Baseline characteristics
Table 1 shows the baseline characteristics of the participants. Among the 3,951,398 participants, 263,754 (6.7%) were identified as having a mental disorder. The proportion of females among those with any mental disorder was higher than that among those without (62.5 vs. 44.3%, respectively; P < 0.001). The mean age of those with any mental disorder was higher (56.5 years; SD, 13.6 years) than that of those without (46.6 years; SD, 13.8 years). Patients with mental disorders tended to smoke less, drink less alcohol, engage in more regular physical activities, and have more comorbid physical illnesses than those without mental disorders.
Incidence of outcome
The average follow-up period was 11.1 years (SD, 1.5 years). Among all eligible participants, 249,830 died, with an incidence rate of 5.72 per 1000 person-years. Among them, 12,290 died by suicide, with an incidence rate of 0.28 per 1000 person-years.
Risk of suicide death according to mental disorders
Table 2 shows the risk of death due to suicide based on mental disorders. Overall, suicide mortality rates were high among people with mental disorders. Among them, the mortality was highest among those with personality disorders (2.49 per 1000 person-years), followed by that among those with substance use disorder (2.21 per 1000 person-years), alcohol use disorder (2.20 per 1000 person-years), bipolar disorder (1.88 per 1000 person-years), and schizophrenia spectrum disorder (1.79 per 1000 person-years).
Cox proportional regression analyses showed that the risks of suicide were increased among those with personality disorder (adjusted hazard ratio [aHR], 7.69; 95% CI, 4.96–11.92), bipolar disorder (aHR, 6.05; 95% CI, 4.86–7.54), schizophrenia spectrum disorder (aHR, 5.91; 95% CI, 5.12–6.82), OCD (aHR, 4.66; 95% CI, 3.36–6.47), substance use disorder (aHR, 4.53; 95% CI, 3.94–5.19), alcohol use disorder (aHR, 4.43; 95% CI, 3.84–5.11), PTSD (aHR, 3.37; 95% CI, 1.69–6.72), depressive disorder (aHR, 2.98; 95% CI, 2.79–3.17), anxiety disorder (aHR 2.66; 95% CI, 2.40–2.94), insomnia (aHR, 2.62; 95% CI, 2.46–2.80), and eating disorder (aHR, 1.95, 95% CI, 1.01–3.74) compared to those without each disease after full adjustment (Fig. 1).
A Suicide. B All-cause death. HR hazard ratio, CI confidence interval. Adjusted for age, sex, low income, Charlson comorbidity index, smoking, alcohol drinking, and regular physical activity.
The results of the analyses including the other mental disorders are presented in Supplementary Table 1.
Subgroup analyses
Table 3 shows the results of subgroup analyses according to sex and age. There were significant interaction effects between mental disorders with sex and age among those with schizophrenia spectrum disorder, bipolar disorder, depressive disorder, alcohol use disorder, and substance use disorder; women and younger age groups showed a prominently increased risk of suicide.
Supplementary Table 2 shows the subgroup analyses according to the duration of the mental disorders. In patients with depression, insomnia, alcohol use, and substance use disorders, the risk of suicide tended to increase with the duration of illness. Bipolar disorder had the highest risk at <1 year. Schizophrenia spectrum disorder, OCD, and personality disorders showed a U-shaped association, with the lowest risk in individuals with a disease duration of 2–4 years.
Secondary analyses
Table 4 presents the risk of all-cause mortality for each mental disorder. After full adjustment, except for ADHD and PTSD, which showed non-significant results, the risk of all-cause death increased for all other mental disorders. Intellectual disability was associated with the highest risk (aHR, 3.89; 95% CI, 3.03–4.98). Schizophrenia spectrum disorder (aHR, 2.31; 95% CI, 2.21–2.43), alcohol use disorder (aHR 2.68; 95% CI, 2.57–2.80), substance use disorder (aHR, 2.62; 95% CI, 2.51–2.73), dementia (aHR, 2.36; 95% CI, 2.28–2.44), and personality disorder (aHR, 2.01; 95% CI, 1.67–2.43) showed that the risk of all-cause death increased more than twice compared to not having each disease (Fig. 1).
Subgroup analyses according to sex, age, and illness duration are presented in Supplementary Table 3 and Supplementary Table 4.
