Skip Navigation
Skip to contents

Epidemiol Health : Epidemiology and Health

OPEN ACCESS
SEARCH
Search

Articles

Page Path
HOME > Epidemiol Health > Volume 48; 2026 > Article
Original Article
The impact of health behavior trajectories on all-cause and cause-specific mortality: an analysis using the National Health Insurance Service-National Health Screening Cohort in Korea
Jeong Min Yang1orcid, Jieun Hwang2,3orcid
Epidemiol Health 2026;48:e2026029.
DOI: https://doi.org/10.4178/epih.e2026029
Published online: June 12, 2026

1Office of Public Healthcare, National Medical Center, Seoul, Korea

2Department of Health Administration, Dankook University College of Health Science, Cheonan, Korea

3Institute of Convergence Healthcare, Dankook University, Cheonan, Korea

Correspondence: Jieun Hwang Department of Health Administration, Dankook University College of Health Science, 119 Dandae-ro, Dongnam-gu, Cheonan 31116, Korea E-mail: hwang0310@dankook.ac.kr
• Received: November 25, 2025   • Revised: May 12, 2026   • Accepted: June 1, 2026

© 2026, Korean Society of Epidemiology

This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

prev next
  • 4,163 Views
  • 134 Download
  • 1 Web of Science
  • 1 Crossref
  • OBJECTIVES
    Non-communicable diseases account for approximately 75% of global deaths and nearly 80% of deaths in Korea. Although the World Health Organization emphasizes the long-term management of multiple health behaviors (HBs), most studies have examined individual behaviors over short periods, thereby overlooking behavioral changes and their long-term effects. This study assessed how combined health behavior trajectories (HBTs) were associated with mortality risk, with consideration of sex-specific and age-specific differences.
  • METHODS
    Using the National Health Insurance Service-National Health Screening Cohort in Korea, we analyzed 167,343 Korean adults aged 40–79 years, including 95,226 male participants and 72,117 female participants. The mean follow-up duration was 10.34 years for male participants and 10.65 years for female participants. Group-based multi-trajectory modeling was used to identify trajectories of tobacco use, alcohol consumption, physical activity, and body mass index. Associations with all-cause, cancer, and cardiovascular disease (CVD) mortality were examined using time-dependent Cox proportional hazards models stratified by sex and age.
  • RESULTS
    Six distinct HBTs were identified in each sex, including a decreasing-risk trajectory that reflected improvements across multiple behaviors. Among male participants, the higher severe-risk group had markedly elevated all-cause mortality (hazard ratio [HR], 2.02; 95% confidence interval [CI], 1.63 to 2.51) and CVD mortality (HR, 3.17; 95% CI, 2.01 to 4.99). Among female participants, the severe-risk group had significantly higher risks of all-cause mortality (HR, 1.83; 95% CI, 1.43 to 2.34), cancer mortality (HR, 1.91; 95% CI, 1.20 to 3.05), and CVD mortality (HR, 2.68; 95% CI, 1.77 to 4.05). The decreasing-risk trajectory was associated with lower all-cause mortality in both sexes, particularly among individuals aged ≥65 years (male participants: HR, 0.87; 95% CI, 0.80 to 0.94; female participants: HR, 0.88; 95% CI, 0.79 to 0.98).
  • CONCLUSIONS
    Distinct sex-specific HBTs were strongly associated with mortality. Sustained improvements across multiple HBs were associated with substantially lower long-term mortality. These findings support preventive health policies that target middle-aged and older populations.
• Combined long-term trajectories of tobacco use, alcohol consumption, physical activity, and body mass index showed strong sex-specific associations with all-cause, cancer, and cardiovascular disease mortality.
• Trajectories characterized by persistent smoking and alcohol use, particularly when combined with unhealthy body mass index patterns, were associated with substantially higher mortality risks, most notably among male participants.
• A decreasing-risk trajectory, representing sustained improvements across multiple behaviors, was associated with significantly lower all-cause mortality, especially among adults aged ≥65 years.
• Public health policies should prioritize life-course approaches that promote simultaneous improvements in multiple health behaviors to reduce premature mortality in middle-aged and older populations.
Non-communicable diseases (NCDs) are chronic conditions with gradual onset and long-term persistence; therefore, public health efforts generally emphasize prevention and long-term management rather than eradication [1,2]. In 2021, NCDs accounted for 75% of all deaths globally and 79.6% of deaths in Korea [3,4]. A large-scale analysis across 195 countries further showed that NCD-related mortality risk increases sharply from age 40 years onward in both sexes [5].
The World Health Organization emphasizes the long-term management of multiple health behaviors (HBs), including tobacco use, alcohol consumption, physical activity (PA), and body mass index (BMI), to reduce NCD-related mortality [6]. Each of these HBs has been independently associated with cause-specific mortality, including cancer and cardiovascular disease (CVD) mortality [7,8]. Their prevalence in Korea remains higher than the global average, further increasing concern about premature mortality risk [4].
Although considerable research has examined associations between individual HBs and mortality [9], single-behavior studies may not fully capture how multiple risk factors jointly influence long-term health outcomes [10]. Accordingly, increasing attention has been given to examining multiple HBs simultaneously [11]. Smoking, alcohol consumption, PA, and BMI have consistently been identified as core behavioral domains [12,13]. Nevertheless, most studies have assessed HBs cross-sectionally, providing limited evidence on the cumulative effects of sustained unhealthy behaviors on mortality [14].
The interpretation of health behavior trajectories (HBTs) is context-dependent [15]. In East Asia, including Korea, smoking and alcohol consumption among males are strongly shaped by socio-cultural norms, including workplace-based socialization, which may sustain high-risk trajectories and modify their associations with mortality [15]. In addition, Asian populations experience poorer health outcomes than Western populations at comparable levels of behavioral risk, including higher CVD risk at similar levels of smoking and greater metabolic risk at equivalent BMI values [16]. These differences underscore the need for ethnicity-specific, long-term cohort studies on the joint effects of multiple HBs on mortality.
Despite growing interest in multiple HBs, few studies have jointly modeled 4 or more HBs as correlated, long-term trajectories with mortality as the primary outcome [14]. Most studies have examined behavioral clustering at a single time point or modeled individual behavior trajectories separately [17]. In addition, scoring-based approaches often assume equal effects of all HBs on mortality, which may introduce misclassification bias [18]. Even longitudinal studies have typically estimated trajectories for each behavior independently rather than identifying subgroups defined by the joint evolution of multiple behaviors within individuals [19].
Therefore, building on a life-course epidemiology framework, the present study applied group-based multi-trajectory modeling (GBMTM) to simultaneously identify joint trajectories of smoking, alcohol consumption, PA, and BMI among middle-aged and older adults. This study examined how co-occurring trajectories of smoking, alcohol consumption, PA, and BMI were associated with subsequent all-cause, cancer, and CVD mortality.
Data source and study population
The National Health Insurance Service-National Health Screening (NHIS-HEALS) Cohort comprises a 10% random sample (n=514,789) of health-screened individuals aged 40–79 years in 2002–2003, drawn from the National Health Insurance Service (NHIS) biennial screening program, which covers approximately 97% of the Korean population [20]. The database includes anthropometric measurements, laboratory tests, HB questionnaires, insurance eligibility, income, demographic characteristics, medical treatment records, and cause and date of death from 2002 to 2019.
From the 514,789 individuals enrolled in 2002–2003, we sequentially excluded 43,676 former smokers at baseline to minimize the “sick quitter” effect [21] and participants with prior diagnoses of conditions associated with all-cause, cancer, or CVD mortality, yielding 430,871 participants. Participants with missing HB data across the 4 waves from 2002 to 2009 or who died during this period were further excluded to construct HBTs, leaving 171,901 participants. Finally, exclusion of participants with missing covariate data in 2008–2009 resulted in a final analytic sample of 95,226 male participants and 72,117 female participants (Supplementary Material 1).
Missing data were handled using listwise deletion, and differences in baseline characteristics between included and excluded participants were examined using chi-square tests to assess potential selection bias [22] (Supplementary Material 2).
Data from 2002 to 2008/2009 were used to construct the GBMTM and identify HBTs. Mortality follow-up began the day after participants completed the final HBT-related questionnaire in 2008/2009 and continued through December 31, 2019, ensuring complete temporal separation between the trajectory assessment and follow-up periods. The 2008/2009 period was selected as the baseline because evidence suggests that approximately 10 years may be required to detect measurable effects of health behavior changes on mortality outcomes [23].
Independent variables

