Risks of Completed Suicide of Community Individuals with ICD-10 Disorders Across Age Groups: A Nationwide Population-Based Nested Case-Control Study in South Korea
Article information
Abstract
Objective
Suicide is the leading cause of death in 10–39-year-olds in South Korea, and the second highest rate among the OECD countries. However, few studies have investigated the particularity of completed suicide in South Korea.
Methods
Study subjects consisted of 2,838 suicide cases and 56,758 age and sex matched living controls from a national representative sample of 1,025,340 South Koreans. They were obtained from the Korean National Health Insurance Service-National Sample Cohort (NHIS-NSC) with follow-up up to 12 years. We obtained information on primary diagnosis of any ICD-10 disorder along with suicide cases during their lifetime.
Results
Among ICD-10 disorders, depression was the most common disorder (19.10%, n=542), found in victims of completed suicides except for common medical disorders such as hypertensive crisis, respiratory tract infection or arthropathies. After adjusting for sex, age, economic status, disability, and disorders, schizophrenia showed the strongest association with suicide (AOR: 28.56, 95% CI: 19.58–41.66) among all ICD-10 disorders, followed by psoriasis, multiple body injury, epilepsy, sleep disorder, depression, and bipolar disorder. For age groups, ≤19 years was associated with anxiety disorder (AOR=80.65, 95% CI: 13.33–487.93), 20–34 years with epilepsy (AOR=134.92, 95% CI: 33.69–540.37), both 35–49 years (AOR=108.57, 95% CI: 37.17–317.09) and 50–65 years (AOR=189.41 95% CI: 26.59–1349.31), with schizophrenia, and >65 years (AOR=44.7, 95% CI: 8.93–223.63) with psoriasis.
Conclusion
Psychiatric and physical disorders carried greatly increased risks and numbers of suicides in South Korea. Schizophrenia was the strongest risk factor, especially 35–65 years, and depression was the most common in suicide victims among ICD-10 disorders in South Korea.
INTRODUCTION
Suicide is a significant cause of death in many countries. The number of suicide deaths worldwide in 2015 was estimated to be 788,000 and the annual global age-standardized suicide rate was 10.7 per 100,000 population [1]. The estimated total suicide mortality in South Korea was 24.3 per 100,000 in 2017 [2], which was 2.3 times higher than the global rate. South Korea had the second highest suicide mortality rate among OECD countries after Lithuania in 2017 [3]. Suicide was the leading cause of death in individuals between 10 and 39 years, with individuals in their eighties having the highest suicide rate at 70.0 per 100,000 in 2017 among all age groups [4]. The Korean government has created policies to reduce suicide rates such as prohibition of paraquat in 2011 [5], regulation of media reports about suicide to reduce copycat suicides of celebrities [6], and legislation of “Suicide Prevention and Life Respect Culture” Act in 2011. After legislation, suicide rate was decreased from 31.7 per 100,000 in 2011 to 28.1 per 100,000 in 2012. However, high suicide rates in South Korea have continued.
In an important review paper on risk factors for suicide, Hawton and Kees ven Herringen [7] have reported different distal and proximal factors. They stressed that both psychiatric and physical disorders were proximal risk factors for suicide. They were particularly associated with poor physical health, disabilities, and pain [8]. Many previous studies have also suggested that suicide is related to physical disorders and psychiatric disorders [9-11]. Previous research has shown an association between suicide and chronic physical disorders such as asthma, allergy, type I diabetes mellitus, epilepsy, coronary heart disease, stroke, and cancer [8,9,12-17]. However, few previous studies have examined the relative strength of association between psychiatric/physical disorders and completed suicide using a nationwide representative sample.
We hypothesized that there would be large differences in the relative strength of associations of each psychiatric/physical disorder and suicide after adjusting for sex, age, economic status, and disability. To investigate how specific physical and psychiatric disorders contributed to suicide in Korea and associations of suicide with gender and five age groups, we used a representative sample of one million South Koreans from National Health Insurance data known as the National Health Insurance Service-National Sample Cohort (NHIS-NSC) collected during 2002–2013 [18]. All 52 million population of South Korea were entered into the national health insurance service and followed by the national registry. Using the NHIS-NSC cohort, we were able to compare differences in odds ratios associated with specific physical and psychiatric disorders.
