Analysis of Multi-Domain Factors Influencing Mental Health and Suicide: A Comparative Study of Urban and Rural Areas at City, County, and District Levels in Korea Authors: Changhyun Kim, Younglan Kim, Areum Lee, Woojung Lee, Yunmi Choi, Hyun Kim, Miji Lee, Jeong Hun Yang Published: 2025 ABSTRACT BackgroundThis study assessed social determinants in urban and rural nationalities to identify regional factors associated with mental health and suicide.MethodsData on 32 variables representing regional characteristics (sociodemographic, health and welfare, environmental, economic, and mental health status including suicide mortality) were collected at the municipal level (si/gun/gu) between 2017 and 2021. For the purpose of analysis, regions were categorized as urban (comprising “si” and “gu”) and rural (comprising “gun”); and correlations were examined between stress perception rate, depressive mood experience rate, prevalence of depressive symptoms, and suicide rates. Simple linear regression analysis and multiple linear regression analysis were done.ResultsStress perception, depressive mood experience, and depressive symptom prevalence were higher in urban areas whereas suicide rates were higher in rural areas. Positive correlations were observed between stress, depressive mood, and depressive symptoms in both urban and rural areas (P < 0.001). However, the relationship between stress and suicide showed contrasting patterns: a negative correlation in urban areas (P = 0.007) and a positive correlation in rural areas (P = 0.016). In urban areas, sociodemographic and health and welfare factors were primary correlates. Conversely, rural areas showed correlations with certain health and welfare, environmental, and economic factors.ConclusionThe findings reveal distinct patterns in the relationships between stress, depression, and suicide between urban and rural areas. These results underscore the need for tailored suicide prevention policies and mental health interventions that address the unique characteristics of each region. INTRODUCTION Suicidal behavior is a significant public health issue, with more than 700,000 people dying by suicide globally every year.12 Previous research has identified various risk factors, such as individual clinical factors (e.g. mental disorders), sociodemographic, psychosocial factors, and environmental factors.34 These interdependent factors highlight the challenges of prevention.56 Since the launch of the 1st Master Plan for Prevention of Suicide in 2004, Korean government has implemented suicide prevention policies by establishing five-year comprehensive plans as part of its efforts to reduce suicide rates. Although numerous studies have examined suicide risk factors at individual and community levels,7891011 no single factor can accurately assess suicide risk.12 The interplay between risk factors can lead to multicollinearity, in which certain variables appear to be statistically significant despite having minimal independent effects on suicide. Therefore, we emphasize the importance of building an integrated model of the social and environmental variables determined to constitute suicide risk factors in previous studies. Mental health determinants, including depression, anxiety, and stress, are regarded as the primary contributors to increased suicide risk.3413 Depression is an established factor strongly correlated with suicidal ideation and behavior.141516 Stress can also lead to feelings of hopelessness, helplessness, and suicidal ideation.17 Sociodemographic factors including mean age, sex ratio, and urbanicity influence regional suicide rates.18 Economic factors, such as unemployment and financial insecurity, are also associated with higher suicide rates. Regions with lower financial independence and higher divorce rates are more susceptible to suicide.19 Additionally, local health infrastructure, including mental health resources, significantly affects suicide rates, underscoring the critical role of regional healthcare availability.2021 These findings highlight the importance of integrating diverse factors to better understand the regional variations in suicide mortality rates. Although similar ecological studies have previously been conducted,222324 to the best of our knowledge, none