HBTs

Two analytic steps were used to identify HBTs. First, each HB variable was categorized. Second, GBMTM was applied to estimate trajectory patterns.
Four HBs were included in this study: tobacco use, alcohol consumption, PA, and BMI. Tobacco use was classified into 5 categories based on the number of cigarettes smoked per day [24]: (1) 0 cigarette/day, (2) 1–9 cigarette/day, (3) 10–19 cigarette/day, (4) 20–39 cigarette/day, and (5) ≥40 cigarette/day. Alcohol consumption was estimated using self-reported drinking frequency and quantity per occasion from the health screening questionnaire [25]. Alcohol intake was calculated in grams per day and categorized as follows [25]: (1) 0 g/day, (2) 1.0–9.9 g/day, (3) 10.0–19.9 g/day, (4) 20.0–29.9 g/day, and (5) ≥30.0 g/day. PA was measured as the weekly frequency of moderate-to-vigorous exercise based on the question, “How many times per week do you engage in PA intense enough to induce sweating?” Responses were categorized into 5 levels [20]: (1) none, (2) 1–2 times/wk, (3) 3–4 times/wk, (4) 5–6 times/wk, and (5) almost every day. This categorization followed established approaches from prior studies [20] and directly reflected the NHIS-HEALS questionnaire structure, in which PA responses were collected as multiple-choice items and treated as an ordinal variable in the GBMTM. BMI was categorized into 5 groups based on the clinical practice guidelines of the Korean Society for the Study of Obesity [26]: (1) underweight (<18.5 kg/m2), (2) normal weight (18.5–22.9 kg/m2), (3) overweight (23.0–24.9 kg/m2), (4) obesity (25.0–29.9 kg/m2), and (5) severe obesity (≥30.0 kg/m2). GBMTM was applied to estimate HBTs over an approximately 8-year period. Model fit was assessed using the Akaike information criterion, Bayesian information criterion, entropy, and average posterior probability [27]. A 6-group trajectory model was identified as the optimal solution for both male participants and female participants (Supplementary Material 3).
The annual HB characteristics of the male and female HBTs identified through GBMTM are presented in Supplementary Materials 4 and 5. HBT labels were assigned retrospectively based on hazard ratios (HRs) estimated from time-dependent Cox proportional hazards models. Specifically, groups with higher HRs relative to the reference group were labeled “severe risk,” whereas groups with intermediate HRs were labeled “moderate risk.” Within each risk level, groups were further differentiated as “higher” or “lower” according to the relative magnitude of their HRs.
Dependent variables

