Type 2 diabetes mellitus is a rapidly growing chronic disease worldwide and recognized as a major risk factor for cardiovascular disease (CVD) (1). According to the International Diabetes Foundation, the number of adults with type 2 diabetes mellitus worldwide is expected to reach approximately 783 million by 2045 (2). In Korea, the prevalence of type 2 diabetes has increased significantly from 7.6% in 2001 to 13.8% in 2018 and 13.9% in 2020 (3, 4). The increasing prevalence of type 2 diabetes has increased its economic burden worldwide, with costs projected to increase from USD 1.3 trillion in 2015 to USD2.1–2.5 trillion by 2030 (5). In Korea, the prevalence of type 2 diabetes in 2019 was 10.7%, contributing to an economic burden of USD 18.3 billion and a per capita cost of USD 4090 (6). This increase is largely driven by the steady increase in obesity and the adoption of westernized dietary habits (7).
Diabetics are particularly susceptible to the adverse effects of lifestyle and socioeconomic status-related factors as diabetes exacerbate their impact, further increasing susceptibility to CVD (8, 9). According to previous studies, CVD is influenced by various risk factors, including physical factor such as age, sex, genetic predisposition, as well as lifestyle factors, such as physical inactivity, smoking, and alcohol consumption (10–12). Furthermore, environmental and socioeconomic factors, including air pollution, noise exposure, work environment, and income, are also associated with heightened CVD risk (13, 14).
According to a previous study, the risk of myocardial infarction increases in the general population as the cumulative unemployment period lengthens (15). This association is largely attributed to chronic stress linked to prolonged unemployment, which over activates the hypothalamic-pituitary-adrenal (HPA) axis, disrupts neuroendocrine pathways, and induces chronic inflammation (16, 17). Chronic stress, characterized by prolonged cortisol elevation, also worsens insulin resistance, making diabetes management more challenging and accelerating disease progression (18). However, previous studies have primarily focused on the general population, overlooking the heightened vulnerability of high-risk groups such as individuals with type 2 diabetes. Considering that diabetes is a major risk factor for CVD, the added stress of job insecurity may further increase CVD risk among these individuals. To address this research gap, our study focuses on middle-aged paid workers with type 2 diabetes to investigate the combined impact of cumulative unemployment on the risk of CVD.
Method
Data source and study population
This study used data from the National Health Insurance Service (NHIS), which covers over 97% of the South Korean population (19). The NHIS database includes detailed demographic and socioeconomic information, such as employment status and industry sector, collected through structured lifestyle questionnaires. Additionally, the NHIS conducts biennial health examinations for adults, including various clinical and biochemical tests (20). This database also provides comprehensive records of outpatient visits, hospitalizations, diagnostic codes, medical procedures, and prescribed medications. The study population comprised workers aged 40–49 years and 50–59 years as of the index date (January 1, 2011) who had participated in a national health examination in 2009–2010. The index date was January 2011, with the participants’ employment status tracked in 2012–2016 to determine cumulative unemployment durations. Patients with type 2 diabetes were identified through prescriptions for antidiabetic medications associated with International Classification of Diseases, 10th revision (ICD-10) codes E11–E14 (21).
The exclusion criteria were as follows: (i) missing information on socio-economic status, chronic disease factors, or lifestyle behaviors; (ii) participants employed in industrial sectors ‘T’ (activities of households as employers; undifferentiated goods- and services-producing activities of households for own use) or ‘U’ (activities of extraterritorial organizations and bodies); (iii) a previous history of CVD diagnosis or hospitalization due to CVD; (iv) occurrence of all-cause mortality or CVD within 1 year from the index date; (iv) implausibly long follow-up intervals; and (vi) missing employment insurance records in 2013–2016 (supplementary material, www.sjweh.fi/article/4272, table S1).
The Institutional Review Board (IRB) of Severance Hospital approved this study (IRB number: 4-2024-0615), which adhered to the ethical principles of the Declarations of Helsinki and Istanbul. The need for informed consent was waived owing to the retrospective nature of this study.
Main outcomes and secondary outcomes
The primary outcome of this study was the incidence of CVD, defined as ≥3 hospitalizations and outpatient visits combined, associated with ICD-10 codes I21–I23, I50, or I63–I64 as identified through insurance claims data (22). Secondary outcomes included all-cause mortality, ischemic heart disease (IHD), ICD-10 codes I21–I23, heart failure (HF), ICD-10 code I50, and stroke, ICD-10 codes I63–I64,. The index date was 1 January 2011. Participants were followed until the occurrence of CVD, all-cause mortality, or 31 December 2023, whichever occurred first.
