INTRODUCTION

Korea is aging faster than any other nation. As of 2023, there are 9.50 million individuals aged 65 or older, constituting 18.3% of the total population. This proportion is projected to rise to 25.5% by 2030 and 40.1% by 2050 [1]. One of the most significant health concerns among the older adult population is depression. The 2020 National Survey on the Elderly revealed that 13.5% of older adults experienced depressive symptoms [2]. One of the primary contributors to late-life depression is social and emotional isolation. As people age, they face an increased risk of alienation and loneliness due to a gradual disconnection from family and society. The proportion of older adults living alone has steadily increased, reaching 19.8% in 2020 [2]. Living alone strongly influences social isolation and loneliness [3], and numerous empirical studies have demonstrated that older adults living alone are more likely to develop depression than those residing with family members [4-8].

While the impact of living alone on depression in older adults is well documented, less attention has been paid to potential mitigating factors. One factor warranting exploration is the role of social welfare programs available in the community. According to the 2020 National Survey on the Elderly, the prevalence of depressive symptoms varied from 2.6% to 20.4% depending on the province of residence [2]. Studies have shown that higher social welfare expenditures for older adults in one’s area of residence can reduce the likelihood of depression [9] and that a larger proportion of the social welfare budget relative to the total budget is associated with lower depressive symptom levels among older adults [10]. As the senior welfare budget expands with increasing social welfare spending, opportunities for older adults to engage in social networks may improve through the provision of jobs, social services, and welfare facilities. These factors likely boost the use of social welfare services, social participation, interpersonal relationships, social support, and social capital—all of which have been shown to alleviate depressive symptoms in older adults [11-17].

The preventive effects of social welfare programs on depressive symptoms may be particularly strong for older adults living alone compared to those living with family members. Social welfare programs may better alleviate the social and emotional isolation experienced by those living alone, thereby reducing their risk of depressive symptoms to a greater extent. However, few empirical studies have examined whether these preventive effects are indeed more substantial among older adults living alone. Although a previous study explored a similar hypothesis, its findings were limited by being conducted solely among older adults in Seoul [6].

This study aimed to explore the role of community-level social welfare expenditures in depressive symptoms among older adults, with a specific focus on living arrangements. Three hypotheses are tested. First, older adults living alone are more likely to experience depressive symptoms than those who do not live alone. Second, higher per capita social welfare expenditure in a region is associated with a lower likelihood of depressive symptoms among its older residents. Third, the protective effect of social welfare expenditure on depressive symptoms is greater for older adults living alone than for those living with others.

METHODS

Data

We constructed multi-level data by linking individual-level and regional-level information. The age of 65—a cut-off used in previous studies to define the older adult population and to determine eligibility for benefits such as Customized Care Services for Older Adults—was employed to select our target population. We used data from the 2019 Community Health Survey for the individual-level analysis. This annual survey, conducted by the Korea Disease Control and Prevention Agency, targets all adults aged 19 and above residing in Korea. To ensure representativeness, approximately 900 participants are selected from each community through multi-stage stratification and random sampling, and the survey is administered face-to-face by trained interviewers [18,19]. In 2019, 229 099 individuals from 255 regions were surveyed, of whom 74 547 were aged 65 or older. After excluding cases with missing values in the variables used for analysis, the final sample comprised 73 226 individuals. For regional-level data, we used information from the e-Local Indicators and Health Insurance Statistics provided by the Korean Statistical Information Service as well as from the Community Health Survey. All data were from 2019 and matched the individual-level data. Because some regional-level indicators were available only for 223 local governments, the individual data were nested within these 223 local governments.

Dependent Variable

The Patient Health Questionnaire-9 (PHQ-9), a validated depression screening tool, served as the dependent variable. The reliability and validity of the PHQ-9 have been demonstrated in Korea [20]. The PHQ-9 score ranges from 0 to 27, with higher scores indicating more severe depressive symptoms. Typically, individuals are classified into 5 categories based on their total score: “none” (0-4 points), “mild” (5-9 points), “moderate” (10-14 points), “moderately severe” (15-19 points), and “severe” (20 points or above) [21]. Among the 73 226 participants, 59 105 (80.7%) were classified as “none,” 10 676 (14.6%) as “mild,” 2285 (3.1%) as “moderate,” 797 (1.1%) as “moderately severe,” and 363 (0.5%) as “severe.” Given that only 0.5% were classified as “severe” and 1.1% as “moderately severe,” these 2 categories were combined into a single “severe” category. Thus, the PHQ-9 was used as an ordinal scale with 4 levels: “none,” “mild,” “moderate,” and “severe.”