have employed such an extensive range of variables to comprehensively evaluate regional contexts, especially differentiating between urban and rural areas at the municipal (si/gun/gu) level. Suicide rates differ between urban and rural areas.25 As urban and rural communities in South Korea exhibit distinct sociodemographic and socioenvironmental characteristics,26 suicide prevention strategies should identify regional risk factors through a detailed examination of these characteristics and tailor interventions to regional needs. This study aimed to classify regions in South Korea into urban and rural areas and identify the determinants of mental health status and suicide mortality rates within these contexts. These findings will contribute to tailoring suicide prevention policies that address regional needs. METHODS Suicide mortality Suicide mortality rates were sourced from the Cause of Death Statistics provided by Korea Statistics through the Korean Statistical Information Service for the five-year period from 2017 to 2021. The number of suicide deaths per 100,000 people was used in this study. Psychological factors related to suicide Our study used data from the Community Health Survey conducted by the Korea Disease Control and Prevention Agency to analyze the psychiatric factors in South Korea. This survey was conducted via one-on-one visits by trained surveyors to select households based on sample-provided data on stress perception rates, depressive mood experience rates, and prevalence of depressive symptoms. Perceived stress The stress perception rate indicated the percentage of people who reported experiencing subjective stress in daily life. This indicator is measured through surveys, capturing the percentage of respondents who put their stress level as “a lot” or “very much.” Perceived stress positively correlates with an increased risk of suicide.272829 Depressive mood experience The depressive mood experience rate measures the percentage of individuals who have experienced persistent depression for two weeks or more in the past year. This variable was measured using the Community Health Survey, in which respondents were asked whether they had felt sad, hopeless, or experienced similar emotions that hindered daily activities for at least two weeks in the past year. Prevalence of depressive symptoms The prevalence of depressive symptoms was measured using the Patient Health Questionnaire-9, a nine-item self-report tool with items rated on a scale of 0 to 3. A score of 10 or higher signifies clinically significant depressive symptoms. Its validity as a screening scale for predicting the risk of suicide attempts and mortality has been established.3031 This study classified respondents scoring 10 or higher on the Community Health Survey as having depressive symptoms. Other factors influencing suicide This study uses municipal-level data from the government and public institutions to compare regions at the municipal level across South Korea. From the literature review, independent variables that previously reported to be associated with mental health and suicide rates were selected: six sociodemographic indicators; twelve health and welfare indicators; four environmental indicators; and seven economic indicators, including metropolitan area status. After performing the multicollinearity test, marriage rate was removed, resulting in the use of 29 indicators for analysis (Table 1). Table 1List of variablesVariablesSourceSocio-demographicMean ageKOSTAT, 「Population and Housing Census」, retrieved from KOSISSex ratioMOIS, 「Resident Registration Population Status」, retrieved from KOSISProportion of single-person householdsKOSTAT, 「Population and Housing Census」, retrieved from KOSISCrude marriage rateKOSTAT, 「Vital Statistics」, retrieved from KOSISCrude divorce rateKOSTAT, 「Vital Statistics」, retrieved from KOSISTeacher-to-student ratioKorean Educational Development Institute, 「Basic Education Statistics」, retrieved from KOSISHealth and welfareRate of healthy lifestyle practicesKDCA, 「Community Health Survey」, retrieved from KOSISHigh-risk drinking rateKDCA, 「Community Health Survey」, retrieved from KOSISObesity prevalence rateKDCA, 「Community Health Survey」, retrieved from KOSISDrinking rateKDCA, 「Community Health