All-cause and cause-specific mortality

All-cause mortality was assessed over a maximum follow-up of 12 years, from the 2008/2009 baseline to December 31, 2019, with a mean follow-up of 3,777 days. Deaths and causes of death were verified through linkage with Resident Registration Death Data from the Korea National Statistical Office using unique personal identification numbers. Mortality ascertainment was complete because both databases are nationwide registries. Cause-specific mortality included cancer mortality, defined using International Classification of Diseases, 10th revision (ICD-10) codes C00–C99 [28], and CVD mortality, defined using ICD-10 codes I00–I99 [29]. The mean follow-up duration by sex-specific HBT is presented in Supplementary Material 6.
Control variables
Potential confounders included factors associated with all-cause and cause-specific mortality: age (<55, 55–64, 65–74, 75–84, and ≥85 years), region (capital, metropolitan, or other areas), medical insurance type, income level (low, middle, or high), disability status (normal, mild, or severe), and Charlson comorbidity index (CCI). The CCI, a validated measure of comorbidity burden in longitudinal studies, was calculated using ICD-10 codes and categorized as 0, 1, 2, or ≥3 [30]. Regions were classified as the Seoul Capital Area, metropolitan cities, or other regions. Income level was defined using household insurance premium deciles (low: 0–3, middle: 4–7, high: 8–10) [31]. Disability status was determined according to national legal criteria and classified as normal, mild, or severe, with grading revisions applied from 2016 onward [32].
Analytical approach and statistics
GBMTM was applied to data on the 4 HBs collected from 2002 to 2009. Differences in mortality risk across HBTs were assessed using chi-square tests and Cohen’s h measures [33]. Kaplan–Meier survival curves and log-rank tests were used to compare survival probabilities by HBT type (Supplementary Material 7). Associations between HBTs and mortality were analyzed using time-dependent Cox proportional hazards models, in which time-varying covariates were updated annually from the 2008/2009 baseline throughout follow-up. Proportional hazards assumptions were evaluated using Schoenfeld residuals and were satisfied for all models.
Among the 6 trajectory groups identified in male participants and female participants, 3 groups were composed of relatively small numbers of participants and were classified as extreme high-risk trajectories: the higher severe-risk group among male participants (1.1%), the lower severe-risk group among male participants (0.9%), and the severe-risk group among female participants (0.9%). Accordingly, sensitivity analyses were conducted by merging these small trajectory groups with adjacent trajectory groups to assess the robustness of the main findings (Supplementary Materials 8 and 9). After trajectory groups with small numbers of participants were merged with adjacent trajectories showing similar risk profiles, the estimated HRs did not attenuate toward the null. Instead, their direction and overall pattern were preserved, with modest increases observed in several estimates.
Data management and statistical analyses were performed using SAS Enterprise Guide 8.3 (SAS Institute Inc., Cary, NC, USA) and RStudio 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria). A p-value of <0.05 was considered statistically significant.
Ethics statement
This study was reviewed and approved by the Institutional Review Board of Dankook University’s Health System in accordance with the principles of the Declaration of Helsinki (IRB No. DKU IRB 2024-10-008). Furthermore, as the NHIS- HEALS data used for the analysis do not contain personally identifiable information, the requirement for informed consent requirement was waived (research ID: NHIS-2024-11-2-055).
Trajectory class of health behavior by sex over time
Figure 1 illustrates the 6 HBTs identified among 95,226 male participants. The stable healthy group (30.3%, reference) was characterized by persistent non-smoking, low alcohol consumption, increasing PA, and stable normal BMI. The decreasing-risk group (28.0%) showed favorable changes across behaviors, including smoking cessation, reduced alcohol intake, increased PA, and BMI improvement. Two moderate-risk trajectories were identified: the higher moderate-risk group (19.0%), characterized by persistently moderate smoking and alcohol use and low PA, and the lower moderate-risk group (20.8%), characterized by sustained smoking and alcohol consumption, low PA, and overweihgt BMI. Two smaller severe-risk trajectories were also observed: the higher severe-risk group (1.1%), characterized by persistent smoking, low PA, and obese BMI, and the lower severe-risk group (0.9%), characterized by continued smoking, low alcohol intake, minimal PA, and persistent underweight.
Figure 2 illustrates the HBTs identified among 72,117 female participants, with a 6-group model showing the best fit. The stable healthy group (34.7%, reference) was characterized by persistent non-smoking and non-drinking, modest increases in PA, and stable normal BMI. The decreasing-risk group (6.8%) showed favorable behavioral changes over time, including reduced alcohol intake and improved BMI. The moderate-risk group (10.2%) exhibited persistent light smoking, low alcohol consumption, moderate PA, and stable normal BMI. In contrast, the severe-risk group (0.9%) was characterized by continued smoking and alcohol consumption, low PA, and overweihgt BMI.
Sample characteristics by sex
Table 1 shows the distribution of mortality by trajectory among male participants (n=95,226). The all-cause mortality rate was 9.1% (n=8,637), the cancer mortality rate was 3.7% (n=3,549), and the CVD mortality rate was 1.5% (n=1,418). All-cause mortality was highest in the lower severe-risk group at 27.1% (n=236) and lowest in the lower moderate-risk group at 7.1% (n=1,407) (p<0.001). Cancer mortality was also highest in the lower severe-risk group (7.1%, n=62) and lowest in the decreasing-risk group (3.1%, n=834; p<0.001). For CVD mortality, the highest rate was again observed in the lower severe-risk group (3.7%, n=32), whereas the decreasing-risk group and lower moderate-risk group had the lowest rates, at 1.2% (n=328 and 240, respectively; p<0.001).
Table 1 shows the distribution of mortality by trajectory among female participants (n=72,117). The all-cause mortality rate was 5.3% (n=3,856), the cancer mortality rate was 1.8% (n=1,286), and the CVD mortality rate was 1.3% (n=926). All-cause mortality was highest in the severe-risk group at 12.5% (n=81) and lowest in the decreasing-risk group at 3.0% (n=147) (p<0.001). Similarly, cancer mortality was highest in the severe-risk group (3.9%, n=25) and lowest in the decreasing-risk group (1.4%, n=69; p<0.001). For CVD mortality, the severe-risk group also had the highest rate (3.9%, n=25), whereas the decreasing-risk group had the lowest rate (0.6%, n=31; p<0.001).
Adjusted associations between health behavior trajectories and all-cause and cause-specific mortality
Table 2 summarizes the associations between HBTs and mortality among male participants. For all-cause mortality, the lower and higher severe-risk groups had significantly higher risks than the stable healthy group (HR, 2.11; 95% CI, 1.79 to 2.49 and HR, 2.02; 95% CI, 1.63 to 2.51, respectively), whereas the decreasing-risk group had a lower risk (HR, 0.91; 95% CI, 0.85 to 0.97). For cancer mortality, significantly higher risks were observed in the higher severe-risk, higher moderate-risk, and lower moderate-risk groups (HRs, 2.54, 1.94, and 1.74, respectively; all p<0.05). For CVD mortality, the higher and lower severe-risk groups had markedly increased risks (HR, 3.17; 95% CI, 2.01 to 4.99 and HR, 2.00; 95% CI, 1.33 to 3.01, respectively).
Table 2 summarizes the associations between HBTs and mortality among female participants. For all-cause mortality, the severe-risk and moderate-risk groups had significantly higher risks than the stable healthy group (HR, 1.83; 95% CI, 1.43 to 2.34; HR, 1.25; 95% CI, 1.09 to 1.44), whereas the decreasing-risk group had a lower risk (HR, 0.90; 95% CI, 0.82 to 0.99). For cancer mortality, a significantly elevated risk was observed only in the severe-risk group (HR, 1.91; 95% CI, 1.20 to 3.05). For CVD mortality, both the severe-risk and moderate-risk groups had increased risks compared with the reference group (HR, 2.68; 95% CI, 1.77 to 4.05 and HR, 1.43; 95% CI, 1.08 to 1.89, respectively).
Stratified associations between health behavior trajectories and all-cause and cause-specific mortality
Figure 3 shows the age-stratified analyses, using a cutoff of 65 years, for male participants and female participants. Among male participants, HRs for all-cause, cancer, and CVD mortality were generally higher among those younger than 65 years. Notably, male participants younger than 65 years in the higher severe-risk group had the highest CVD mortality risk (HR, 5.76; 95% CI, 3.29 to 10.10), whereas male participants aged ≥65 years in the decreasing-risk group had a lower risk of all-cause mortality (HR, 0.87; 95% CI, 0.80 to 0.94). Among female participants, HRs for all-cause and cancer mortality were higher among those aged <65 years, whereas CVD mortality risk was higher among those aged ≥65 years. Similarly, female participants aged ≥65 years in the decreasing-risk group had a lower risk of all-cause mortality (HR, 0.88; 95% CI, 0.79 to 0.98).
