Individual- and Community-Level Factors Associated with Healthy Lifestyle Practices among Low-Income Patients with Diabetes Mellitus: A Multilevel Analysis

INTRODUCTION

Globally, diabetes directly contributed to 1.6 million deaths in 2021, 47% of which occurred before 70 years of age (World Health Organization [WHO], 2024). This burden is also evident in South Korea, where approximately 5.33 million adults aged ≥30 years have diabetes and nearly 14.07 million have prediabetes (Park et al., 2025). Although the age-standardized diabetes mortality rate declined from 23.0 per 100,000 population in 2012 to 21.7 in 2024, advances in medical management have not eliminated the substantial socioeconomic and clinical burden associated with population aging (Statistics Korea, 2025). This burden is especially pronounced among socioeconomically disadvantaged groups, in whom disparities in glycemic control and mortality persist despite improvements in clinical care (Hill-Briggs et al., 2021). In South Korea, individuals in the lowest income quintile face a significantly greater risk of diabetes-related complications, underscoring the need for interventions tailored to the Korean sociocultural context (Choi et al., 2020).

Effective self-management is essential for achieving optimal glycemic control and reducing the risk of complications in individuals with diabetes (American Diabetes Association Professional Practice Committee, 2024). However, sustained self-management is strongly influenced by socioeconomic disparities, cultural norms, and environmental constraints (Hill-Briggs et al., 2021). Compared with higher-income groups, low-income populations have a higher prevalence of diabetes, poorer glycemic control(e.g., elevated HbA1c levels), and higher rates of complications and mortality (Rawshani et al., 2016). The persistence of these disparities indicates that education directed solely at individuals may be insufficient because low-income patients often face structural barriers that limit their ability to act on health information (Habibi Soola, Davari, & Rezakhani Moghaddam, 2022). Although South Korea's universal health insurance system provides broad medical coverage, income-related disparities and food insecurity may continue to restrict diabetes self-management and healthcare access among socioeconomically disadvantaged individuals (Park, Eggleston, Do, & Cutler, 2026). Recent nationwide evidence further indicates that socioeconomic and area-level deprivation remains an important determinant of adverse diabetes outcomes in South Korea (Choi et al., 2020). These disparities are exacerbated by systemic barriers, including limited access to affordable healthy foods, insufficient social support for lifestyle modifications, and structural inequities in healthcare access (Habibi Soola et al., 2022; Hill-Briggs et al., 2021).

Healthy lifestyle practices, including regular physical activity, a balanced diet, smoking cessation, and moderate alcohol consumption, are central to diabetes self-management. Adopting at least two such practices has been associated with a 40.9% lower risk of cardiovascular disease mortality (Liu et al., 2018). When combined with medication adherence, lifestyle modifications that improve physical activity and dietary behavior can enhance glycemic control and long-term health outcomes (Han et al., 2022).

The social determinants of health (SDOH) framework recognizes that health outcomes, including diabetes management, are shaped not only by individual factors but also by community resources, environmental constraints, and systemic policies (Dahlgren & Whitehead, 2006). Applying this framework to healthy lifestyle practices among low-income individuals with diabetes facilitates a broader understanding of the structural barriers and community-level disparities that perpetuate inequities (Hill-Briggs et al., 2021). For example, a qualitative study of low-income African American adults with type 2 diabetes found that neighborhood insecurity, limited recreational facilities, and socioeconomic inequality constrained physical activity, particularly among those with mobility limitations, who experienced additional safety concerns and social stigma (Park, Zachary, Gittelsohn, Quinn, & Surkan, 2020). These findings align with the SDOH framework, illustrating how community conditions, such as crime and inadequate infrastructure, can interact with individual socioeconomic circumstances to exacerbate health disparities. Although multilevel approaches to diabetes prevention and management are increasingly recognized as important, evidence on community- and neighborhood-level determinants remains limited relative to evidence on individual-level factors (Patel, Kornas, Gerstle, Diemert, & Rosella, 2025). Consequently, a substantial portion of existing diabetes research remains centered on individual variables and may overlook how community characteristics, such as healthcare-provider density or regional social capital, combine to create high-risk environments (Patel et al., 2025). Addressing this gap requires moving beyond the identification of at-risk individuals to examine environments that impede effective disease management.