Survey」, retrieved from KOSISProportion of social welfare budget in general accountingMOIS, 「Financial yearbook of local government」, retrieved from KOSISNo. of medical institutions per 1,000 populationNational Health Insurance Service, 「Statistical Yearbook on The Usage of Medical Services by Region」, MOIS, 「Resident Registration Population Status」, retrieved from KOSISNo. of hospital beds per 1,000 populationNational Health Insurance Service, 「Statistical Yearbook on The Usage of Medical Services by Region」, MOIS, 「Resident Registration Population Status」, retrieved from KOSISNo. of mental health institutions per 100,000 populationNational Health Insurance Service, 「Statistical Yearbook on The Usage of Medical Services by Region」, MOIS, 「Resident Registration Population Status」No. of senior leisure welfare facilities per 1,000 elderly populationMinistry of Health and Welfare, 「Status of Elderly Welfare Facilities」, MOIS, 「Resident Registration Population Status」, retrieved from KOSISNo. of in-home senior welfare facilities per 1,000 elderly populationNational Health Insurance Service, 「Long Term Care Insurance Statistical Yearbook」, MOIS, 「Resident Registration Population Status」, retrieved from KOSISNo. of social welfare volunteers per 1,000 populationKorea National Council on Social Welfare, 「Social Welfare Volunteering Statistical Yearbook」, MOIS, 「Resident Registration Population Status」Volunteer activity rateKorea National Council on Social Welfare, 「Social Welfare Volunteering Statistical Yearbook」EnvironmentGreen space ratioKorea Land and Housing Corporation, 「Urban Planning Statistics」, retrieved from KOSISNo. of cultural facilities per 100,000 populationMinistry of Culture, Sports and Tourism, 「National Cultural Infrastructure Comprehensive Bibliography」, MOIS, 「Resident Registration Population Status」, retrieved from KOSISAir pollution level (PM2.5)ME, 「Environmental Statistics Yearbook」, retrieved from KOSISAir pollution level (PM10)ME, 「Environmental Statistics Yearbook」, retrieved from KOSISEconomyGRDP per capitaKOSTAT (Income Statistics Division), MOIS, 「Resident Registration Population Status」, retrieved from KOSISFiscal independence rateMOIS, 「Fiscal Independence Rate」, retrieved from KOSISFiscal autonomy rateMOIS, 「Fiscal Autonomy Rate」, retrieved from KOSISNo. of businesses per 1,000 populationKOSTAT, 「Census on establishment 」, MOIS, 「Resident Registration Population Status」, retrieved from KOSISLand price index for residential areasKorea Real Estate Board, 「Land Price Change Rate Survey」, retrieved from R-ONELand price change rateKorea Real Estate Board, 「Land Price Change Rate Survey」, retrieved from R-ONEHousing supply rateKOSTAT, 「Population and Housing Census」, retrieved from KOSISKOSTAT = Statistics Korea, KOSIS = Korean Statistical Information Service, MOIS = Ministry of the Interior and Safety, KDCA = Korea Disease Control and Prevention Agency, ME = Ministry of Environment. List of variables KOSTAT = Statistics Korea, KOSIS = Korean Statistical Information Service, MOIS = Ministry of the Interior and Safety, KDCA = Korea Disease Control and Prevention Agency, ME = Ministry of Environment. Study setting Periods As the data release date varies from institution to institution, the variables analyzed in this study were based on the most recent five-year period (2017–2021). The time frame of 5 years was derived from the data available at the outset of the study. Urban and rural areas The South Korean regional classification was as follows: 17 provincial level local governments and 226 municipal level local governments. South Korean municipalities are classified as cities (si), counties (gun), or districts (gu). Administrative cities and districts (si, gu) are urban areas by definition and counties (gun) are rural. Multicollinearity The Variance Inflation Factor (VIF) values were determined to check for multicollinearity among the independent variables. Variables exhibiting VIF value greater than 10 were identified as having high multicollinearity and were removed from the model. Statistical analysis The collected data covered five years from 2017 to 2021 and were averaged over their annual values for use in the analysis. This methodology was employed to maintain data uniformity and reduce variability