This study used the 2002–2019 NHIS-HEALS to identify sex-specific HBTs and examine their associations with mortality. Analysis of HBTs identified 5 maintenance trajectories and 1 improvement trajectory common to both male participants and female participants. Male participants had higher levels of smoking and alcohol consumption than female participants, and a distinct underweight trajectory was observed only among male participants. Severe-risk trajectories characterized by extreme BMI were strongly associated with all-cause and CVD mortality in both sexes. In contrast, a moderate-risk trajectory characterized by high smoking and alcohol consumption despite normal BMI was specifically associated with increased cancer mortality among male participants. Age-stratified analyses indicated higher mortality risks among male participants younger than 65 years, whereas female participants showed cause-specific age differences. Notably, the decreasing-risk trajectory was consistently associated with lower mortality risk across sexes, particularly among adults aged ≥65 years.
Differentiated HBTs were observed between male participants and female participants, with more pronounced unhealthy patterns among male participants, consistent with previous findings. In Asian contexts, smoking and alcohol consumption are often embedded in male-dominated workplace and social norms and may function as means of social networking and stress relief. In contrast, female participants tend to maintain more stable HBs and exhibit greater risk avoidance, partly because of socio-cultural expectations and a pronounced sick-quitter effect [15,34,35]. From a life-course epidemiology perspective, these sex differences indicate that unhealthy behavioral exposure is not merely a matter of individual choice but is also structurally shaped by gendered social and labor roles [36]. This finding underscores the need to conceptualize HBTs as socially patterned processes and highlights the importance of sex-specific public health strategies, including structural workplace-based interventions for male and life-stage-sensitive preventive approaches for female [13,15].
Consistent with these sex differences, associations between HBTs and mortality also differed by sex. Among male participants, both severe-risk trajectories characterized by severe BMI were strongly associated with all-cause and CVD mortality, and the high-risk trajectory among female participants showed similar patterns. Prolonged exposure to unhealthy behaviors was associated with higher mortality risk, particularly when smoking co-occurred with persistent obesity or underweight. Even modest BMI deviations were associated with increased CVD mortality [14,37]. Notably, favorable behavioral changes remained associated with lower mortality risk even in later life, supporting the persistence of behavioral plasticity beyond midlife and the importance of HB modification for secondary prevention and healthy aging in older populations [38].
Several biological mechanisms may explain the associations between HBTs and mortality. Severe BMI can accelerate endothelial dysfunction and atherosclerosis, increasing thrombotic and stroke risk; these effects may be further exacerbated by smoking and alcohol consumption through lipid oxidation and impaired high-density lipoprotein function [39]. Among underweight individuals, limited myocardial energy reserves combined with smoking and alcohol use may increase CVD mortality through atherosclerotic and arrhythmic pathways [40]. In contrast, among male participants with normal BMI but high tobacco and alcohol use, moderate-risk trajectories were associated with cancer mortality, consistent with evidence that concurrent smoking and alcohol consumption synergistically enhance carcinogen absorption and promote carcinogenesis [12,41].
In the age-stratified analysis, the decreasing-risk group aged ≥65 years had lower all-cause mortality risk than the reference group in both sexes. This finding is consistent with evidence that long-term smoking and alcohol cessation reduce mortality risk to levels comparable to those among individuals who maintain healthy lifestyles, with benefits strengthening over time [42,43]. Mortality reductions in the decreasing-risk group became evident only over longer follow-up periods, particularly among female participants. Cohort effects may partially explain these age-stratified patterns: older Korean males were more likely to have initiated smoking and heavy alcohol consumption earlier in life under prevailing socio-cultural norms, whereas younger cohorts reflect behavioral shifts following tobacco-control policies [44].
Among male participants, stronger HRs were observed in those younger than 65 years, consistent with a selective survival mechanism [45]. Male participants in high-risk trajectories may have been disproportionately exposed to smoking and alcohol from earlier in life, accelerating disease progression and increasing premature mortality before age 65 years [46]. This pattern may have left a selectively healthier survivor pool at older ages. Among female participants, stronger associations for all-cause and cancer mortality were observed in those younger than 65 years, whereas higher CVD mortality risk was noted among those aged ≥65 years. This divergence may reflect distinct biological pathways: younger female participants may be particularly susceptible to CVD through hormonal mechanisms, as earlier estrogen loss amplifies cardiovascular vulnerability, whereas cancer risk accumulates over prolonged exposure and may manifest more prominently at older ages [47]. Consequently, the selective survival effect was more pronounced among male participants, whereas female participants exhibited differential age-specific and cause-specific mortality patterns driven by biological susceptibility and cumulative behavioral exposure.
In Korea, community health policies have largely emphasized post-onset disease management or prevention targeting specific diseases, leaving premature mortality risk insufficiently addressed at the population level [15]. The findings of this study suggest that the existing NHIS General Health Screening Program, which provides biennial health examinations covering cardiovascular risk factors and cancer screening for adults aged ≥40 years, could be strengthened by incorporating trajectory-based risk stratification to identify high-risk behavioral clusters earlier [48]. Specifically, male participants in high-risk trajectories characterized by concurrent smoking, alcohol consumption, and extreme BMI represent a priority target group for workplace-based behavioral interventions, given the strong occupational and social embeddedness of these behaviors among Korean males [15]. Furthermore, the finding that the decreasing-risk trajectory was associated with lower all-cause mortality in both midlife and older adulthood, particularly among those aged ≥65 years, underscores the importance of sustained, multi-behavior health promotion strategies across the life course rather than single-point or disease-specific interventions. Evidence from Japan indicates that community-based, prevention-oriented health policies can reduce disease risk, suggesting potential benefits of adopting similar approaches in Korea [49].
This study has several limitations. First, dietary habits and sleep could not be included because of data constraints, and PA was measured only as weekly frequency, without information on duration or intensity. Second, although a 2-stage modeling strategy was used with complete temporal separation between the trajectory assessment period (2002–2008/2009) and mortality follow-up, and although former smokers and individuals with major diseases at baseline were excluded, reverse causality cannot be completely ruled out. Subclinical or undiagnosed conditions during trajectory assessment may have led to misclassification into lower-risk groups, potentially underestimating associations between unhealthy behaviors and mortality or overestimating the protective association of the decreasing-risk group. All findings should therefore be interpreted as associations rather than causal effects. Third, residual confounding from unmeasured factors, such as education and family structure, may remain despite extensive covariate adjustment. Fourth, listwise deletion of missing values may have introduced selection bias because the analytic sample tended to be younger and more socioeconomically advantaged than excluded participants, potentially underestimating associations between unhealthy HBs and mortality while overestimating the protective association of the decreasing-risk group. Fifth, 3 trajectory groups comprising less than 1% of the study population—the higher and lower severe-risk groups among male participants and the severe-risk group among female participants—had limited numbers of deaths, and their HR estimates may be subject to overestimation or imprecision despite sensitivity analyses confirming the consistency of the overall direction and pattern of associations.
The data that support the findings of this study are available from National Health Insurance Service but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Jeong Min Yang.
Supplementary materials are available at https://doi.org/10.4178/epih.e2026029.