Multilevel analysis allows individual- and community-level factors to be examined simultaneously, making it particularly valuable for investigating the social determinants of health (Hill-Briggs et al., 2021; Patel et al., 2025). This study therefore applied the SDOH framework (Dahlgren & Whitehead, 2006) to examine how individual-level factors and community-level determinants—including social and physical environment features, and healthcare system accessibility—collectively influence healthy lifestyle practices (physical activity, smoking, and alcohol consumption) among low-income patients with diabetes. By analyzing these determinants simultaneously, the study addresses an important gap in diabetes health disparities research and provides evidence to inform precision public health strategies. The findings may support the development of targeted multilevel interventions that address individual needs and community barriers, thereby informing policies and practice strategies to reduce health inequities and improve diabetes management in vulnerable populations.

METHODS

RESULTS

1. Sample Characteristics

The analytical sample comprised 1,673 low-income individuals with diabetes. Among them, 59.4% were women and 52.3% were aged ≥70 years. Most participants had a middle school education or lower (73.8%), were unmarried, separated, widowed, or divorced (71.4%), and were unemployed (85.3%). Regarding health-related characteristics, 42.5% had obesity (BMI ≥25 kg/m²), 59.5% slept 6~8 h daily, and 73.1% rated their health as unhealthy. In addition, 50.7% had lived with diabetes for ≥10 years, and 69.4% had comorbid hypertension (Table 1).

Table 1.

Differences in Healthy Lifestyle Practices according to Participants’ Demographic and Health-related Characteristics(N=1,673)

3. Multilevel Regression Analysis

1) Individual-level factors

In the multivariate analysis, several individual-level factors were significantly associated with healthy lifestyle practices. Men had higher levels of physical activity (β=.06, p=.014), smoking (β=.29, p<.001), and alcohol consumption (β=.24, p<.001) than women. Participants with a middle school education or lower were more likely to be physically active than those with a high school education or higher (β=.08, p=.005). Unemployed participants had lower levels of physical activity (β=.18, p<.001) and alcohol consumption (β=.11, p<.001) than employed participants. Compared with participants of normal weight, underweight participants had higher levels of physical activity (β=.15, p<.001) and smoking (β=.08, p=.008), whereas overweight (β=.06, p=.041) and obese (β=.10, p<.001) participants were more likely to smoke. Participants aged ≥70 years had higher levels of smoking (β=.30, p=.004) and alcohol consumption (β=.33, p=.008) than those aged 19~39 years. Unmarried, separated, widowed, or divorced participants smoked less than those who were married or living with a partner (β=-.12, p<.001). Frequent stress was associated with lower smoking (β=-.06, p=.002), whereas comorbid hypertension was associated with higher smoking (β=.04, p=.032). Depression, sleep duration, and self-reported health status were not significantly associated with any of the three behaviors (all p>.05) (Table 2).

Table 2.

Fixed Effects of the Linear Mixed-Effects Model

2) Community-level factors

The linear mixed-effects model revealed minimal variability in healthy lifestyle practices (physical activity, smoking, alcohol consumption) attributable to community-level factors. All nested community variables—including standardized mortality rate, unmet healthcare rates, green space availability, and density of bars/tobacco stores—had standard deviations (StdDev <0.00007), indicating negligible influence on health behaviors. Residual variance was substantially higher (StdDev: 0.362~0.457), suggesting that unmeasured individual-level factors or random effects predominantly drive behavioral differences (Table 3).

Table 3.