Independent variable and covariates
Job insecurity, an independent variable, was defined as the cumulative unemployment period from 2011 to 2015, categorized into three groups based on employment status, which is derived from insurance type. The “stable” group comprised individuals who experienced no or <1 year of unemployment over the 5-year period. The “partially stable” group included individuals with 1–2 years of unemployment, while the “unstable” group included those unemployed for ≥3 years within the same period.
The analysis included the following covariates: age, household income quartile, residential area (Seoul, metropolitan, other), hypertension, dyslipidemia, duration of type 2 diabetes drug prescription, number of oral antidiabetic drugs taken last, type 2 diabetes complications, uncontrolled fasting blood sugar (FBS), fatty liver index (FLI), smoking history, alcohol consumption, and physical activity. Baseline covariates were assessed using insurance claims data and lifestyle questionnaires from a two-year look-back period prior to the index date 1 January 2011. Industrial clusters were classified into three industrial sectors based on the Korean Standard Industrial Classification (KSIC) and prior research: blue-collar (categories A-F, H), service (I, M, N, Q, R), and white-collar (G, J, K, L, O, P, S) (23) (supplementary table S1). Hypertension was defined as either a prescription for antihypertensive drugs according to ICD-10 codes I10–13, I15, or a systolic blood pressure (BP) of ≥140 mmHg or ≥90 mmHg (24). Dyslipidemia was identified through prescriptions for anti-dyslipidemia drugs corresponding to ICD-10 code E78 or total cholesterol ≥240 mg/dL (25). The duration of type 2 diabetes was categorized into <1 and ≥1 year. Oral antidiabetic drug use was categorized into four groups: 1, 2, >3 classes and insulin therapy. Type 2 diabetes-related complications were assessed based on the specificity of the ICD-10 codes related to type 2 diabetes (26). Uncontrolled FBS level was defined as ≥126 mg/dL. The FLI was categorized based on the standard threshold of FLI ≥30 (27). Individuals were classified as non-, ex-, and current smokers based on their smoking status using lifestyle questionnaires. Alcohol consumption was assessed by weekly intake levels, categorized as mild (<210 g for males and <140 g for females), moderate (<420 g for males and <350 g for females), and severe (≥420 g for males and ≥350 g for females) (28). The metabolic equivalent of the task (MET-minutes/week) for each participant’s physical activity was calculated based on their history of vigorous activity (7 MET), moderate activity (4 MET), and walking (2.9 MET) (29). Physical activity levels were further categorized into two groups: <600 MET-minutes per week and ≥600 or more MET-minutes per week (30).
Statistical analysis
All the analyses were stratified by sex. Baseline characteristics were summarized using counts and percentages for categorical variables and medians with interquartile ranges (IQR) for continuous variables. Cumulative hazard curves for CVD by job insecurity groups were estimated using Cox proportional hazards models, with differences among groups assessed via log-rank tests. Differences in job insecurity experiences were examined using log-rank tests. Adjusted hazard ratios (HRadj) and 95% confidence intervals (CI) for the primary and secondary outcomes were estimated using multivariate Cox proportional hazards models. To validate the results, analyses included a crude model (unadjusted) and a final model adjusted for age, residential area, household income, industry sector, chronic disease-related factors, and lifestyle factors.
We conducted several sensitivity analyses. First, stratified analyses were conducted based on household age and income to assess the risk of CVD associated with job insecurity. Second, smoking intensity was further adjusted by incorporating smoking pack-years, calculated as the product of the number of cigarette packs smoked per day and the duration of smoking in years.
All statistical analyses were performed using SAS software version 8.2 (SAS Institute Inc, Cary, NC, USA) and R software version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria). Results were considered statistically significant at a two-sided P<0.05.
Results
Following the recruitment of 146 175 individuals, 128 704 participants were included in the final analysis after applying the exclusion criteria (Supplementary figure S1). Table 1 presents the baseline characteristics of male participants by job insecurity, and table 2 presents those of female participants by job insecurity. Of these, 107 071 were males (83.2%) and 21 633 were females (16.8%). Among males, 68.6% (N=73 446) were in the stable group, 19.8% (N=21 163) in the partially stable group, and 11.6% (N=12 462) in the unstable group. Among females, 55.8% (N=12 065) were in the stable group, 24.9% (N=5388) in the partially stable group, and 19.3% (N=4180) in the unstable group.