Individual-level Explanatory Variables

Among the individual-level explanatory variables, living arrangement was the variable of primary interest. Based on household survey data, living arrangement was categorized as either living alone (single-person household) or living with others. Control variables included gender, age, education level, income level, self-rated health, and the presence of hypertension or diabetes—factors that previous studies have consistently identified as significantly affecting depression in older adults [4-7,9,10,22,23]. Although previous studies often considered the number of chronic diseases, the 2019 Community Health Survey did not include prevalence data for various chronic conditions, so only the presence of hypertension and diabetes was used as a control. Age was categorized into 5-year intervals. Education was divided into 4 categories—less than elementary school, elementary school graduate, middle school graduate, and high school graduate or above—based on the distribution among older adults. Income was classified into 4 groups: less than 500 000 Korean won (KRW) per month, 500 000 KRW to less than 1 million KRW, 1 million KRW to less than 2 million KRW, and 2 million KRW or more. For self-rated health, the responses “very good” and “good” were combined into “good,” and “very bad” and “bad” into “bad,” resulting in 3 categories: “good,” “fair,” and “bad.”

Regional-level Explanatory Variables

The regional-level explanatory variable of primary interest was per capita social welfare expenditure in each municipality. Data on regional social welfare budgets were obtained from the Ministry of the Interior and Safety’s Local Finance Yearbook [24]. We chose per capita social welfare expenditure rather than the proportion of the social welfare budget because using the latter could underestimate program intensity in wealthy areas and overestimate it in poorer regions since the total budget is the denominator. Additional regional covariates included the local government’s financial independence rate, the number of leisure and welfare facilities for older adults, the number of psychiatric clinics, and a social cohesion score. Per capita social welfare expenditure was calculated by dividing the municipality’s social welfare budget by its population size. The number of leisure and welfare facilities for older adults was computed per 1000 older adults, while the number of psychiatric clinics was calculated per 100 000 individuals. Because the number of psychiatric clinics was right-skewed, its logarithm was used in regression analyses. The social cohesion score, used as an indicator of social capital, was derived from 2 questions in the 2019 Community Health Survey: “People in our neighborhood can be trusted” and “There is a tradition of mutual help among residents during life events in the neighborhood.” A “no” response was scored as 0 and a “yes” as 1; the total score for these 2 questions was then averaged for each municipality [22].

Statistical Analysis

Given the nested structure of our data and the ordinal nature of the dependent variable, we conducted a multi-level ordered logistic regression analysis. Before performing the multi-level analysis, chi-square tests were used to examine the relationship between categorical individual-level variables and depressive symptoms, while Pearson correlation coefficients were calculated for the continuous regional-level variables to assess their association with the mean PHQ-9 score in each region. The intra-class correlation was computed to confirm the need for a multi-level model. After this confirmation, we developed several models. Model 1 included all individual-level and regional-level variables (except the social cohesion score) to assess their effects on depressive symptoms, including living alone and per capita social welfare expenditure. Model 2 added a cross-level interaction term between living arrangements and per capita social welfare expenditure to examine whether the impact of increased social welfare expenditure on reducing depressive symptoms was greater among older adults living alone. Because of the high correlation between per capita social welfare expenditure and the social cohesion score (Pearson correlation coefficient=0.816; p<0.001, Figure 1), models 3 and 4 were created by adding the social cohesion index to models 1 and 2, respectively. Finally, model 5 was constructed by adding an interaction term between the social cohesion index and living alone to model 4 to test the robustness of the interaction effect. All statistical analyses were performed using Stata version 16.1 (StataCorp., College Station, TX, USA), with significance levels set at 0.05, 0.01, and 0.001.