Supplementary Material 1.

Flow chart for sample selection
epih-48-e2026029-Supplementary-1.docx

Supplementary Material 2.

Comparison between the analytic and excluded samples by sex
epih-48-e2026029-Supplementary-2.docx

Supplementary Material 3.

Model fit statistics for estimated patterns of trajectory
epih-48-e2026029-Supplementary-3.docx

Supplementary Material 4.

Annual health behavior characteristics by trajectory group among males
epih-48-e2026029-Supplementary-4.docx

Supplementary Material 5.

Annual health behavior characteristics by trajectory group among females
epih-48-e2026029-Supplementary-5.docx

Supplementary Material 6.

Mean follow-up periods by sex-specific health behavior trajectories
epih-48-e2026029-Supplementary-6.docx

Supplementary Material 7.

Kaplan-Meier survival curves by health behavior trajectories among males and females
epih-48-e2026029-Supplementary-7.docx

Supplementary Material 8.

Sensitivity analysis results after merging small health behavior trajectory among males
epih-48-e2026029-Supplementary-8.docx

Supplementary Material 9.

Sensitivity analysis results after merging small health behavior trajectory among females
epih-48-e2026029-Supplementary-9.docx

Conflict of interest

The authors have no conflicts of interest to declare for this study.

Funding

This research was supported by the publication fee support from the Researcher Support Project of the Public Health and Medical Research Institute of the National Medical Center.

Acknowledgements

This article is based on the doctoral dissertation Jeong Min Yang in 2025. We would like to thank colleagues from the Department of Public Health, Graduate School of Dankook University, who provided their advice in the preparation of this manuscript.

Author contributions

Both authors contributed equally to conceiving the study, analyzing the data, and writing this paper.