Table 1
Baseline characteristics of male workers with type 2 diabetes by job insecurity status. [FBS=fasting blood sugar; GTP=glutamyl transpeptidase; IQR=interquartile range; MET=metabolic equivalent of the task]
Table 2
Baseline characteristics of female workers with type 2 diabetes, by job insecurity status. [FBS=fasting blood sugar; GTP=glutamyl transpeptidase; IQR=interquartile range; MET=metabolic equivalent of the task]
Age was calculated as of the index date (1 January 2011). The median age was 51 years (IQR: 46–55 years) for males and 52 years (IQR: 47–56 years) for females. Stable workers were somewhat younger than those in the partially stable and unstable groups. For males, the unstable group had a higher prevalence of low household income, residence in non-metropolitan areas, blue-collar occupations, hypertension, absence of dyslipidemia, over one year of oral antidiabetic medication use, use of a single type of diabetes medication, absence of diabetes complications, status as current smoker, moderate alcohol consumption, and insufficient physical activity (<600 minutes/week). For females, the unstable group had a higher proportion with low household income, blue-collar employment, no diabetes complications, FLI scores <30, and non-smokers.
During a mean follow-up period of 12.6 years, 6535 males (6.1%) and 848 females (3.9%) experienced incident CVD. In the unstable group, 990 males (7.9%) and 222 females (5.3%) experienced CVD, compared with 1464 males (6.9%) and 204 females (3.8%) in the partially stable group, and 4081 males (5.6%) and 422 females (3.5%) in the stable group. The 5-year cumulative incidence of CVD after the index date (January 2011) was 3.8%, 4.6%, and 5.2% for the stable, partially stable, and unstable groups in males; and 2.2%, 2.5%, and 3.4% in females, respectively. The differences across employment stability groups were statistically significant in both men (P=2.0 × 10−16) and women (P=5.0 × 10−7) (figure 1).
Table 3 shows the results of the univariate and multivariate Cox regression models that assessed the association between job insecurity and CVD. Among males, the partially stable and unstable groups showed significantly elevated risk of CVD compared with the stable group (HRadj 1.12, 95% CI 1.05–1.19 and HRadj1.25, 95% CI 1.16–1.34, respectively). Among females, only the unstable group showed a significantly increased risk (HRadj 1.33, 95% CI 1.13–1.57), while the partially stable group did not differ significantly from the stable group (HRadj aHR 1.00, 95% CI 0.85–1.19) (table 3).
Table 3
Adjusted hazard ratios (HR) and 95% confidence intervals (CI) of cardiovascular diseases by employment status.
a Rate were expressed per 100,000 person–years. b Final model was adjusted for adjusted for age, residential area, household income, Industrial Sector, hypertension, dyslipidemia, duration of diabetes, last class of oral antidiabetic drugs, uncontrolled fasting blood sugar, fatty liver index, smoking status, alcohol consumption, and physical activity. c CVD was defined as ≥3 combined hospitalizations or outpatient visits associated with ICD-10 codes I21–I23, I50, or I63–I64.
For both males and females, the risk of all-cause mortality significantly increased as job insecurity increased. The HF showed a pattern similar to that of the primary outcome, with a significant increase in risk among males as job insecurity increased. For females, a significant association between job insecurity and HF was observed only in the unstable group. For IHD, the risk significantly increased in the unstable group for males, whereas no significant increase was observed among females. For stroke, the risk significantly increased with cumulative unstable employment status among males, whereas it remained non-significant among females (supplementary table S2).
According to the analyses stratified by household age and income (table 4), the association between job insecurity and the risk of CVD was consistent across all categories. A significant association between job insecurity and an increased risk of CVD was observed among males in the 50–59 years age group (HRadj 1.92, 95% CI 1.39–2.66) and in females in the 40–49 years age group (HRadj 1.24, 95% CI 1.15–1.35). A significant association between job insecurity and an increased risk of CVD was observed in males in the low-income group (HRadj 1.31, 95% CI 1.20–1.42) and in females in the high-income group (HRadj 1.46, 95% CI 1.14–1.87).
Table 4
The association between job insecurity and the risk of cardiovascular diseases a stratified by age and income. All models were adjusted for age, residential area, household income, industrial sector, hypertension, dyslipidemia, duration of diabetes, oral antidiabetic drug, uncontrolled fasting blood sugar, fatty liver index, smoking status, alcohol consumption, and physical activity. [HR=hazard ratio; CI=confidence interval]
a CVD was defined as ≥3 combined hospitalizations or outpatient visits associated with ICD-10 codes I21–I23, I50, or I63–I64. b Rate were expressed per 100,000 person–years.
Supplementary table S3 presents that adjusting for smoking pack-years, alone or with smoking status, did not alter the association between job insecurity and CVD risk. After adjusting for smoking pack-years and smoking status, males in the partially stable (HRadj 1.11, 95% CI 1.05–1.18) and unstable groups (HRadj 1.24, 95% CI 1.15–1.33) had higher CVD risk than the stable group. Among females, only the unstable group showed increased risk (HRadj 1.34, 95% CI 1.13–1.58).