Ethics Statement

The Seoul National University Hospital Institutional Review Board (IRB) exempted this study from IRB review (IRB No. E-2407-011-1548).

RESULTS

Table 1 presents the sample characteristics and the distribution of individual-level variables by the severity of depressive symptoms. Among the 73 226 participants, 80.7% exhibited no depressive symptoms, 14.6% had mild symptoms, 3.1% had moderate symptoms, and 1.6% had severe symptoms. Approximately 25.7% of the participants lived alone, and those living alone were significantly more likely to experience more severe depressive symptoms (p<0.001). Additionally, all control variables—including gender, age, education level, household income level, self-rated health status, and the presence of hypertension or diabetes—were significantly associated with depressive symptom severity (p<0.001).

Table 2 shows the distribution of regional-level variables. The average per capita social welfare expenditure across the 223 regions was 1.49 million KRW. The province (do) region (n=1) had 1.16 million KRW, cities (si) (n=74) averaged 1.29 million KRW (standard deviation [SD], 0.35), counties (gun) (n=80) averaged 2.02 million KRW (SD, 0.36), and districts (gu) (n=68) averaged 1.04 million KRW (SD, 0.23). The average social cohesion index across these regions was 1.47. The province (do) region scored 1.71, cities (si) averaged 1.41 (SD, 0.26), counties (gun) averaged 1.77 (SD, 0.11), and districts (gu) averaged 1.16 (SD, 0.19). Scatter plots of municipality-level per capita social welfare expenditure and the mean depressive symptom categories by living arrangement indicate that older adults living alone display greater variability in depressive symptom severity compared with those not living alone. Moreover, the negative association between per capita social welfare expenditure and depressive symptoms is more pronounced among those living alone (Figure 2).

The intra-class correlation coefficient from the null model indicated that 5.9% of the total variance in depressive symptoms was attributable to differences between regions, supporting the use of a multi-level model. Table 3 presents the results of the multi-level ordered logistic regressions. Model 1, which included individual-level and regional-level variables (excluding the social cohesion score), showed that living alone was associated with greater depressive symptom severity (OR, 1.07; p=0.011) and that higher per capita social welfare expenditure was significantly linked to lower depressive symptom severity (OR, 0.71; p=0.010). Most individual-level variables remained statistically significant, while the local government’s financial independence rate, the number of leisure and welfare facilities for older adults, and the number of psychiatric clinics were not significantly associated with depressive symptoms. In model 2, which added the cross-level interaction between living alone and per capita social welfare expenditure, the results demonstrated that the protective effect of social welfare expenditure was stronger among older adults living alone (OR, 0.92; p =0.046). An increase in social welfare expenditure significantly reduced depressive symptom severity (OR, 0.73; p =0.019), whereas other regional-level variables showed no significant associations. In model 3, which incorporated the social cohesion score into model 1, per capita social welfare expenditure was no longer significantly associated with depressive symptoms (OR, 0.85; p=0.234); however, a higher social cohesion score was significantly linked to lower depressive symptoms (OR, 0.50; p=0.001). Model 4, which added the social cohesion score to model 2, indicated that living alone remained associated with greater depressive symptom severity (OR, 1.22; p=0.006) and that the effect of social welfare expenditure continued to be more pronounced among older adults living alone (OR, 0.92; p=0.044). In contrast, model 5—which added an interaction term between the social cohesion score and living alone—revealed that the effect of social welfare expenditure did not differ according to living arrangement.

DISCUSSION

In this study, we examined the role of community-level social welfare expenditure in depressive symptoms among older adults, with a specific focus on differences according to living arrangements. Overall, our results support our 3 hypotheses. First, older adults living alone experienced more severe depressive symptoms compared with those living with others. Second, higher community social welfare expenditure was associated with reduced depressive symptom severity. Third, the protective effect of social welfare expenditure on depressive symptoms was stronger among older adults living alone than among those not living alone.

