric comorbidities and CCI were used to examine suicide risk between groups in our study. The analyses of dying from other causes as a competing risk were not performed in the present study because suicide always occurred much earlier than other mortality causes, especially natural death. Third, the NHIRD lacks information on environmental factors, psychosocial stress, lifestyle, and other variables that could influence the study’s outcomes. Therefore, we could not comprehensively investigate the contributions of these factors to suicide risk.

In conclusion, the present study observed that, compared with individuals who did not have GAD, patients with GAD were more likely to die by suicide during the follow-up period (2003–2017) after adjusting for psychiatric comorbidities, specifically schizophrenia, bipolar disorder, major depressive disorder, neurodevelopmental disorders, AUD, and SUD. GAD was individually significantly associated with the risk of suicide, and having additional psychiatric comorbidities with GAD increased this risk. Identifying GAD comorbidity is clinically relevant and important for suicide prevention in patients with severe mental disorders. To mitigate the risk of suicide in individuals with GAD, suicide prevention strategies targeting individuals with GAD and related psychiatric comorbidities should be developed and implemented. Further studies are required to clarify the pathomechanisms underlying anxiety and suicide.