among variables. Statistical analysis was performed in two stages using SPSS 25 (IBM Corp., Armonk, NY, USA), with each step detailed as follows: Step 1. simple linear regression analysis A simple linear regression analysis was conducted to investigate the relationships between individual independent variables and mental health status. For each analysis, linear regression coefficient-parameters and a P value of variables were reported. Variables with a P value ≤ 0.05 were retained as covariates for the next step. Step 2. multiple linear regression analysis Multiple linear regression analysis was performed using covariates that demonstrated significant results in the simple linear regression analysis. Before finalizing the regression models, multicollinearity among covariates was assessed using VIF value. In this phase, we evaluated the simultaneous effects of several independent variables on suicide rates and analyzed the relative importance and combined effects of each variable. Ethics statement Ethical approval was obtained from our Institutional Review Board (IRB) of Chungnam National University Sejong Hospital (IRB No. 2024-09-002) before the start of the study. As the data used in this study were anonymized, no personal information was collected or used. Informed consent was waived. RESULTS Comparison of characteristics of areas In South Korea, 76 cities (si) and 69 districts (gu) are categorized as minicipal level local governments that correspond to urban areas. A total of 146 urban municipalities were divided into the Seoul Metropolitan Area (41.8%) and non-metropolitan areas (58.2%). Minicipal level local governments corresponding to rural areas are designated as “gun” (counties). However, we classified the Jeju Special Self-Governing Province as a single rural local government. Including the counties and Jeju Special Self-Governing Province, there are 82 rural municipalities; of which 6.1% are located in the Seoul Metropolitan Area and 93.9% in non-metropolitan areas (Table 2). Table 2Comparison of characteristics of areas (Seoul metropolitan and non-metropolitan status)AreasSeoul metropolitan regionNon-metropolitan regionTotalP valueUrban areas61 (41.8)85 (58.2)146 (100.0)< 0.001Rural areas5 (6.1)77 (93.9)82 (100.0)Values are presented as number (%).Fischer’s exact test was performed. Comparison of characteristics of areas (Seoul metropolitan and non-metropolitan status) Values are presented as number (%). Fischer’s exact test was performed. The baseline characteristics of the urban and rural areas were compared using χ2 tests and t-tests (Table 3). Among the sociodemographic factors, rural areas had a higher mean age (50.812 vs. 43.466, P < 0.001), sex ratio (102.030 vs. 99.594, P = 0.005), and proportion of single-person households (35.283 vs. 31.155, P < 0.001), whereas urban areas had higher rates of marriage (4.369 vs. 3.460, P < 0.001) and student-teacher ratios (15.342 vs. 8.961, P < 0.001). Table 3Comparison of characteristics of areas (other variables)VariablesUrban areas (n = 146)Rural areas (n = 82)T valueP valueAge43.466 ± 3.02150.812 ± 3.76916.0880.000*** Sex ratio99.594 ± 4.429102.030 ± 6.9252.8720.005** Proportion of single-person households31.155 ± 5.25435.283 ± 3.3817.2020.000*** Crude marriage rate4.369 ± 0.7753.460 ± 0.948−7.8270.000*** Crude divorce rate2.084 ± 0.3682.136 ± 0.2871.1890.236Teacher-to-student ratio15.342 ± 3.0788.961 ± 3.348−14.5520.000*** Rate of healthy lifestyle practices32.365 ± 7.53123.707 ± 5.323−10.1060.000*** High-risk drinking rate17.191 ± 2.61118.846 ± 3.3663.8490.000*** Obesity prevalence rate30.806 ± 2.62032.324 ± 2.8644.0580.000*** Drinking rate58.762 ± 2.59954.197 ± 3.621−10.0540.000*** Proportion of social welfare budget in general accounting42.437 ± 12.83121.199 ± 4.915−17.8080.000*** No. of medical institutions per 1,000 Population1.425 ± 0.8630.928 ± 0.206−6.6320.000*** No. of hospital beds per 1,000 Population16.177 ± 10.87914.250 ± 10.211−1.3000.195No. of mental health institutions per 100,000 population1.670 ± 1.2773.172 ± 1.6657.0800.000*** No. of senior leisure welfare facilities per 1,000 elderly population5.044 ± 3.87516.912 ± 6.02115.9970.000*** No. of in-home senior welfare facilities per 1,000 elderly population1.855 ± 1.1531.314 ± 