Although our study focused on per capita social welfare expenditure while controlling for other regional-level variables in a series of multi-level models, further theoretical and empirical work is needed to clarify the independent mechanisms through which social welfare expenditure affects depressive symptoms in older adults. Indeed, our results revealed a high correlation between social welfare expenditure and the social cohesion score, and including both measures in our models weakened the independent association between social welfare expenditure and depressive symptoms. The positive correlation between social welfare expenditure and social cohesion raises additional questions about their causal interrelationships, methodological challenges, and policy implications. First, social cohesion may capture some indirect effects of social welfare expenditure [25,26], as welfare programs often foster community cohesion by providing spaces and opportunities for social interaction that improve mental health outcomes. Second, social welfare expenditure itself may result from high levels of social cohesion rather than solely acting as its cause. Third, the high correlation between these 2 variables introduces multicollinearity, making it challenging to disentangle their independent effects. Beyond these methodological issues, an important policy implication is that communities with low social capital may also allocate less social welfare expenditure, thereby exacerbating health inequalities, particularly in mental health among older adults.

Our findings that living alone, older age, lower education, lower income, and poorer subjective health are associated with more severe depressive symptoms are consistent with previous research [4-7,9,10,22,23]. Likewise, the association between higher regional social welfare expenditure and reduced depressive symptom severity aligns with earlier studies [9,10]. One previous study that used the proportion of the social welfare budget as an indicator found no significant association with depressive symptoms [22]. A possible explanation for this discrepancy is that per capita social welfare expenditure may better reflect the intensity of welfare programs than the budget proportion. Moreover, although 2 prior studies reported that leisure welfare facilities for older adults reduced depressive symptom severity, our study did not observe this effect. This difference may be due to those studies focusing on older adults in a single large metropolitan area (Seoul [6] and Busan [27], respectively).

Our study suggests that the protective effect of regional social welfare expenditure on depressive symptoms may differ by living arrangement. However, this effect is not entirely robust. While models 3 and 4 demonstrated significant interaction effects between social welfare expenditure and living alone, model 5—after adding an interaction between the social cohesion score and living alone—eliminated these effects. This does not necessarily imply that the protective effects of social welfare expenditure are uniform across living arrangements; rather, the high correlation between social welfare expenditure and social cohesion introduces multicollinearity, leading to unstable estimates and complicating the isolation of each variable’s individual impact and its interaction with living alone.

Although our study indicates that the impact of social welfare expenditure may vary by living arrangement, findings in other contexts might differ. For example, a study focusing on older adults in Seoul did not observe such an association [6]. This inconsistency could be explained by the possibility that social welfare programs exert a greater preventive effect on depressive symptoms in non-metropolitan areas. Indirect evidence for this is provided by Lee and Kahng [28], who found that an increased proportion of the social welfare budget was associated with a decrease in the suicide rate among older adults, with a relatively greater impact in rural areas compared to urban ones.

The present study has important policy and practice implications. Non-medical interventions, such as enhanced social welfare policies, may improve depressive symptoms among older adults, particularly those living alone. Although the estimated effect sizes are modest, the finding that depressive symptom severity can be reduced through social welfare programs calls for increased attention to and investment in such interventions—especially given the growing population of older adults living alone.

This study has several limitations. First, our main explanatory variable—per capita social welfare expenditure—was calculated by dividing total regional social welfare spending by the population size; however, the proportion of this expenditure dedicated specifically to older adults may vary across regions. Even if total expenditure is similar, the types of welfare programs available for older adults can differ by region. Second, using social welfare expenditure as a macro-level indicator makes it difficult to pinpoint which specific programs contribute most to alleviating depressive symptoms. Third, the study’s design limits our ability to clearly delineate the causal relationships among living arrangements, social welfare expenditure, and depressive symptoms.

Despite these limitations, the study offers valuable insights into policy-amenable regional factors that influence depressive symptom severity among older adults in Korea. Our findings highlight the vulnerability of older adults living alone and the potential of social welfare programs to mitigate their depressive symptoms. The study also emphasizes the importance of addressing both individual- and community-level factors when considering the mental health needs of older adults. Moreover, our results suggest promising avenues for future research at the intersection of individual vulnerability and supportive social policies.