METHODS
A nationwide population-based cohort and study population
This study was performed by the Korea Psychological Autopsy Center (KPAC), an affiliate of the Korean Ministry of Health and Welfare operated by the Sungkyunkwan University Research and Business Foundation, and School of Public Health, Seoul National University in collaboration with Health Insurance Review & Assessment Service (HIRA) and Statistics Korea. As the Korean health insurance system is a universal health insurance system provided by the government, every citizen is obligatorily inducted into the Korean National Health Insurance in their lifetime without exception. Their medical records are handled by the HIRA and the National Health Insurance Service (NHIS).
A nested case-control study design was constructed using the National Health Insurance Service-National Sample Cohort (NHIS-NSC) which consisted of representative 1,025,340 individuals (2.21%) sampled from a total of 46,405,433 eligible citizens of South Korea. As the universal insurance provider in South Korea, the NHIS developed National Health Insurance Database (NHID) based on claims data containing all citizens’ information on demographics such as sex, age [in 5-year units], economic status, disability, and healthcare utilization [date of hospital visits/admissions, primary and accessory diagnosis, type of medical department, and drugs/treatments administered]. The NHIS-NSC was established based on the 2002 NHID. The NHIS-NSC was randomly selected using systematic stratified random sampling according to sex, age (18 age groups), and economic status from eligible population in the 2002 NHID. The cohort was followed until 2013 (up to 12 years).
The income variable in the NHIS-NSC cohort was based on the insurance premium of insured employees/insured selfemployed individuals. The NHIS imposes a premium according to income levels (0 for medical aid beneficiaries and an insurance premium decile for the insured). To adjust for economic status, we recategorized the income variable as having 4 categories which consisted of level 1 (beneficiaries: lowest income), level 2 (1st to 4th deciles), level 3 (5th to 7th deciles), and level 4 (8th to 10th deciles: highest income).
The NHIS was authorized to link the NHIS-NSC with the Death Statistics Database based on data of the Korean Statistical Information Service (KOSIS). The KOSIS defines “suicide” status of each dead person through two kinds of statistics. One is cause of death statistics resulting from the integration of national statistics and the other is intentional self-harm among all cases of unnatural death investigations by the National Police Agency which include their Coroner’s findings from the National Forensic Service. We obtained linked data and included all individuals who died by suicide from such data. Suicide cases were identified from causes of death codes which had ICD-10 (International Classification of Disease, 10th version) codes X60-X84 (Intentional self-harm). Using this definition, we identified 2,838 suicide cases (0.27%) from all ages during 2002–2013 from the NHIS-NSC. Each suicide case was matched by gender, age [in 5-year unit], and date of death to 20 controls who were alive until 2013 from representative subsamples in NHIS-NSC. This control section method was based on a previous study [8,9,12-17], which assessed the relationship between suicide and physical disorders. The selection of large numbers of controls enabled maximizing the power for accurate effect estimation for low prevalence disorders. Consequentially, 2,838 suicide cases and 56,758 sex- and age- matched living controls were included in the present study. This study was approved by the Institutional Review Board (IRB) of Seoul National University (No. E1711/003-014). Informed consent was waived by IRB because we obtained encrypted data for privacy.
Psychiatric and physical disorders as potential risk factors
Information on primary diagnosis was retrieved from claims of NHIS-NSC for both suicide cases and corresponding controls until the date of death. To comprehensively evaluate the effect of psychiatric and physical disorders on the risk of suicide, we examined all diseases associated with suicide cases during their lifetime among all ICD-10 disorders. As shown in Table 1, a total of 39 disorders were included in the analysis, including 9 psychiatric disorders (dementia, alcohol use disorder, schizophrenia, bipolar disorder, depression, anxiety disorder, adjustment disorder, somatoform disorder, and sleep disorder) and 30 physical disorders (cancer, benign tumor, iron deficiency anemia, disorder of thyroid gland, diabetes mellitus, other nutritional deficiencies, epilepsy, migraine, vascular disease, cataract, glaucoma, visual disturbances and blindness, conductive and sensorineural hearing loss, hypertensive disease, ischemic heart disease, cerebrovascular disease, acute upper respiratory infection, influenza and pneumonia, chronic lower respiratory disease, noninfective enteritis and colitis, other diseases of intestines, alcoholic liver disease, atopic dermatitis, psoriasis, arthropathies, renal failure, urolithiasis, multiple body injury, effects of foreign body, and burns and corrosions).
