Mortality of Suicide and Cerebro-Cardiovascular Diseases by Occupation in Korea, 1997–2020

Although studies on occupational mortality have been conducted in Korea, the results for occupations with high mortality around 2010 are inconsistent. This study aimed to examine occupational mortality from overwork-related suicide and cerebro-cardiovascular diseases (CCVD) from 1997 to 2020. We used microdata of the Causes of Death Statistics (CDS) and Economically Active Population Survey (EAPS) to obtain indirect standardized mortality ratio (SMR) and standardized proportional mortality ratio (PMR) of suicide (X60–X84) and CCVD deaths (I20–I25 and I60–I69) by gender and eight occupational categories. The trend of SMR of suicide and CCVD by occupation was similar within individual genders. The SMR of managers (MNG) was the highest for men and women in 2012–2017 and 2008–2020, respectively, whereas the SMR of professionals and related workers (PRF) was consistently low. Despite the similar socioeconomic status of MNG and PRF, we suggest that their mortality should be analyzed separately in Korea. SMR of suicide and CCVD in female MNG were consistently highest, although the PMR was low. Female MNG may have been more directly affected by the economic crisis. There is a need for work-related stress management, early intervention, and prevention policies in occupations vulnerable to mortality.


Introduction
Suicide and cerebro-cardiovascular diseases (CCVD) are the leading cause of death globally and are known to be influenced by a complex interaction between various factors [1,2]. In particular, suicide is not only influenced by economic problems, health problems, and personally stressful environments but also work-related stresses [1]. Some East Asian countries, such as Taiwan, Japan, and South Korea, consider suicide and CCVD overwork-related diseases and provide compensation for death [3][4][5][6]. Death by CCVD and suicide have been found to be related to long working hours and various work-related stresses in Korea [7][8][9].
Mortality from overwork-related diseases has been reported to vary by occupation, which is an important factor affecting health [10]. In general, occupational mortality was closely related to socioeconomic status. Although the classification of occupation differed by study, in most Western countries, suicide and CCVD mortality rates continued to be higher in unskilled occupations with low socioeconomic status than for managers and professionals with high socioeconomic status [11][12][13][14][15][16]. This may be the effect of social inequality in various areas such as income, education, and medical service [17]. However, Japan is known to be a representative country with higher mortality for upper non-manual workers (managers and professionals) after the economic bubble burst [18,19], whereas the results for Korea are conflicting. In recent years, studies have reported that the mortality

Ethics Statement
Both CDS and EAPS provided microdata without personal identification information. This study was approved by the Institutional Review Board at Hanyang University, Seoul, Republic of Korea (HYUIRB-202012-005-2).

Statistical Analysis
SMR can be calculated even when the number of deaths is small and variance is relatively low [23]. In general, the economically active population (including employed and unemployed people) is used as the reference population. However, as mentioned above, in order to avoid the effect of the death of the unemployed [16,18], only the deaths among employed people were considered the target population. Since SMR may have denominator-numerator bias due to unlinked data to the reference population and deaths, PMR was used to overcome this bias.
SMR is calculated by comparing the actual number of suicide and CCVD deaths with the expected number of suicide and CCVD deaths in the reference population. PMR is calculated by comparing suicide and CCVD deaths among total deaths in the study group with suicide and CCVD deaths among total deaths in the reference group. We stratified the age by five years for adjustment and analyzed the data by gender and occupation with estimated 95% confidence intervals (95% CI). All statistical procedures for SMR and PMR were performed using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA) and R software version 4.2.0 [24].

Ethics Statement
Both CDS and EAPS provided microdata without personal identification information. This study was approved by the Institutional Review Board at Hanyang University, Seoul, Republic of Korea (HYUIRB-202012-005-2).

Statistical Analysis
SMR can be calculated even when the number of deaths is small and variance is relatively low [23]. In general, the economically active population (including employed and unemployed people) is used as the reference population. However, as mentioned above, in order to avoid the effect of the death of the unemployed [16,18], only the deaths among employed people were considered the target population. Since SMR may have denominator-numerator bias due to unlinked data to the reference population and deaths, PMR was used to overcome this bias.
SMR is calculated by comparing the actual number of suicide and CCVD deaths with the expected number of suicide and CCVD deaths in the reference population. PMR is calculated by comparing suicide and CCVD deaths among total deaths in the study group with suicide and CCVD deaths among total deaths in the reference group. We stratified the age by five years for adjustment and analyzed the data by gender and occupation with estimated 95% confidence intervals (95% CI). All statistical procedures for SMR and PMR were performed using SAS software version 9.4 (SAS Institute Inc., Cary, North Carolina, USA) and R software version 4.2.0 [24].    The CCVD SMR of females was the highest for SFF in 1997-2008, followed by CLK, whereas the PMR of both occupations was low or insignificant (Table 4). After 2009-2011, the CCVD SMR of female MNG rose dramatically. In particular, the SMR of MNG was the highest at 7.93 (95% CI 5.89-9.97) in 2015-2017. Conversely, the PMR of female MNG was low or not significant for 24 years.