Figure 1.
Trajectories of health behaviors (A) tbacco use, (B) alcohol consumption, (C) PA, and (D) BMI among male participants. This figure illustrates 6 male health behavior trajectories, color-coded as follows: higher moderate-risk group (red, 19.0%; stable moderate smoking and alcohol use, normal BMI), higher severe-risk group (green, 1.1%; stable low smoking, obese BMI), stable healthy group (blue, 30.3%; non-smoking, low alcohol consumption, normal BMI), decreasing-risk group (black, 28.0%; concurrent reductions in smoking, alcohol consumption, and BMI with increasing PA), lower severe-risk group (yellow, 0.9%; stable low smoking, underweight BMI), and lower moderate-risk group (purple, 20.8%; stable moderate smoking and alcohol use, overweight BMI). BMI, body mass index; PA, physical activity.
epih-48-e2026029f1.jpg
Figure 2.
Trajectories of health behaviors (A) tbacco use, (B) alcohol consumption, (C) PA, and (D) BMI among female participants. This figure illustrates 6 female health behavior trajectories, color-coded as follows: moderate-risk group (red, 10.2%; stable low smoking and alcohol consumption, normal BMI), stable healthy group (green, 34.7%; non-smoking, no alcohol consumption, stable normal BMI), low-risk group (blue, 23.8%; non-smoking, no alcohol consumption, stable slightly overweight BMI), mild-risk group (black, 23.5%; non-smoking, no alcohol consumption, stable overweight BMI), decreasing-risk group (yellow, 6.8%; non-smoking, decreasing alcohol consumption and BMI with increasing PA), and severe-risk group (purple, 0.9%; stable low smoking and alcohol consumption, overweight BMI). BMI, body mass index; PA, physical activity.
epih-48-e2026029f2.jpg
Figure 3.
Stratified analysis of the association of health behavior trajectories (HBTs) with mortality risk by sex and age (A) male participants aged <65 years; (B) male participants aged ≥65 years; (C) female participants aged <65 years; (D) female participants aged ≥65 years. This figure illustrates age-stratified hazard ratios for all-cause, cancer, and cardiovascular disease (CVD) mortality across HBT groups, stratified by sex and age. Square, diamond, and triangle markers represent all-cause, cancer, and CVD mortality, respectively, and horizontal lines indicate 95% confidence intervals. The reference group for all comparisons is the stable healthy group.
epih-48-e2026029f3.jpg
epih-48-e2026029f4.jpg
Table 1.
General characteristics of health behavior trajectories and mortality among male and female participants at baseline (2008–2009)
Characteristics Total All-cause mortality
p-value (effect size)1 Cancer mortality
p-value (effect size)1 CVD mortality
p-value (effect size)1
Death Death Death
 Disability <0.001 <0.001 <0.001
  Normal 87,387 (91.8) 7,205 (8.2) - 3,091 (3.5) - 1,128 (1.3) -
  Mild 6,041 (6.3) 967 (16.0) (0.22) 332 (5.5) (0.09) 194 (3.2) (0.13)
  Severe 1,798 (1.9) 465 (25.9) (0.44) 126 (7.0) (0.15) 96 (5.3) (0.23)
 CCI <0.001 <0.001 <0.001
  0 65,317 (68.6) 4,610 (7.1) - 1,964 (3.0) - 770 (1.2) -
  1 24,370 (25.6) 2,862 (11.7) (0.15) 1,141 (4.7) (0.08) 468 (1.9) (0.06)
  2 4,736 (5.0) 915 (19.3) (0.34) 323 (6.8) (0.17) 141 (3.0) (0.13)
  ≥3 803 (0.8) 250 (31.1) (0.59) 121 (15.1) (0.43) 39 (4.9) (0.22)
Female
 Total 72,117 (100) 3,856 (5.3) 1,286 (1.8) 926 (1.3)
 Health behavior trajectory <0.001 <0.001 <0.001
  Moderate risk 7,382 (10.2) 330 (4.5) (0.43) 112 (1.5) (0.25) 72 (1.0) (0.20)
  Stable healthy 25,045 (34.7) 1,317 (5.3) - 409 (1.6) - 294 (1.2) -
  Low risk 17,198 (23.8) 921 (5.4) (0.47) 302 (1.8) (0.27) 217 (1.3) (0.23)
  Mild risk 16,916 (23.5) 1,060 (6.3) (0.51) 369 (2.2) (0.30) 287 (1.7) (0.26)
  Decreasing risk 4,927 (6.8) 147 (3.0) (0.35) 69 (1.4) (0.24) 31 (0.6) (0.16)
  Severe risk 649 (0.9) 81 (12.5) (0.72) 25 (3.9) (0.40) 25 (3.9) (0.40)
 Age (yr) <0.001 <0.001 <0.001
  <55 31,309 (43.4) 347 (1.1) - 202 (0.6) - 41 (0.1) -
  55-64 25,974 (36.0) 709 (2.7) (0.33) 373 (1.4) (0.24) 120 (0.5) (0.14)
  65-74 11,149 (15.5) 1,516 (13.6) (0.76) 478 (4.3) (0.42) 394 (3.5) (0.38)
  75-84 3,633 (5.0) 1,254 (34.5) (1.26) 228 (6.3) (0.51) 363 (10.0) (0.64)
  ≥85 52 (0.1) 30 (57.7) (1.73) 5 (9.6) (0.63) 8 (15.4) (0.81)
 Region <0.001 0.002 <0.001
  Capital area 29,987 (41.6) 1,375 (4.6) - 517 (1.7) - 317 (1.1) -
  Metropolitan area 15,872 (22.0) 742 (4.7) (0.44) 246 (1.5) (0.25) 174 (1.1) (0.21)
  Other areas 26,258 (36.4) 1,739 (6.6) (0.52) 523 (2.0) (0.28) 435 (1.7) (0.26)
 Medical insurance <0.001 <0.001 <0.001
  National Health Insurance (region) 19,272 (26.7) 1,426 (7.4) - 441 (2.3) - 365 (1.9) -
  National Health Insurance (workplace) 52,727 (73.1) 2,406 (4.6) (0.43) 840 (1.6) (0.25) 552 (1.0) (0.20)
  Medical Aid 118 (0.2) 24 (20.3) (0.94) 5 (4.2) (0.41) 9 (7.6) (0.56)
 Income level <0.001 0.051 0.025
  Low 16,979 (23.5) 1,019 (6.0) - 311 (1.8) - 249 (1.5) -
  Mid 24,640 (34.2) 1,201 (4.9) (0.49) 399 (1.6) (0.27) 319 (1.3) (0.24)
  High 30,498 (42.3) 1,636 (5.4) (0.45) 576 (1.9) (0.26) 358 (1.2) (0.23)
 Disability <0.001 <0.001 <0.001
  Normal 67,669 (93.8) 3,319 (4.9) - 1,148 (1.7) - 777 (1.1) -
  Mild 3,631 (5.0) 398 (11.0) (0.67) 115 (3.2) (0.36) 106 (2.9) (0.34)
  Severe 817 (1.1) 139 (17.0) (0.85) 23 (2.8) (0.34) 43 (5.3) (0.46)
 CCI <0.001 <0.001 <0.001
  0 44,009 (61.0) 1,891 (4.3) - 673 (1.5) - 431 (1.0) -
  1 21,672 (30.1) 1,335 (6.2) (0.50) 439 (2.0) (0.29) 325 (1.5) (0.25)
  2 5,472 (7.6) 470 (8.6) (0.59) 124 (2.3) (0.30) 121 (2.2) (0.30)
  ≥3 964 (1.3) 160 (16.6) (0.84) 50 (5.2) (0.46) 49 (5.1) (0.45)

Values are presented as number (%).

CVD, cardiovascular disease; CCI, Charlson comorbidity index.