Random Effects of the Linear Mixed-Effects Model(N=1,673)

3) Model fit

Table 4 presents the fit indices for the linear mixed-effects models. The AIC and BIC were lowest for the smoking model (AIC=1,538.3, BIC=1,732.7), followed by alcohol consumption (AIC=2,113.8, BIC=2,308.1) and physical activity (AIC=2,295.7, BIC=2,490.0). The log-likelihood values were also highest (least negative) for the smoking model (logLik=-733.2), indicating that, among the three behaviors, the model for smoking demonstrated the best fit to the data (Table 4).

Table 4.

Model Fit(N=1,673)

DISCUSSION

Using data from the 2019 CHS and the SDOH framework, this study examined individual- and community-level factors associated with healthy lifestyle practices among low-income patients with diabetes. Whereas previous studies have largely focused on individual determinants, this study simultaneously examined contextual factors using a multilevel approach, thereby providing a more comprehensive understanding of health behaviors in this population. In South Korea, previous studies have primarily addressed individual socioeconomic disparities in diabetes outcomes and health behaviors (Choi et al., 2020; Lee, Han, & Kim, 2025), with limited consideration of contextual influences. By incorporating individual- and community-level factors within a single analytical framework, this study extends previous Korean research.

In this sample of 1,673 low-income patients with diabetes, 63.2% were physically inactive, 22.0% were current smokers, and 31.3% reported alcohol consumption during the past year. Individual-level factors—including sex, education, employment status, BMI, age, marital status, and stress—were significantly associated with healthy lifestyle practices. In contrast, the included community-level factors—standardized mortality rate, unmet medical care rate, green-space availability, and the densities of alcohol-serving establishments and tobacco retail stores—contributed little to the estimated variability in these behaviors. Within the specified models, individual-level determinants therefore accounted for most of the observed variation, emphasizing the importance of individual socioeconomic circumstances in the South Korean context.

Clear sex differences in health behaviors were observed. Women reported lower levels of physical activity than men. This finding is consistent with a meta-analysis of 65,344 patients, which found that men with diabetes engaged in 1.65 times more moderate-to-vigorous physical activity than women across age groups (Whipple et al., 2022). Studies of South Korean populations have likewise identified sex disparities in health behaviors, with women often encountering greater barriers to physical activity because of social roles and caregiving responsibilities (Kim, Kim, & Lee, 2024). These barriers may be especially consequential among low-income women, who may face combined constraints related to caregiving and limited access to supportive environments (Batalha, Marcal, Main, & Ghisi, 2025).

Smoking behavior also differed by sex. Although current smoking was more prevalent among men, women may experience a disproportionate smoking-related health burden, particularly for coronary heart disease. Women who smoke have been reported to exhibit higher nicotine dependence and more psychosocial barriers to smoking cessation, including stress and concerns about weight gain (Tramunt et al., 2023). These observations suggest that sex differences in health behavior reflect social and structural conditions as well as individual choices, supporting the need for sex-sensitive intervention strategies.

Education and employment emerged as significant predictors of healthy lifestyle practices among low-income individuals with diabetes. Previous South Korean studies have similarly reported that lower socioeconomic status, including lower education and income, is associated with poorer diabetes management and health behaviors (Choi et al., 2020; Park et al., 2020). Interestingly, lower educational attainment was associated with greater physical activity in the present study, contrary to evidence linking lower education with physical inactivity (Allet et al., 2016). Because previous South Korean studies have not consistently distinguished among domains of physical activity, the present findings may reflect differences in activity domains. Differences in health literacy, physical fitness, and engagement in health-promoting behaviors across educational groups may also contribute, as educational attainment has been associated with physical fitness among individuals with type 2 diabetes (Allet et al., 2016).

Unemployment was associated with lower physical activity, possibly because unemployed individuals have fewer structured daily routines, fewer opportunities for occupational activity, and more limited access to health-related resources (Shawahna, Jaber, Zmiro, & Kashkoush, 2024). In a Palestinian study, employed individuals with diabetes engaged in physical activity 2.3 times more frequently than unemployed individuals, supporting the possibility that employment facilitates healthier lifestyle practices (Shawahna et al., 2024).