Notes

Conflict of Interest

The authors have no conflicts of interest associated with the material presented in this paper.

Funding

This study was supported by the Korean Society for Preventive Medicine, with a funding from the Korea Disease Control and Prevention Agency.

Acknowledgements

None.

Author Contributions

Conceptualization: Do YK, Lee JS, Yoon ES. Data curation: Lee JS. Formal analysis: Lee JS, Do YK. Funding acquisition: Lee JS. Methodology: Lee JS, Do YK. Writing – original draft: Lee JS. Writing – review & editing: Lee JS, Yoon ES, Song Y, Park S, Do YK.

Figure. 1.

The relationship between social welfare expenditure and social cohesion. The line represents a fitted linear regression, using municipality as the unit of analysis (n=223).

Figure. 2.

The relationship between social welfare expenditure and depressive symptoms based on living alone status (A) not liviing alone, (B) living alone. The vertical axis represents the municipality-level mean score of depressive symptom categories used in the main analysis (0: none, 1: mild, 2: moderate, 3: severe). The line in each scatter plot represents a fitted linear regression, using municipality as the unit of analysis (n=223).

REFERENCES

• 1. Statistics Korea. 2023 Elderly statistics. Daejeon: Statistics Korea; 2023. (Korean)
• 2. Lee YK, Kim SJ, Hwang NH, Lim JM, Joo BH, Namgung EH, et al. The 2020 national survey on the elderly. Sejong: Ministry of Health and Welfare; 2020. (Korean)
• 3. World Health Organization. Social isolation and loneliness among older people: advocacy brief; 2021 [cited 2024 Jul 18]. Available from: https://www.who.int/publications/i/item/9789240030749
• 4. Park JH, Kim KW, Kim MH, Kim MD, Kim BJ, Kim SK, et al. A nationwide survey on the prevalence and risk factors of late life depression in South Korea. J Affect Disord 2012;138(1-2):34-40. https://doi.org/10.1016/j.jad.2011.12.038ArticlePubMed
• 5. Fukunaga R, Abe Y, Nakagawa Y, Koyama A, Fujise N, Ikeda M. Living alone is associated with depression among the elderly in a rural community in Japan. Psychogeriatrics 2012;12(3):179-185. https://doi.org/10.1111/j.1479-8301.2012.00402.xArticlePubMed
• 6. Ha SC, Nam E. Community welfare resources and late-life depression among older adults in Seoul: a multi-level analysis. Int J Geriatr Psychiatry 2022;37(3):1-10. https://doi.org/10.1002/gps.5690Article
• 7. Do YK, Malhotra C. The effect of coresidence with an adult child on depressive symptoms among older widowed women in South Korea: an instrumental variables estimation. J Gerontol B Psychol Sci Soc Sci 2012;67(3):384-391. https://doi.org/10.1093/geronb/gbs033ArticlePubMed
• 8. Koo JH, Son N, Yoo KB. Relationship between the living-alone period and depressive symptoms among the elderly. Arch Gerontol Geriatr 2021;94: 104341. https://doi.org/10.1016/j.archger.2021.104341ArticlePubMed
• 9. Ko JE, Lee SH. A multilevel modeling of factors affecting depression among older Korean adults. Ment Health Soc Work 2012;40(1):322-351. (Korean)
• 10. Kim MI, Eo Y, Kim SE. A study of depression in the elderly by individual and community effects. Health Soc Welf Rev 2019;39(2):192-221. (Korean). https://doi.org/10.15709/hswr.2019.39.2.192Article
• 11. Lim JC, Ju KH, Lim BW. Effects of work participation on the depression of the elderly; focusing on the mediating role of self-esteem. Korean J Gerontol Soc Welf 2012;(57):29-50. (Korean). https://doi.org/10.21194/kjgsw..57.201209.29Article
• 12. Kim N, Jun MS. The influence of elderly social activity and depression on elderly well being. J Korea Contents Assoc 2017;17(3):496-506. (Korean). https://doi.org/10.5392/JKCA.2017.17.03.496Article