ICD-10 disorders associated with suicide cases during their lifetime
Statistical analysis
For descriptive analysis, we calculated the absolute number of suicide cases and the rate of suicide per disorder in the total study population (for which cases and controls were combined) and in subgroups stratified by sex and age (<19 years, 20–34 years, 35–49 years, 50–65 years, and >65 years). To determine whether the risk of suicide was increased with selected psychiatric and physical disorders, odds ratios (ORs) and two-tailed 95% confidence intervals (CIs) were estimated by using conditional logistic regression models for matched casecontrol pairs. Crude models were inherently adjusted for sex and age by the study design with each disorder as exposure. In adjusted models, we included economic status (four categories), disability, and all psychiatric and physical disorders. To examine whether the effect of the disorder on the risk of suicide differed by sex and age, the same analysis was conducted for subgroups stratified by sex and age group. All statistical analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC, USA).
RESULTS
Demographic profiles of suicide subjects and non-suicide subjects
Table 2 shows demographic characteristics (sex, age, economic status and disability) and clinical characteristics of the NHIS-NSC cohort by suicide cases and non-suicide controls from 2002–2013. Of 1,025,340 representative samples of Koreans, 2,838 (0.28%) died by suicide during 12 years. Schizophrenia (RR=14.70), sleep disorder (RR=11.93), depression (RR=9.27), bipolar disorder (RR=9.03), and anxiety disorder (RR=6.92) had significantly high relative ratio (RR) of completed suicide compared to non-suicide controls among psychiatric disorders. Among physical disorders which included cases of suicide, multiple body injury (RR=7.33) showed the highest relative ratio. As shown in Table 2, psychiatric disorder had higher effect on suicide than physical disorder in view of illness in South Korea.
Demographic and clinical characteristics of the NHIS-NSC cohort by suicide cases and age- and sex-matched non-suicidal living controls from 2002 to 2013 (N=1,025,340)
Relation between suicide and psychiatric/physical disorders
Table 3 shows odds ratios and adjusted odds ratios (AOR) of all ICD-10 disorders associated with suicide cases during their lifetime after adjusting for sex, age, economic status, disability, and all covariates listed. Among all ICD-10 disorders, schizophrenia showed the strongest association with suicide cases (OR=17.44, 95% CI: 14.06–21.63). Its association with suicide cases was much stronger after adjusting for all covariates listed (AOR=28.56, 95% CI: 19.58–41.66). AOR of schizophrenia for suicide cases was the highest, followed by psoriasis, multiple body injury, epilepsy, sleep disorder, depression, and bipolar disorder. AOR of schizophrenia was as much as two times higher than the AOR of psoriasis, the second highest disorder that showed strong association with suicide cases.
Risk of suicide by each ICD-10 disorder associated with suicide cases had during their lifetime compared to age- and sex-matched non-suicidal living controls in the NHIS-NSC cohort from 2002 to 2013 (N=1,025,340)
Relationship between suicide and sex differences
Table 4 shows sex differences between psychiatric/physical disorders and suicide. There were marked differences in adjusted odds ratio (AOR) for suicide between sex in both psychiatric and physical disorders. In males, schizophrenia showed the strongest association with suicide cases among all ICD-10 disorders after adjusting for age, economic status, disability, and all covariates listed. However, no male patient with schizophrenia was found in non-suicidal control. Psoriasis (AOR=27.83, 95% CI: 11.53–67.19) and bipolar disorder (AOR=24.83, 95% CI: 11.26–54.84) also showed stronger association with suicide cases than other ICD-10 disorders in males. In females, schizophrenia (AOR=22.17, 95% CI: 11.97–41.05) showed the second highest association with suicide cases. Multiple body injury had the highest association with suicide cases (AOR=33.67, 95% CI: 9.28–122.19) among all ICD-10 disorders. Epilepsy (AOR=19.76, 95% CI: 6.59–59.25) and sleep disorder (AOR=15.19, 95% CI: 10.01–23.04) also showed stronger associations with suicide cases than other ICD-10 disorders in females. Those results indicated that psychiatric and physical disorders related suicide were different according to sex. Females showed much higher suicide rates among individuals with schizophrenia than males (7.63% vs. 5.03%).