Age and marital status were associated with smoking and alcohol consumption. Consistent with previous South Korean research, social and demographic factors, including marital status and social support, may shape health behaviors (Park, Prochnow, Amo, Curran, & Smith, 2023). Interestingly, older adults aged ≥70 years had higher levels of smoking and alcohol consumption than those aged 19~39 years. Although this finding contrasts with the assumption that health behaviors improve with age, risk behaviors may persist into older adulthood, particularly among socioeconomically disadvantaged populations. Such persistence may reflect the cumulative effects of socioeconomic disadvantage across the life course (Stringhini et al., 2010).

Because glycemic control may deteriorate and the risk of complications may increase with age (Templer, Abdo, & Wong, 2024), targeted smoking-cessation and alcohol reduction programs are particularly important for older, low-income populations with diabetes. Marital status also played a significant role. Unmarried individuals demonstrated higher levels of smoking than their married counterparts. This finding is consistent with previous research indicating that smoking-related health behaviors among individuals with diabetes are influenced by social and interpersonal factors, including family and partner support (Tramunt et al., 2023). In contrast, single individuals may lack such direct support, which can negatively influence their ability to maintain health-promoting behaviors. Moreover, this study found that participants who reported higher levels of stress were less likely to smoke. Because previous evidence generally indicates that psychosocial stress and adverse social determinants impede health-promoting behaviors and diabetes self-management (Hill-Briggs et al., 2021), this inverse association should be interpreted cautiously.

Underweight participants reported higher levels of physical activity than those of normal weight, whereas overweight and obese participants had higher levels of smoking. Previous research has generally associated overweight and obesity with lower engagement in recommended physical activity (Shawahna et al., 2024). The present findings indicate that BMI-related patterns differed across behaviors and should not be interpreted uniformly. Interventions for this population may therefore need to address both physical and psychosocial barriers to healthy behavior.

The standard deviations of the community-level variables were minimal, indicating that these factors explained little of the observed variability. Neighborhood environments can facilitate or hinder physical activity among adults with type 2 diabetes (Park et al., 2020); therefore, the limited community-level effects in this study should be interpreted cautiously. South Korea’s relatively small geographic area and limited regional variability may partly explain the findings. More importantly, aggregating 255 local administrative districts into 17 metropolitan and provincial units may have attenuated local contextual effects. Aggregated indicators may also fail to capture neighborhood-level exposures accurately. The weak estimated effects may therefore reflect measurement and aggregation limitations rather than the absence of meaningful community influences.

Among the three behaviors, the smoking model showed the best fit according to the reported indices.

Several limitations should be considered. Community indicators were aggregated at the municipal level (e.g., green space calculated as total park area per district). Such aggregation may obscure meaningful local variation and underestimate community effects. Additionally, the cross-sectional design of the study precludes any inference of causality between health behaviors and their determinants, limiting the ability to discern the directionality of observed associations. Future studies should refine the measurement of community-level factors by incorporating more granular geographic data, such as neighborhood-level (dong) variables or spatial analyses using geographic information systems (GIS), to better reflect individuals’ actual environments. Longitudinal designs are needed to evaluate how changes in community policies influence health behaviors over time. The analysis should also be replicated using more recent CHS data to capture post-COVID-19 behavioral patterns.

CONCLUSION

Healthy lifestyle practices among low-income patients with diabetes were more strongly associated with individual-level characteristics than with the community-level factors examined. These findings support the development of tailored interventions that account for differences in sex, age, education, employment status, and body mass index. However, the limited community-level effects should be interpreted cautiously because the use of aggregated regional indicators may have obscured local variation. Public health strategies should therefore prioritize individualized support while continuing to consider community barriers that may not have been fully captured in this study.

Acknowledgments

This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean Government (MSIT) (2021R1A2C2092656).