0.441−5.0510.000*** No. of social welfare volunteers per 1,000 population19.102 ± 12.78015.064 ± 9.626−2.4910.013* Volunteer activity rate10.706 ± 2.03710.634 ± 3.086−0.1900.850Green space ratio59.188 ± 22.51874.378 ± 14.7016.1460.000*** No. of cultural facilities per 100,000 population6.358 ± 5.27318.876 ± 10.55310.0600.000*** Air pollution level (PM2.5)22.017 ± 2.81418.384 ± 3.493−8.0610.000*** Air pollution level (PM10)39.376 ± 4.26833.219 ± 5.041−9.3390.000*** GRDP per capita40.591 ± 47.05834.331 ± 13.692−1.4980.136Fiscal independence rate25.892 ± 12.08911.365 ± 6.067−12.0640.000*** Fiscal autonomy rate50.878 ± 11.79758.584 ± 4.1137.1560.000*** No. of businesses per 1,000 population103.336 ± 53.981107.252 ± 15.0810.8210.413Land price index for residential areas89.221 ± 3.58691.512 ± 2.5665.5650.000*** Land price change rate3.710 ± 1.3242.832 ± 0.799−6.2360.000*** Housing supply rate88.944 ± 10.881104.682 ± 6.98013.2770.000*** Perceived stress26.724 ± 2.90723.405 ± 3.578−7.1750.000*** Depressive mood experience6.189 ± 1.4534.911 ± 1.570−6.0520.000*** Prevalence of depressive symptoms3.169 ± 0.9612.478 ± 0.922−5.3490.000*** Suicide mortality26.996 ± 5.03031.831 ± 5.6566.4420.000*** Values are presented as mean ± standard deviation.*P < 0.05, **P < 0.01, ***P < 0.001. Comparison of characteristics of areas (other variables) Values are presented as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001. When comparing health and welfare factors between urban and rural areas, urban areas showed higher healthy lifestyle practices rates (32.365 vs. 23.707, P < 0.001), drinking rates (58.762 vs. 54.197, P < 0.001), proportion of social welfare budget in general accounting (42.437 vs. 21.199, P < 0.001), number of medical institutions per 1,000 population (1.425 vs. 0.928, P < 0.001), number of senior leisure welfare facilities per 1,000 elderly population (1.855 vs. 1.314, P < 0.001), and number of social welfare volunteers per 1,000 population (19.102 vs. 15.064, P = 0.013). In contrast, rural areas had higher high-risk drinking rates (18.846 vs. 17.191, P < 0.001), obesity prevalence rate (32.324 vs. 30.806, P < 0.001), number of mental health institutions per 100,000 population (3.172 vs. 1.670, P < 0.001), and number of senior leisure welfare facilities per 1,000 elderly population (16.912 vs. 5.044, P < 0.001). Regarding environmental factors, rural areas had a higher green space ratio (74.378 vs. 59.188, P < 0.001) and number of cultural facilities per 100,000 population (18.876 vs. 6.358, P < 0.001), whereas urban areas had higher levels of air pollution level (PM2.5: 22.017 vs. 18.384, P < 0.001; PM10: 39.376 vs. 33.219, P < 0.001). Regarding economic factors, urban areas showed a higher fiscal independence rate (25.892 vs. 11.365, P < 0.001) and land price change rate (3.710 vs. 2.832, P < 0.001), whereas rural areas had higher fiscal autonomy rate (58.584 vs. 50.878, P < 0.001), residential land price index for residential areas (91.512 vs. 89.221, P < 0.001), and housing supply rate (104.682 vs. 88.944, P < 0.001). Comparison of psychological factors of urban and rural areas When comparing the dependent variables between urban and rural areas, urban regions exhibited higher stress perception rates (26.724 vs. 23.405, P < 0.001), depressive mood experience rates (6.189 vs. 4.911, P < 0.001), and depression symptom prevalence rates (3.169 vs. 2.478, P < 0.001), whereas the suicide mortality rate (31.831 vs. 26.996, P < 0.001) was higher in rural areas (Table 3). To examine differences based on whether an area was located in the Seoul metropolitan area, we analyzed the dependent variables by categorizing the areas into Seoul metropolitan cities, Seoul metropolitan rural areas, non-metropolitan cities, and non-metropolitan rural areas (Fig. 1). The stress perception rate (28.479 vs. 25.464 vs. 24.940 vs. 23.305), depressive mood experience rate (6.614 vs. 5.884 vs. 5.304 vs. 4.886), and depression symptom prevalence rate (3.435 vs. 2.977 vs. 2.544 vs. 2.473) were the highest in Seoul metropolitan cities, followed by non-metropolitan cities, Seoul metropolitan rural areas, and non-metropolitan rural areas. Conversely, suicide rate (31.965 vs. 29.775 vs. 29.026 vs. 24.168) followed the opposite trend, being highest in non-metropolitan