Discussion
Our study identified a significant association between job insecurity, measured by cumulative unemployment duration, and the risk of CVD among patients with type 2 diabetes. The risk of CVD increased with greater job insecurity in both sexes, demonstrating a dose–response relationship, where higher levels of job insecurity were associated with a progressively greater risk. For males, significant associations were observed in both the partially stable and unstable groups, whereas for females, the association was significant only in the unstable group. Most secondary outcomes exhibited trends similar to those observed for CVD, showing increased risks with greater job insecurity.
One French study reported a significant association between the duration of past unemployment and the prevalence of non-fatal CVD events, such as myocardial infarction (ORadj 1.75, 95% CI 1.12–2.74), after adjusting for demographic, lifestyle, and employment-related factors (15). A previous study conducted in Italy demonstrated that unemployed males had significantly higher risks of all-cause mortality (ORadj 1.82, 95% CI 1.37–2.41) and coronary heart disease (ORadi 2.58, 95% CI 1.05–6.37), whereas no such associations were observed in females (all-cause mortality ORadj 1.29, 95% CI 0.87–1.92) (31). Our study revealed a significant positive association between prolonged unemployment duration and CVD incidence, in line with existing evidence. For all-cause mortality, risks were elevated in both partially stable and unstable groups for both sexes. For CVD, the risk was significant in both the partially stable and unstable groups for males, whereas it was significant only in the unstable group for females. According to social-role theory, these differences may arise from culturally defined gender roles, where men often face pressure as primary breadwinners and women navigate stress from balancing professional and caregiving responsibilities (32, 33). Such role-based expectations could explain the observed gender differences in responses to job insecurity.
In Anfossi et al's systematic review (34), six studies examined job insecurity as a risk factor for ischemic heart disease. While most studies showed a positive, but not statistically significant, association, the overall evidence from these general or occupationally heterogeneous populations was limited and inconsistent. In contrast, our investigation focused on individuals with type 2 diabetes—a population with elevated baseline cardiometabolic vulnerability. We found a robust and statistically significant association between job insecurity and adverse cardiovascular outcomes, suggesting that the detrimental effects of job insecurity on cardiovascular health may be amplified in high-risk clinical subgroups.
In age-stratified analyses, the highest risk was observed among males in their 50s in the unstable job group, while among females, the strongest association was found in the 40–49 age group. These age- and sex-specific patterns align with prior literature indicating that middle-aged men and women may face distinct stressors related to employment and social roles. Previous studies have noted that men in their 50s often experience increased financial pressures, while women in their 40s frequently manage both occupational and caregiving responsibilities, potentially contributing to differential vulnerability to job-related stress (35–37).
In a previous study (38), compared to low-income females, low-income males faced a higher risk of CVD owing to job insecurity. Similarly, in our study, males in the unstable employment group were particularly vulnerable and exhibited a higher risk of CVD in the low-income group. In contrast, females in the unstable employment group with high household income demonstrated a heightened risk of CVD. This disparity may stem from traditional gender roles. Males often shoulder financial responsibilities, leading to increased stress in low-income situations, while high-income females may face heightened stress from social expectations and professional demands (32, 33, 39). These findings underscore the importance of considering both socioeconomic and gender-specific factors when addressing the health impacts of job insecurity.
This study had several strengths. First, to our knowledge, this is the first study to demonstrate an association between job insecurity and the risk of CVD among individuals with type 2 diabetes in Asian countries. Second, it used a representative cohort comprising over 97% of the general population of the Republic of Korea (19), with comprehensive adjustment for a wide array of potential confounders, including socioeconomic status, physical activity, medication use, and lifestyle habits. Third, the long follow-up period allowed for the evaluation of longitudinal associations between job insecurity and cardiovascular outcomes, thereby enhancing the reliability of the observed effects over time.
However, our study had some limitations. First, due to the limited data available in the NHIS database, certain unmeasured confounding factors, such as dietary habits and genetic predispositions, could not be included in our analysis. Second, as this was an observational study, a definitive causal relationship between job insecurity and CVD could not be established. Third, residual confounding may persist due to limitations in capturing the complexity of smoking behavior. Specifically, our adjustments for smoking status and smoking pack-years may not fully account for time-varying changes in smoking behavior or inaccuracies in self-reported smoking data, potentially affecting the precision of our risk estimates. Fourth, the generalizability of the findings may be limited owing to differences in occupational and cultural backgrounds between countries. However, these results still offer valuable insights compared to other similar studies conducted in Asian countries, where socioeconomic and employment contexts might be more comparable to those in Korea. Fifth, owing to missing data on the cause of death, we could not analyze specific causes of mortality. Therefore, we focused on the association between job insecurity and overall health outcomes.
In summary, prolonged job insecurity demonstrates a gender difference in its impact on the risk of newly diagnosed CVD, particularly among individuals with type 2 diabetes. Targeted policies and comprehensive strategies addressing job insecurity and health management in individuals with type 2 diabetes are required to effectively prevent these outcomes.