Gender differences in risks of suicide by each ICD-10 disorder in the NHIS-NSC cohort from 2002 to 2013 (N=1,025,340)
Relationship between suicide and age group
Table 5 shows age differences between psychiatric/physical disorders and suicide. After adjusting for sex, economic status, disability, and all covariates listed in each age group, there were marked differences in OR associated with suicide according to each age group. In the age group of less than 19 years old, anxiety disorder (AOR=80.65, 95% CI: 13.33– 487.93) showed the strongest association with suicide. Schizophrenia showed the highest association with suicide both in the age group of 35–49 years (AOR=108.57, 95% CI: 37.17–317.09) and the age group of 50–65 years (AOR=189.41, 95% CI: 26.59–1349.31). For the age group of 20–34 years, schizophrenia had the second highest association with suicide (AOR=56.74, 95% CI: 26.13–123.22) after epilepsy (AOR=134.92, 95% CI: 33.69–540.37). For the age group of more than 65 years, psoriasis (AOR=44.70, 95% CI: 8.93–223.63) showed the strongest association with suicide while alcoholic liver disease (AOR=21.29, 95% CI: 3.83–118.39) had the second strongest association. As shown in Table 5, schizophrenia showed high associations with the risk of suicide in for age groups of 20–34, 35–49, and 50–65 years. On the contrary, for age group of more than 65 years or less than 19 years, there was no significant association between schizophrenia and suicide. Those results indicated that associations of psychiatric/physical disorders with suicide were different among age groups in South Korea.
Age differences in risks of suicide by each ICD-10 disorder in the NHIS-NSC cohort from 2002 to 2013 (N=1,025,340)
DISCUSSION
To the best of our knowledge, this is the first study to compare odds ratios of suicide in all psychiatric and physical disorders across age groups from a large national representative cohort. The main findings from this study were: 1) suicide was strongly associated with medical disorders, especially schizophrenia in psychiatric disorders compared to non-suicide control; 2) sex had a significant effect on the relationship (odds ratio) between suicide risk and medical disorders; 3) the associations between suicide risk and medical disorders showed large differences across age groups.
Our results are consistent with previous studies showing that psychiatric disorders had the strongest association with suicide [9-11]. They were present in about 90% of completed suicides led by depression and schizophrenia [9,19,20]. Suicide is also associated with poor physical health such as epilepsy [16,21,22], cancer [23-25], and pain-related illnesses [7].
Although the present study also showed associations between psychiatric/physical disorders and completed suicide, the unique finding of the present study was that schizophrenia had the highest odds ratios for completed suicide among psychiatric and physical disorders in South Korea. Up to date, meta-analysis results of suicide risk in schizophrenia have changed. According to earlier literatures [26-28], the prevalence of suicide in patients affected by schizophrenia was approximately 10–13%. However, recent studies [29] showed that average lifetime suicide rate in schizophrenia was about 4.9%. Therefore, the main finding that suicide rate was 5.03% in males and 7.63% in females in the present was meaningful in that such rates were above average of 4.9%.
Another finding of this study was that suicide rates showed great differences across sex and age groups. For sex, males were associated with depressions, bipolar disorders, and psoriasis while females were associated with schizophrenia, sleep disorder, epilepsy, and alcoholic liver disease. The middle age group (35–65 years) was more strongly associated with psychiatric disorders such as schizophrenia, depressions, anxiety disorders, alcohol use disorders, and sleep disorders whereas the younger age group (20–34 years) was not only associated with psychiatric disorders, but also strongly with physical disorders such as epilepsy and cancer. The elderly group (>65 years) was associated with psoriasis and alcoholic liver disease while children and adolescent age group (<19 years) was associated with anxiety disorders. These findings suggest there are psychiatric comorbidities in patients with physical disorders in both young and old age groups. Previous studies have reported that these physical disorders show risks of comorbid depression and suicide attempt or completed suicide in young age with epilepsy [16,21,22,30] and cancer [31] and in old age groups with psoriasis [32] and alcohol related problems [33].
According to these three main findings, schizophrenia is the strongest cause of suicide in South Korea. There are several reasons for this. First, treatment and follow-up were poorly managed for schizophrenia in South Korea. Thus, patients with schizophrenia could have high risk of exposure to suicide due to laxity of management. In most countries, depression or bipolar disorder is the highest and strongest cause of suicide based on evidences of many literatures [9,19,34]. However, our study showed a different result. Second, the diagnosis of psychiatric disorders such as depression was not successful in South Korea. Finally, it might be due to perception of specific psychiatric disorders such as schizophrenia in South Korea. Previous studies have indicated perceived cognitive dysfunction and stigmatization in the onset of suicidal ideation in people with schizophrenia [35]. Significant associations between regional suicide rates and the intention to seek informal help, self-stigma, and shame have been reported in previous studies [36]. Therefore, prejudice and misconception might have affected suicide risk in adults with schizophrenia belonging to age groups of 35–49 and 50–65 years in South Korea.