1 Effect sizes are presented as Cohen’s h.

Table 2.
Adjusted associations of health behavior trajectories on all-cause and cause-specific mortality among male and female participants
Variables 1-yr 3-yr 5-yr 7-yr Entire period
Male
 All-cause mortality
  Health behavior trajectory
   Higher moderate risk 1.00 (0.93, 1.07) 1.06 (0.99, 1.13) 1.22 (1.14, 1.31)* 1.42 (1.33, 1.52)* 1.66 (1.56, 1.78)*
   Higher severe risk 0.88 (0.71, 1.09) 0.94 (0.76, 1.17) 1.17 (0.94, 1.45) 1.45 (1.17, 1.80)* 2.02 (1.63, 2.51)*
   Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
   Decreasing risk 0.99 (0.92, 1.06) 0.99 (0.92, 1.06) 0.99 (0.92, 1.05) 0.97 (0.91, 1.04) 0.91 (0.85, 0.97)*
   Lower severe risk 1.14 (0.97, 1.35) 1.21 (1.03, 1.43)* 1.46 (1.23, 1.72)* 1.79 (1.51, 2.11)* 2.11 (1.79, 2.49)*
   Lower moderate risk 0.98 (0.90, 1.06) 1.02 (0.94, 1.10) 1.14 (1.05, 1.23)* 1.26 (1.17, 1.37)* 1.37 (1.27, 1.48)*
 Cancer cause mortality
  Health behavior trajectory
   Higher moderate risk 1.11 (0.99, 1.24) 1.19 (1.06, 1.33)* 1.40 (1.25, 1.57)* 1.64 (1.47, 1.83)* 1.94 (1.74, 2.17)*
   Higher severe risk 1.06 (0.77, 1.46) 1.16 (0.84, 1.60) 1.47 (1.07, 2.02)* 1.87 (1.36, 2.57)* 2.54 (1.85, 3.48)*
   Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
   Decreasing risk 1.12 (1.00, 1.25) 1.11 (0.99, 1.24) 1.17 (1.00, 1.25) 1.11 (0.99, 1.24) 1.07 (0.95, 1.19)
   Lower severe risk 0.76 (0.54, 1.07) 0.82 (0.58, 1.16) 1.00 (0.71, 1.42) 1.25 (0.88, 1.76) 1.50 (1.06, 2.12)*
   Lower moderate risk 1.18 (1.05, 1.34)* 1.25 (1.10, 1.41)* 1.42 (1.25, 1.59)* 1.59 (1.41, 1.79)* 1.74 (1.54, 1.96)*
 CVD cause mortality
  Health behavior trajectory
   Higher moderate risk 0.97 (0.82, 1.16) 1.02 (0.85, 1.21) 1.16 (0.98, 1.38) 1.33 (1.12, 1.58)* 1.53 (1.29, 1.82)*
   Higher severe risk 1.46 (0.92, 2.30) 1.54 (0.98, 2.44) 1.87 (1.19, 2.95)* 2.28 (1.45, 3.60)* 3.17 (2.01, 4.99)*
   Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
   Decreasing risk 1.12 (0.95, 1.32) 1.11 (0.95, 1.31) 1.11 (0.94, 1.31) 1.08 (0.92, 1.27) 0.98 (0.84, 1.16)
   Lower severe risk 1.17 (0.78, 1.76) 1.23 (0.82, 1.85) 1.45 (0.96, 2.17) 1.74 (1.16, 2.62)* 2.00 (1.33, 3.01)*
   Lower moderate risk 1.05 (0.86, 1.28) 1.09 (0.89, 1.32) 1.20 (0.99, 1.46) 1.31 (1.08, 1.59)* 1.39 (1.15, 1.69)*
Female
 All-cause mortality
  Health behavior trajectory
   Moderate risk 1.02 (0.89, 1.17) 1.05 (0.92, 1.21) 1.13 (0.98, 1.30) 1.21 (1.06, 1.39)* 1.25 (1.09, 1.44)*
   Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
   Low risk 0.98 (0.89, 1.08) 0.98 (0.89, 1.08) 0.98 (0.89, 1.07) 0.96 (0.87, 1.06) 0.95 (0.86, 1.04)
   Mild risk 1.01 (0.92, 1.11) 1.00 (0.91, 1.10) 1.00 (0.91, 1.09) 0.97 (0.89, 1.07) 0.87 (0.72, 1.05)
   Decreasing risk 0.83 (0.68, 1.00) 0.83 (0.69, 1.00) 0.82 (0.68, 0.99)* 0.80 (0.66, 0.96)* 0.90 (0.82, 0.99)*
   Severe risk 1.13 (0.88, 1.45) 1.17 (0.91, 1.49) 1.32 (1.03, 1.69)* 1.56 (1.22, 2.00)* 1.83 (1.43, 2.34)*
 Cancer cause mortality
  Health behavior trajectory
   Moderate risk 0.95 (0.75, 1.21) 0.96 (0.75, 1.23) 1.03 (0.80, 1.32) 1.12 (0.87, 1.43) 1.21 (0.94, 1.55)
   Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
   Low risk 1.01 (0.86, 1.19) 1.03 (0.86, 1.22) 1.03 (0.86, 1.22) 1.01 (0.85, 1.20) 1.01 (0.85, 1.20)
   Mild risk 1.16 (0.99, 1.35) 1.18 (1.00, 1.39) 1.16 (0.98, 1.37) 1.14 (0.97, 1.34) 1.09 (0.93, 1.28)
   Decreasing risk 1.04 (0.79, 1.37) 1.08 (0.81, 1.44) 1.07 (0.81, 1.43) 1.05 (0.79, 1.40) 1.14 (0.86, 1.52)
   Severe risk 1.16 (0.75, 1.79) 1.19 (0.74, 1.90) 1.39 (0.87, 2.22) 1.67 (1.05, 2.67)* 1.91 (1.20, 3.05)*
 CVD cause mortality
  Health behavior trajectory
   Moderate risk 1.17 (0.87, 1.58) 1.21 (0.90, 1.63) 1.29 (0.96, 1.74) 1.43 (1.08, 1.89)* 1.43 (1.08, 1.89)*
   Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
   Low risk 1.14 (0.93, 1.39) 1.14 (0.93, 1.39) 1.13 (0.92, 1.38) 1.07 (0.88, 1.29) 1.04 (0.86, 1.26)
   Mild risk 1.25 (1.04, 1.52)* 1.25 (1.03, 1.51)* 1.24 (1.02, 1.50)* 1.19 (1.00, 1.43) 1.11 (0.93, 1.33)
   Decreasing risk 0.93 (0.61, 1.40) 0.92 (0.61, 1.38) 0.89 (0.59, 1.34) 0.88 (0.60, 1.29) 0.98 (0.67, 1.45)
   Severe risk 1.70 (1.10, 2.62)* 1.75 (1.14, 2.71)* 1.97 (1.28, 3.04)* 2.25 (1.48, 3.41)* 2.68 (1.77, 4.05)*

Values are presented as hazard ratio (95% confidence interval); All covariates were controlled.

CVD, cardiovascular disease.

* p<0.05.