Results
Over the 24-year period, SMR of suicide and CCVD were highest in SFF and MNG among males and females, respectively (Appendix A). Appendix B displays the result of analyzing the SMR of male suicide and CCVD in 2007-2018 by period and four occupation groups, simila r to the recent linked data study [20]. As a result of sorting them into four occupational groups, the SMR of upper non-manual workers (MNG and PRF) was lower than that of other groups. However, as a result of analyzing MNG and PRF separately in eight groups, the SMR of MNG increased to first (CCVD) and second (suicide).

Discussion
The mortality ratios of suicide and CCVD were identified according to eight occupational categories and gender in 1997-2020. Interestingly, the suicide and CCVD mortality ratios by occupation showed similar trends within individual genders, with marked changes observed in 2012 for men and 2008 for women. In men, SFF in suicide and CLK in CCVD had the highest SMR in 1997-2011. Both suicide and CCVD SMRs were the highest for MNG in 2012-2017. Additionally, for both suicide and CCVD in 2018-2020, LMT had the highest SMR but lower PMR. Suicide and CCVD SMR of SFF were highest among women in 2000-2008. Both female MNG's suicide and CCVD had the highest SMR in 2009-2020, but the PMR was low or not significant. Japanese studies also showed similar patterns of cardiovascular disease, suicide, and cancer mortality in the same occupation [18,19]. Large prospective studies found that depression and psychological distress were common risk factors for an increase in all-cause mortality, including cardiovascular disease and suicide mortality [25,26]. CCVD may induce suicide by various psychopathological means [27]. However, studies on the relationship between overwork-related diseases are scarce. The CCVD SMR of CLK was the highest in 1997-2011 but only slightly higher in suicide for the same period. CLK, a typically sedentary job, is known to have a higher risk of CCVD [28].
Suicide and CCVD mortality studies in various European countries [11,15,16], the United States [13], Canada [14], and New Zealand [12] reported that MNG and PRF had the lowest risk of suicide, but manual workers such as LMT, QMS, and SFF had a high risk. Japanese studies, however, found that workers with high socioeconomic status, including MNG and PRF, had higher mortality risks of ischemic heart disease, cerebrovascular disease, and suicide after the economic bubble burst [18,19,29]. In contrast, the results of studies on occupational mortality in Korea are inconsistent. Some studies using unlinked data showed that male mortality of upper non-manual workers (MNG and PRF) turned out to be higher than that of manual workers in the mid-2010s, unlike in the late 1990s [9,19], whereas others reported that the mortality of manual workers remained the highest after the late 2000s [20,21].
It has been argued that one of the reasons for this discrepancy in the results may be due to differences in the data used for mortality analysis [20]. Numerator-denominator bias occurs when death and census data are not linked, which is known to have the potential to influence mortality estimates [30]. However, the effect and direction of the numeratordenominator bias are still controversial. Similar results were found with linked data from 1995 to 2008 [31] and unlinked data from 1995 to 1999 [19], with the lowest mortality for MNG and PRF. Additionally, studies using the same unlinked data reported different results in MNG suicide mortality [9,21].
Another reason may be that MNG and PRF were not analyzed separately in previous studies. A previous Korean study reported a consistently lower suicide mortality rate for MNG during 1993-2016 [21]. However, this result may be because MNG and PRF were included in the same group. The present findings show that the SMR of MNG has rapidly increased since around 2010, whereas the SMR of PRF has been consistently low. To clarify this, we compared the SMR of eight occupational groups by sorting them into four occupational groups, similar to a recent linked data study [20]. Upper non-manual workers (MNG and PRF) had lower SMR of suicide and CCVD than other groups. However, as a result of separately analyzing MNG and PRF, the SMR of MNG was the second highest in suicide and the first in CCVD (Appendix B).
In Europe, the increase in mortality associated with occupations of low socioeconomic status appears to have been directly affected by socioeconomic inequality [17,19]. However, socioeconomic status seems to be insufficient to explain the mortality difference by occupation in Korea after 2010. The economic crisis in Korea may have increased the work-related stress of middle managers, which can influence high suicide mortality [32]. MNG who have died by suicide in Korea experienced various work-related stresses, mainly related to excessive responsibilities compared to other occupations [7,8]. MNG are divided into senior management positions representing institutions or companies and management positions including middle managers within the companies [33]. In Korea, there was a significant decrease in the number of jobs centered on middle management positions between 2008 and 2015, which indicates that restructuring due to the economic crisis may have mainly impacted the middle management positions [34]. A recent study comparing mortality inequality by occupation in Europe, Japan, and Korea speculated that the working conditions and social culture in Korea and Japan that cause overwork of MNG are different from those in Europe [19].