This study was based on the Korean National Health Insurance Service-National Sample Cohort (NHIS-NSC) from 2002 to 2013. It is the largest study examining the relationship between psychiatric/physical disorders and suicide in South Korea. A particular strength of this study was that it compared the risk of suicide in people with different psychiatric and physical disorders within the same total population through casecontrol research. This allowed comparison of adjusted odds ratios of individuals with and without psychiatric and physical disorders. Results of this investigation also enabled us to report that specific psychiatric/physical disorders could affect suicide risk associated with age (five specific age groups).
This study has several limitations. First, this study was conducted using major diagnoses. We were unable to be aware of dual diagnoses between psychiatric and physical disorders or dual diagnosis between psychiatric disorders such as sleep disorders, depressions, and so on. Sleep disorders might be correlated with depressions. Some serious physical disorders are often accompanied by pain, disability, limited social activity, and worry, leading to many problems with respect to physical, psychological, and social well-being [37,38]. Second, psychiatric disorders are underestimated when clinicians make a diagnosis. In Asian-culture, community population are likely to conceal their psychiatric problems. As a result, diagnosis of psychiatric disorders is easily underestimated in Korea. Third, it was impossible to differentiate diagnoses resulting from suicide from diagnoses of physical injuries such as ‘effects of foreign body’.
This study showed that suicide was associated with specific psychiatric and physical disorders such as schizophrenia, psoriasis, multiple body injury, epilepsy, sleep disorder, depression, bipolar disorder, and atopic dermatitis. This study also showed that there were considerable differences in the association between psychiatric/physical disorders and suicide risk depending on sex and age (five age groups). To prevent suicide, it is important to have national suicide prevention strategies to address at-risk populations suffering from each psychiatric and physical disorder based on sex and age group. The extremely high odds ratios of suicides in patients with these psychiatric disorders and physical disorders may suggest poor support systems and high stigmatization for such disorder in South Korea. Medical systems should deal with patients’ emotional traumas, low self-esteem, and comorbid psychiatric disorders after diagnosis and provide follow-up for each disorder according to age-specific environments.
Acknowledgements
This study was mainly supported by the Korea Psychological Autopsy Center (KPAC) (Director: Hong Jin Jeon, MD, PhD). The KPAC is operated by the Research & Business Foundation of Sungkyunkwan University on commission of the Korean Ministry of Health and Welfare from Jan 2016. This research was partially supported by the Original Technology Research Program for Brain Science through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (No. NRF-2016M 3C7A1947307; PI HJJ), and the Bio & Medical Technology Development Program of NRF funded by the Korean government, MSIP (No. NRF-2017 M3A9F1027323; PI HJJ). Statistical analyses of this study were supported by the Department of Public Health Science Graduate School of Public Health, Seoul National University (Dean: Ho Kim, PhD).
Notes
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: Hong Jin Jeon. Data curation: Eun Jin Na, Hyewon Lee, Woojae Myung, Hong Jin Jeon. Formal analysis: Hyewon Lee, Ho Kim. Funding acquisition: Hong Jin Jeon. Investigation: Eun Jin Na, Hyewon Lee, Woojae Myung, Kwan Woo Choi, Ho Kim, Hong Jin Jeon. Methodology: Hyewon Lee, Ho Kim. Project administration: Hong Jin Jeon. Resources: Hong Jin Jeon. Software: Hyewon Lee, Ho Kim. Supervision: Maurizio Fava, David Mischoulon, Jong-Woo Paik, Jin pyo Hong, Ho Kim, Hong Jin Jeon. Validation: Hyewon Lee, Ho Kim. Visualization: Hyewon Lee, Ho Kim. Writing—original draft: Eun Jin Na, Hyewon Lee. Writing— review & editing: Eun Jin Na, Hyewon Lee, Woojae Myung, Maurizio Fava, David Mischoulon, Jong-Woo Paik, Jin pyo Hong, Kwan Woo Choi, Ho Kim, Hong Jin Jeon.