Figure & Data

References

    Citations

    Citations to this article as recorded by  
    • Combined health risk factors and all-cause and cause-specific mortality in the Korean National Health Examination Baseline cohort
      Jeong Min Yang, Geon Heo, Sung-il Cho, Wankyo Chung, Jieun Hwang
      Scientific Reports.2026;[Epub]     CrossRef

    Figure
    • 0
    • 1
    • 2
    • 3
    Related articles
    The impact of health behavior trajectories on all-cause and cause-specific mortality: an analysis using the National Health Insurance Service-National Health Screening Cohort in Korea
    Image Image Image Image
    Figure 1. Trajectories of health behaviors (A) tbacco use, (B) alcohol consumption, (C) PA, and (D) BMI among male participants. This figure illustrates 6 male health behavior trajectories, color-coded as follows: higher moderate-risk group (red, 19.0%; stable moderate smoking and alcohol use, normal BMI), higher severe-risk group (green, 1.1%; stable low smoking, obese BMI), stable healthy group (blue, 30.3%; non-smoking, low alcohol consumption, normal BMI), decreasing-risk group (black, 28.0%; concurrent reductions in smoking, alcohol consumption, and BMI with increasing PA), lower severe-risk group (yellow, 0.9%; stable low smoking, underweight BMI), and lower moderate-risk group (purple, 20.8%; stable moderate smoking and alcohol use, overweight BMI). BMI, body mass index; PA, physical activity.
    Figure 2. Trajectories of health behaviors (A) tbacco use, (B) alcohol consumption, (C) PA, and (D) BMI among female participants. This figure illustrates 6 female health behavior trajectories, color-coded as follows: moderate-risk group (red, 10.2%; stable low smoking and alcohol consumption, normal BMI), stable healthy group (green, 34.7%; non-smoking, no alcohol consumption, stable normal BMI), low-risk group (blue, 23.8%; non-smoking, no alcohol consumption, stable slightly overweight BMI), mild-risk group (black, 23.5%; non-smoking, no alcohol consumption, stable overweight BMI), decreasing-risk group (yellow, 6.8%; non-smoking, decreasing alcohol consumption and BMI with increasing PA), and severe-risk group (purple, 0.9%; stable low smoking and alcohol consumption, overweight BMI). BMI, body mass index; PA, physical activity.
    Figure 3. Stratified analysis of the association of health behavior trajectories (HBTs) with mortality risk by sex and age (A) male participants aged <65 years; (B) male participants aged ≥65 years; (C) female participants aged <65 years; (D) female participants aged ≥65 years. This figure illustrates age-stratified hazard ratios for all-cause, cancer, and cardiovascular disease (CVD) mortality across HBT groups, stratified by sex and age. Square, diamond, and triangle markers represent all-cause, cancer, and CVD mortality, respectively, and horizontal lines indicate 95% confidence intervals. The reference group for all comparisons is the stable healthy group.
    Graphical abstract
    The impact of health behavior trajectories on all-cause and cause-specific mortality: an analysis using the National Health Insurance Service-National Health Screening Cohort in Korea
    Characteristics Total All-cause mortality
    p-value (effect size)1 Cancer mortality
    p-value (effect size)1 CVD mortality
    p-value (effect size)1
    Death Death Death
     Disability <0.001 <0.001 <0.001
      Normal 87,387 (91.8) 7,205 (8.2) - 3,091 (3.5) - 1,128 (1.3) -
      Mild 6,041 (6.3) 967 (16.0) (0.22) 332 (5.5) (0.09) 194 (3.2) (0.13)
      Severe 1,798 (1.9) 465 (25.9) (0.44) 126 (7.0) (0.15) 96 (5.3) (0.23)
     CCI <0.001 <0.001 <0.001
      0 65,317 (68.6) 4,610 (7.1) - 1,964 (3.0) - 770 (1.2) -
      1 24,370 (25.6) 2,862 (11.7) (0.15) 1,141 (4.7) (0.08) 468 (1.9) (0.06)
      2 4,736 (5.0) 915 (19.3) (0.34) 323 (6.8) (0.17) 141 (3.0) (0.13)
      ≥3 803 (0.8) 250 (31.1) (0.59) 121 (15.1) (0.43) 39 (4.9) (0.22)
    Female
     Total 72,117 (100) 3,856 (5.3) 1,286 (1.8) 926 (1.3)
     Health behavior trajectory <0.001 <0.001 <0.001
      Moderate risk 7,382 (10.2) 330 (4.5) (0.43) 112 (1.5) (0.25) 72 (1.0) (0.20)
      Stable healthy 25,045 (34.7) 1,317 (5.3) - 409 (1.6) - 294 (1.2) -
      Low risk 17,198 (23.8) 921 (5.4) (0.47) 302 (1.8) (0.27) 217 (1.3) (0.23)
      Mild risk 16,916 (23.5) 1,060 (6.3) (0.51) 369 (2.2) (0.30) 287 (1.7) (0.26)
      Decreasing risk 4,927 (6.8) 147 (3.0) (0.35) 69 (1.4) (0.24) 31 (0.6) (0.16)
      Severe risk 649 (0.9) 81 (12.5) (0.72) 25 (3.9) (0.40) 25 (3.9) (0.40)
     Age (yr) <0.001 <0.001 <0.001
      <55 31,309 (43.4) 347 (1.1) - 202 (0.6) - 41 (0.1) -
      55-64 25,974 (36.0) 709 (2.7) (0.33) 373 (1.4) (0.24) 120 (0.5) (0.14)
      65-74 11,149 (15.5) 1,516 (13.6) (0.76) 478 (4.3) (0.42) 394 (3.5) (0.38)
      75-84 3,633 (5.0) 1,254 (34.5) (1.26) 228 (6.3) (0.51) 363 (10.0) (0.64)
      ≥85 52 (0.1) 30 (57.7) (1.73) 5 (9.6) (0.63) 8 (15.4) (0.81)
     Region <0.001 0.002 <0.001
      Capital area 29,987 (41.6) 1,375 (4.6) - 517 (1.7) - 317 (1.1) -
      Metropolitan area 15,872 (22.0) 742 (4.7) (0.44) 246 (1.5) (0.25) 174 (1.1) (0.21)
      Other areas 26,258 (36.4) 1,739 (6.6) (0.52) 523 (2.0) (0.28) 435 (1.7) (0.26)
     Medical insurance <0.001 <0.001 <0.001
      National Health Insurance (region) 19,272 (26.7) 1,426 (7.4) - 441 (2.3) - 365 (1.9) -
      National Health Insurance (workplace) 52,727 (73.1) 2,406 (4.6) (0.43) 840 (1.6) (0.25) 552 (1.0) (0.20)
      Medical Aid 118 (0.2) 24 (20.3) (0.94) 5 (4.2) (0.41) 9 (7.6) (0.56)
     Income level <0.001 0.051 0.025
      Low 16,979 (23.5) 1,019 (6.0) - 311 (1.8) - 249 (1.5) -
      Mid 24,640 (34.2) 1,201 (4.9) (0.49) 399 (1.6) (0.27) 319 (1.3) (0.24)
      High 30,498 (42.3) 1,636 (5.4) (0.45) 576 (1.9) (0.26) 358 (1.2) (0.23)
     Disability <0.001 <0.001 <0.001
      Normal 67,669 (93.8) 3,319 (4.9) - 1,148 (1.7) - 777 (1.1) -
      Mild 3,631 (5.0) 398 (11.0) (0.67) 115 (3.2) (0.36) 106 (2.9) (0.34)
      Severe 817 (1.1) 139 (17.0) (0.85) 23 (2.8) (0.34) 43 (5.3) (0.46)
     CCI <0.001 <0.001 <0.001
      0 44,009 (61.0) 1,891 (4.3) - 673 (1.5) - 431 (1.0) -
      1 21,672 (30.1) 1,335 (6.2) (0.50) 439 (2.0) (0.29) 325 (1.5) (0.25)
      2 5,472 (7.6) 470 (8.6) (0.59) 124 (2.3) (0.30) 121 (2.2) (0.30)
      ≥3 964 (1.3) 160 (16.6) (0.84) 50 (5.2) (0.46) 49 (5.1) (0.45)
    Variables 1-yr 3-yr 5-yr 7-yr Entire period
    Male
     All-cause mortality
      Health behavior trajectory
       Higher moderate risk 1.00 (0.93, 1.07) 1.06 (0.99, 1.13) 1.22 (1.14, 1.31)* 1.42 (1.33, 1.52)* 1.66 (1.56, 1.78)*
       Higher severe risk 0.88 (0.71, 1.09) 0.94 (0.76, 1.17) 1.17 (0.94, 1.45) 1.45 (1.17, 1.80)* 2.02 (1.63, 2.51)*
       Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
       Decreasing risk 0.99 (0.92, 1.06) 0.99 (0.92, 1.06) 0.99 (0.92, 1.05) 0.97 (0.91, 1.04) 0.91 (0.85, 0.97)*
       Lower severe risk 1.14 (0.97, 1.35) 1.21 (1.03, 1.43)* 1.46 (1.23, 1.72)* 1.79 (1.51, 2.11)* 2.11 (1.79, 2.49)*
       Lower moderate risk 0.98 (0.90, 1.06) 1.02 (0.94, 1.10) 1.14 (1.05, 1.23)* 1.26 (1.17, 1.37)* 1.37 (1.27, 1.48)*
     Cancer cause mortality
      Health behavior trajectory
       Higher moderate risk 1.11 (0.99, 1.24) 1.19 (1.06, 1.33)* 1.40 (1.25, 1.57)* 1.64 (1.47, 1.83)* 1.94 (1.74, 2.17)*
       Higher severe risk 1.06 (0.77, 1.46) 1.16 (0.84, 1.60) 1.47 (1.07, 2.02)* 1.87 (1.36, 2.57)* 2.54 (1.85, 3.48)*
       Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
       Decreasing risk 1.12 (1.00, 1.25) 1.11 (0.99, 1.24) 1.17 (1.00, 1.25) 1.11 (0.99, 1.24) 1.07 (0.95, 1.19)
       Lower severe risk 0.76 (0.54, 1.07) 0.82 (0.58, 1.16) 1.00 (0.71, 1.42) 1.25 (0.88, 1.76) 1.50 (1.06, 2.12)*
       Lower moderate risk 1.18 (1.05, 1.34)* 1.25 (1.10, 1.41)* 1.42 (1.25, 1.59)* 1.59 (1.41, 1.79)* 1.74 (1.54, 1.96)*
     CVD cause mortality
      Health behavior trajectory
       Higher moderate risk 0.97 (0.82, 1.16) 1.02 (0.85, 1.21) 1.16 (0.98, 1.38) 1.33 (1.12, 1.58)* 1.53 (1.29, 1.82)*
       Higher severe risk 1.46 (0.92, 2.30) 1.54 (0.98, 2.44) 1.87 (1.19, 2.95)* 2.28 (1.45, 3.60)* 3.17 (2.01, 4.99)*
       Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
       Decreasing risk 1.12 (0.95, 1.32) 1.11 (0.95, 1.31) 1.11 (0.94, 1.31) 1.08 (0.92, 1.27) 0.98 (0.84, 1.16)
       Lower severe risk 1.17 (0.78, 1.76) 1.23 (0.82, 1.85) 1.45 (0.96, 2.17) 1.74 (1.16, 2.62)* 2.00 (1.33, 3.01)*
       Lower moderate risk 1.05 (0.86, 1.28) 1.09 (0.89, 1.32) 1.20 (0.99, 1.46) 1.31 (1.08, 1.59)* 1.39 (1.15, 1.69)*
    Female
     All-cause mortality
      Health behavior trajectory
       Moderate risk 1.02 (0.89, 1.17) 1.05 (0.92, 1.21) 1.13 (0.98, 1.30) 1.21 (1.06, 1.39)* 1.25 (1.09, 1.44)*
       Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
       Low risk 0.98 (0.89, 1.08) 0.98 (0.89, 1.08) 0.98 (0.89, 1.07) 0.96 (0.87, 1.06) 0.95 (0.86, 1.04)
       Mild risk 1.01 (0.92, 1.11) 1.00 (0.91, 1.10) 1.00 (0.91, 1.09) 0.97 (0.89, 1.07) 0.87 (0.72, 1.05)
       Decreasing risk 0.83 (0.68, 1.00) 0.83 (0.69, 1.00) 0.82 (0.68, 0.99)* 0.80 (0.66, 0.96)* 0.90 (0.82, 0.99)*
       Severe risk 1.13 (0.88, 1.45) 1.17 (0.91, 1.49) 1.32 (1.03, 1.69)* 1.56 (1.22, 2.00)* 1.83 (1.43, 2.34)*
     Cancer cause mortality
      Health behavior trajectory
       Moderate risk 0.95 (0.75, 1.21) 0.96 (0.75, 1.23) 1.03 (0.80, 1.32) 1.12 (0.87, 1.43) 1.21 (0.94, 1.55)
       Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
       Low risk 1.01 (0.86, 1.19) 1.03 (0.86, 1.22) 1.03 (0.86, 1.22) 1.01 (0.85, 1.20) 1.01 (0.85, 1.20)
       Mild risk 1.16 (0.99, 1.35) 1.18 (1.00, 1.39) 1.16 (0.98, 1.37) 1.14 (0.97, 1.34) 1.09 (0.93, 1.28)
       Decreasing risk 1.04 (0.79, 1.37) 1.08 (0.81, 1.44) 1.07 (0.81, 1.43) 1.05 (0.79, 1.40) 1.14 (0.86, 1.52)
       Severe risk 1.16 (0.75, 1.79) 1.19 (0.74, 1.90) 1.39 (0.87, 2.22) 1.67 (1.05, 2.67)* 1.91 (1.20, 3.05)*
     CVD cause mortality
      Health behavior trajectory
       Moderate risk 1.17 (0.87, 1.58) 1.21 (0.90, 1.63) 1.29 (0.96, 1.74) 1.43 (1.08, 1.89)* 1.43 (1.08, 1.89)*
       Stable healthy 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference)
       Low risk 1.14 (0.93, 1.39) 1.14 (0.93, 1.39) 1.13 (0.92, 1.38) 1.07 (0.88, 1.29) 1.04 (0.86, 1.26)
       Mild risk 1.25 (1.04, 1.52)* 1.25 (1.03, 1.51)* 1.24 (1.02, 1.50)* 1.19 (1.00, 1.43) 1.11 (0.93, 1.33)
       Decreasing risk 0.93 (0.61, 1.40) 0.92 (0.61, 1.38) 0.89 (0.59, 1.34) 0.88 (0.60, 1.29) 0.98 (0.67, 1.45)
       Severe risk 1.70 (1.10, 2.62)* 1.75 (1.14, 2.71)* 1.97 (1.28, 3.04)* 2.25 (1.48, 3.41)* 2.68 (1.77, 4.05)*
    Table 1. General characteristics of health behavior trajectories and mortality among male and female participants at baseline (2008–2009)

    Values are presented as number (%).

    CVD, cardiovascular disease; CCI, Charlson comorbidity index.

    Effect sizes are presented as Cohen’s h.

    Table 2. Adjusted associations of health behavior trajectories on all-cause and cause-specific mortality among male and female participants

    Values are presented as hazard ratio (95% confidence interval); All covariates were controlled.

    CVD, cardiovascular disease.

    p<0.05.


    Epidemiol Health : Epidemiology and Health
    TOP