Most European countries have reported that the 2008 economic crisis had a negative impact on health, but the results for gender vary by country [35,36]. While the United Kingdom did not find a direct link between the economic crisis and women's suicide [35], Greece had a high suicide rate among women in management positions in 2000-2009 [11]. A previous Korean study has reported that women are less affected by the economic downturn than men in Korea, with a decrease in the suicide rate among women and an increase in the suicide rate among men [21]. However, our study results show that SMR of female suicide and CCVD in MNG increased sharply after the 2008 global economic crisis. Suicide was highest in the 1997 economic crisis as well. The high SMR of female MNG may indicate that national economic difficulties and social changes affect female MNG more directly than females in other occupations in the Korean labor market. The promotion of Korean women to managerial positions is rare, despite their having a high level of education similar to that of men [37]. The high mortality rate of female MNG may be because of overwork, increased responsibilities, preferential dismissal, decreased employment rate, and decreased managerial population due to the economic crisis [34,38]. Therefore, for female MNG with the highest mortality risk, a fundamental and long-term policy approach is essential for resolving occupational gender discrimination and preventing overwork. It is a part that requires a lot of research and social attention for preventive intervention.
Although the SMR of MNG increased rapidly for women after 2008 and for men after 2012, even though SFF withdrew from the top spot, the SMR of SFF is still high in suicide and CCVD [19,20,31]. SFF is also known as an occupation most affected by the economic crisis, with higher mortality [11,19,39]. The causes are presumed to be overwork due to a continuous decline in the labor force, financial difficulties due to changes in the industrial structure, and difficulties in accessing medical services [19,40,41]. Additionally, work environments with physical risk factors are known to be more stressful for female workers than for males, which can lead to mortality risk factors [29].
The SMR of MNG and SFF was higher than that of other occupations and significant, but the PMR was low or insignificant, which was more pronounced in women than in men. This may be an effect of the decreasing population of the relevant occupation. Suicide, CCVD deaths, and total deaths in female MNG surged from 2009 to 2017, while the population of female MNG continued to decline from 2012 to 2017. SMRs of suicide and CCVD for female MNG were noticeably higher than for males. However, the PMR was low or not significant, which may have been overestimated, mainly due to the relatively small number of female MNG in the population [29]. Similarly, male SFF's suicide and CCVD deaths decreased, and their population is also rapidly decreasing. Although both suicide and CCVD had high SMR, the PMR was low or insignificant for SFF.
This study had several limitations. First, the results are not representative of suicide deaths for the entire working-age population in Korea. To compare the mortality by occupation, we analyzed the data of employed people only. The Armed Forces and other unclassified people among the employed population, unemployed people among the economically active population, and the economically inactive population were excluded from the study. Second, although the linkage table was followed for reclassification from the old version to the 7th version of KSCO by the Korean Statistical Classification of Statistics Korea [33], some reclassification of occupations may not be accurate. Third, changes in the CDS regarding recording the occupation of the deceased may have affected the change in mortality by occupation over a 24-year period. Around the year 2000, the method of recording the occupation was changed from "occupation before death" to "occupation at the time of illness or accident". However, according to the present results, the pattern of mortality by occupation did not change significantly around 2000. Moreover, after 2018, the CDS used administrative data to determine the deceased's occupation, as this information was excluded from the death certificate. Although these changes seem to have influenced the mortality trend by occupation after 2018, it is noteworthy that the mortality of female MNG continues to be the highest. Finally, SMR and PMR have a disadvantage in that they do not provide information on the causal relationship between occupation and mortality. Additionally, since SMR was calculated using unlinked data, there may be a numeratordenominator bias in mortality estimates, though the PMR was analyzed to compensate for this bias. In order to understand the causes of mortality by occupation, further research linking CDS with data that have variables that can affect suicide in economically active populations is needed.

Conclusions
This study investigates the trends and changes in occupational mortality over 24 years in Korea by analyzing the SMR and PMR of suicide and CCVD, which are types of overworkrelated deaths. After the global economic crisis, both male and female MNG had the highest mortality ratios for suicide and CCVD, whereas the SMR of PRF was consistently low. As MNG and PRF are known to have similarly high socioeconomic status, previous studies analyzed the mortality by sorting them into the same group. However, we found differences in mortality between MNG and PRF, which suggests that they should be analyzed separately. Female MNG had the highest SMR and low PMR compared to other occupations, which may be due to decreased population, increased responsibilities, and priority layoffs during the economic crisis. Occupational differences between suicide and CCVD mortality highlight the need for management of work-related stress, early treatment, and preventive intervention in the workplace, especially in high-mortality occupations.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analysis, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results. CCVD-cerebro-cardiovascular diseases; N-observed death; SMR-indirect standardized mortality ratio; PMR-standardized proportional mortality ratio; CI-95% confidence intervals.