• 13. Kim MI, Lee E, Kim J. The effect of the use of social welfare services on life satisfaction among the elderly: multiple mediation effects of depression and self-rated health. J Soc Sci 2019;30(2):141-161. (Korean). http://dx.doi.org/10.16881/jss.2019.04.30.2.141Article
• 14. Park JH, Min S, Eoh Y, Park SH. The elderly living in single-person households in South Korea: a latent profile analysis of self-esteem, life satisfaction, and depression. Qual Life Res 2021;30(4):1083-1092. https://doi.org/10.1007/s11136-020-02693-1ArticlePubMed
• 15. Son H, Cho HJ, Cho S, Ryu J, Kim S. The moderating effect of social support between loneliness and depression: differences between the young-old and the old-old. Int J Environ Res Public Health 2022;19(4):2322. https://doi.org/10.3390/ijerph19042322ArticlePubMedPMC
• 16. Han KM, Han C, Shin C, Jee HJ, An H, Yoon HK, et al. Social capital, socioeconomic status, and depression in community-living elderly. J Psychiatr Res 2018;98: 133-140. https://doi.org/10.1016/j.jpsychires.2018.01.002ArticlePubMed
• 17. Lee HJ, Lee DK, Song W. Relationships between social capital, social capital satisfaction, self-esteem, and depression among elderly urban residents: analysis of secondary survey data. Int J Environ Res Public Health 2019;16(8):1445. https://doi.org/10.3390/ijerph16081445ArticlePubMedPMC
• 18. Korea Disease Control and Prevention Agency. Korea Community Health Survey [cited 2024 Jul 18]. Available from: https://chs.kdca.go.kr/chs/main.do (Korean)
• 19. Lee YK, Choi SH, Go YS, So SH, Park JH, Won JS, et al. Korea Community Health Survey 2019 raw data guideline. Cheongju: Korea Centers for Disease Control and Prevention; 2020. (Korean)
• 20. Choi HS, Choi JH, Park KH, Joo KJ, Ga H, Ko HJ, et al. Standardization of the Korean version of Patient Health Questionnaire-9 as a screening instrument for major depressive disorder. J Korean Acad Fam Med 2007;28(2):114-119. (Korean)
• 21. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med 2001;16(9):606-613. https://doi.org/10.1046/j.1525-1497.2001.016009606.xArticlePubMedPMC
• 22. Kim B. Individual and community-level factors affecting depressive symptoms among Korean older adults. J. Korean Soc Wellness 2020;15(4):695-706. (Korean). https://doi.org/10.21097/ksw.2020.11.15.4.695Article
• 23. Choi BJ, Lee JM. Effects of individual status and the community environment on geriatric depression: a multi-level analysis. Korean J Health Serv Manag 2021;15(3):51-65. (Korean). https://doi.org/10.12811/kshsm.2021.15.3.051Article
• 24. Korean Statistical Information Service. Local finance yearbook: the proportion of social welfare budget in the general account [cited 2023 Mar 3]. Available from: https://kosis.kr/statHtml/statHtml.do?orgId=101&tblId=DT_1YL20911&conn_path=I2 (Korean)
• 25. Gelissen JP, van Oorschot WJ, Finsveen E. How does the welfare state influence individuals’ social capital? Eurobarometer evidence on individuals’ access to informal help. Eur Soc 2012;14(3):416-440. https://doi.org/10.1080/14616696.2012.676660Article
• 26. Visser M, Gesthuizen M, Scheepers P. The crowding in hypothesis revisited: new insights into the impact of social protection expenditure on informal social capital. Eur Soc 2018;20(2):257-280. https://doi.org/10.1080/14616696.2018.1442928Article
• 27. Kim K, Lee M. Depressive symptoms of older adults living alone: the role of community characteristics. Int J Aging Hum Dev 2015;80(3):248-263. https://doi.org/10.1177/0091415015590315ArticlePubMed
• 28. Lee C, Kahng SK. The effects of social welfare budget on elderly suicide rate in local governments: focusing on the difference between urban and rural areas. Health Soc Welf Rev 2020;40(1):394-427. (Korean). https://doi.org/10.15709/hswr.2020.40.1.394Article

Figure & Data

References

Citations

Citations to this article as recorded by