References

• Allet, L., Giet, O., Barral, J., Junod, N., Durrer, D., Amati, F., et al. (2016). Educational Level Is Related to Physical Fitness in Patients with Type 2 Diabetes-A Cross-Sectional Study. PloS One, 11(10), e0164176. [https://doi.org/10.1371/journal.pone.0164176]
• American Diabetes Association Professional Practice Committee. (2024). Introduction and methodology: Standards of care in diabetes—2024. Diabetes Care, 47(Supplement 1), S1-S4. [https://doi.org/10.2337/dc24-SINT]
• Angrist, J. D., & Pischke, J. S. (2009). Mostly harmless econometrics: An empiricist's companion. Princeton, NJ: Princeton university press. [https://doi.org/10.1515/9781400829828]
• Batalha, A. P. D. B., Marcal, I. R., Main, E., & Ghisi, G. L. D. M. (2025). Barriers to physical activity in women from ethnic minority groups: A systematic review. BMC Women's Health, 25(1), 330-21. [https://doi.org/10.1186/s12905-025-03877-y]
• Choi, D. W., Lee, S. A., Lee, D. W., Joo, J. H., Han, K. T., Kim, S., et al. (2020). Effect of socioeconomic deprivation on outcomes of diabetes complications in patients with type 2 diabetes mellitus: A nationwide population-based cohort study of South Korea. BMJ Open Diabetes Research & Care, 8(1), e000729. [https://doi.org/10.1136/bmjdrc-2019-000729]
• Craig, C. L., Marshall, A. L., Sjostrom, M., Bauman, A. E., Booth, M. L., Ainsworth, B. E., et al. (2003). International physical activity questionnaire: 12-country reliability and validity. Medicine and Science in Sports and Exercise, 35(8), 1381-1395. [https://doi.org/10.1249/01.mss.0000078924.61453.fb]
• Dahlgren, G., & Whitehead, M. (2006). European strategies for tackling social inequities in health: Levelling up part 2 (Report No. E89384). Copenhagen: WHO Regional Office for Europe. https://iris.who.int/bitstream/handle/10665/107791/E89384.pdf
• Habibi Soola, A., Davari, M., & Rezakhani Moghaddam, H. (2022). Determining the predictors of self-management behaviors in patients with type 2 diabetes: An application of socio-ecological approach. Frontiers in Public Health, 10, 820238. [https://doi.org/10.3389/fpubh.2022.820238]
• Han, H., Cao, Y., Feng, C., Zheng, Y., Dhana, K., Zhu, S., et al. (2022). Association of a healthy lifestyle with all-cause and cause-specific mortality among individuals with type 2 diabetes: A prospective study in UK Biobank. Diabetes Care, 45(2), 319-329. [https://doi.org/10.2337/dc21-1512]
• Hill-Briggs, F., Adler, N. E., Berkowitz, S. A., Chin, M. H., Gary-Webb, T. L., Navas-Acien, A., et al. (2021). Social determinants of health and diabetes: A scientific review. Diabetes Care, 44(1), 258-279. [https://doi.org/10.2337/dci20-0053]
• Kim, S., Kim, D. J., & Lee, H. (2024). Socioeconomic inequalities in the prevalence, non-awareness, non-treatment, and non-control of diabetes among South Korean adults in 2021. Plos One, 19(11), e0313988. [https://doi.org/10.1371/journal.pone.0313988]
• Korean Society for the Study of Obesity. (2022, December 9). Clinical practice guidelines for obesity 2022(8th ed.). Retrieved from https://general.kosso.or.kr/html/?pmode=BBBS0001300003&page=1&smode=view&seq=1383&searchValue=&searchTitle=strTitle
• Lee, Y. J., Han, K. D., & Kim, J. H. (2025). Nationwide study of income-based disparities in diabetic limb preservation in South Korea. Scientific Reports, 15(1), 43721. [https://doi.org/10.1038/s41598-025-27511-x]
• Liu, G., Li, Y., Hu, Y., Zong, G., Li, S., Rimm, E. B., et al. (2018). Influence of lifestyle on incident cardiovascular disease and mortality in patients with diabetes mellitus. Journal of the American College of Cardiology, 71(25), 2867-2876. [https://doi.org/10.1016/j.jacc.2018.04.027]
• Park, J. H., Prochnow, T., Amo, C., Curran, L., & Smith, M. L. (2023). Differences in physical activity, sedentary behavior, and mental health of the older population in South Korea based on marital status and gender. International Journal of Environmental Research and Public Health, 20(3), 1726. [https://doi.org/10.3390/ijerph20031726]
• Park, S. E., Ko, S. H., Kim, J. Y., Kim, K., Moon, J. H., Kim, N. H., et al. (2025). Diabetes fact sheets in Korea 2024. Diabetes and Metabolism Journal, 49(1), 24-33. [https://doi.org/10.4093/dmj.2024.0818]
• Park, S., Eggleston, K., Do, Y. K., & Cutler, D. M. (2026). Income-based inequalities in health system performance in the US and South Korea. JAMA Health Forum, 7(3), e260136. [https://doi.org/10.1001/jamahealthforum.2026.0136]
• Park, S., Zachary, W. W., Gittelsohn, J., Quinn, C. C., & Surkan, P. J. (2020). Neighborhood influences on physical activity among low-income African American adults with type 2 diabetes mellitus. The Diabetes Educator, 46(2), 181-190. [https://doi.org/10.1177/0145721720906082]
• Patel, R., Kornas, K., Gerstle, D., Diemert, L. M., & Rosella, L. C. (2025). Multilevel diabetes prevention Interventions to address population inequities in diabetes risk: Scoping review. JMIR Public Health and Surveillance, 11(1), e70267. [https://doi.org/10.2196/70267]
• Rawshani, A., Svensson, A. M., Zethelius, B., Eliasson, B., Rosengren, A., & Gudbjörnsdottir, S. (2016). Association between socioeconomic status and mortality, cardiovascular disease, and cancer in patients with type 2 diabetes. JAMA Internal Medicine, 176(8), 1146-1154. [https://doi.org/10.1001/jamainternmed.2016.2940]
• Shawahna, R., Jaber, M., Zmiro, A., & Kashkoush, S. (2024). Factors associated with physical inactivity among Palestinians with type 2 diabetes mellitus treated in resource-limited settings. Scientific Reports, 14(1), 11256. [https://doi.org/10.1038/s41598-024-60876-z]
• Statistics Korea. (2025, September 25). Trends in mortality rates by cause of death. Retrieved June 21, 2026, from https://www.index.go.kr/unity/potal/main/EachDtlPageDetail.do?idx_cd=F0051
• Stringhini, S., Sabia, S., Shipley, M., Brunner, E., Nabi, H., Kivimaki, M., et al. (2010). Association of socioeconomic position with health behaviors and mortality. JAMA, 303(12), 1159-1166. [https://doi.org/10.1001/jama.2010.297]
• Templer, S., Abdo, S., & Wong, T. (2024). Preventing diabetes complications. Internal Medicine Journal, 54(8), 1264-1274. [https://doi.org/10.1111/imj.16455]
• Tramunt, B., Rouland, A., Durlach, V., Vergès, B., Thomas, D., Berlin, I., et al. (2023). Smoking and diabetes: Sex and gender aspects and their effect on vascular diseases. Canadian Journal of Cardiology, 39(5), 681-692. [https://doi.org/10.1016/j.cjca.2023.01.018]
• Whipple, M. O., Pinto, A. J., Abushamat, L. A., Bergouignan, A., Chapman, K., Huebschmann, A. G., et al. (2022). Sex differences in physical activity among individuals with type 2 diabetes across the life span: A systematic review and meta-analysis. Diabetes Care, 45(9), 2163-2177. [https://doi.org/10.2337/dc22-0576]
• World Health Organization. (2024, November 14). Diabetes [Based on 2021 data]. Retrieved January 28, 2025, from https://www.who.int/news-room/fact-sheets/detail/diabetes