rural areas, followed by Seoul metropolitan rural areas, non-metropolitan cities, and Seoul metropolitan cities. Comparison of mean values of dependent variables between urban and rural areas. In urban areas, stress perception, depressive mood experience, and depression symptom prevalence rates were positively correlated (P < 0.001). However, suicide rate showed a negative correlation with stress perception rate (P = 0.007) and no significant correlation with depressive mood experience rate (P = 0.301) or depression symptom prevalence (P = 0.176) (Table 4). Table 4Correlation among independent variables (urban areas, N = 146)VariablesStressDepressive moodDepressive symptomsSuicideStress----Depressive mood0.575*** (0.000)---Depressive symptoms0.585*** (0.000)0.691*** (0.000)--Suicide−0.221** (0.007)0.086 (0.301)0.113 (0.176)-Values are presented as r (P).**P < 0.01, ***P < 0.001. Correlation among independent variables (urban areas, N = 146) Values are presented as r (P). **P < 0.01, ***P < 0.001. Similarly, in rural areas, stress perception, depressive mood experience, and depression symptom prevalence rates were positively correlated (P < 0.001). However, unlike in urban areas, the suicide rate in rural regions showed a positive correlation with both the stress perception rate (P = 0.016) and depressive mood experience rate (P = 0.028). Depression symptom prevalence did not show a statistically significant relationship in rural areas, as was the case in urban areas (P = 0.165) (Table 5). Table 5Correlation among independent variables (rural areas, N = 82)VariablesStressDepressive moodDepressive symptomsSuicideStress----Depressive mood0.656*** (0.000)---Depressive symptoms0.652*** (0.000)0.726*** (0.000)--Suicide0.266* (0.016)0.243* (0.028)0.155 (0.165)-Values are presented as r (P).*P < 0.05, ***P < 0.001. Correlation among independent variables (rural areas, N = 82) Values are presented as r (P). *P < 0.05, ***P < 0.001. Additionally, in urban areas, a lower mean age was associated with higher stress levels (R2 = 0.113) but a decrease in the suicide rate (R2 = 0.362). However, in rural areas, a lower mean age was also associated with higher stress levels (R2 = 0.121), but the effect on suicide rate was minimal (R2 = 0.017) (Supplementary Figs. 1 and 2). Results of linear regression analysis The linear regression analysis between the stress perception rate, depressive mood experience rate, depression symptom prevalence rate, and suicide rate in urban areas with independent variables is summarized as follows (Table 6). Table 6Examining independent variables, using single and multiple linear regression analysis (urban areas)VariablesStressDepressive moodDepressive symptomsSuicideSingleMultipleSingleMultipleSingleMultipleSingleMultipleSeoul metropolitan area statusa 0.513*** (< 0.001)0.357** (0.001)0.249** (0.002)0.183* (0.042)0.236** (0.004)0.193* (0.036)−0.478*** (< 0.001)−0.176* (0.039)Mean age−0.337*** (< 0.001)−0.212* (0.038)−0.061 (0.465)0.011 (0.895)0.602*** (< 0.001)0.387*** (< 0.001)Sex ratio0.058 (0.486)0.174* (0.035)−0.029 (0.807)0.018** (0.003)0.078 (0.487)0.283** (0.001)0.074 (0.452)Proportion of single-person households−0.113 (0.175)−0.058 (0.489)0.102 (0.219)0.388*** (< 0.001)0.167* (0.031)Crude divorce rate0.005(0.953)0.241** (0.003)0.036 (0.768)0.153 (0.065)0.577*** (< 0.001)0.384*** (< 0.001)Teacher-to-student ratio0.237** (0.004)0.079 (0.286)0.120 (0.150)0.163* (0.049)0.114 (0.146)−0.174* (0.036)<0.001 (0.994)Rate of healthy lifestyle practices0.245** (0.003)−0.056 (0.618)0.014 (0.867)−0.032 (0.701)−0.546*** (< 0.001)−0.202* (0.018)High-risk drinking rate−0.072 (0.391)0.153 (0.065)0.144 (0.084)0.573*** (< 0.001)0.078 (0.323)Obesity prevalence rate0.148 (0.075)0.343*** (< 0.001)0.317** (0.010)0.282** (0.001)0.231* (0.030)0.433*** (< 0.001)−0.028 (0.745)Drinking rate0.185* (0.025)0.201* (0.012)0.101 (0.227)0.159 (0.055)−0.256** (0.002)−0.051 (0.365)Proportion of social welfare budget in general accounting0.129 (0.120)−0.021 (0.802)−0.024 (0.775)−0.315*** (< 0.001)0.107 (0.331)No. of medical institutions per 1,000 population0.068 (0.411)−0.081 (0.329)−0.029 (0.731)−0.126 (0.129)No. of hospital beds per 1,000 population−0.143 (0.085)−0.082 (0.327)−0.003 (0.972)0.321*** (< 0.001)0.016 (0.801)No. of mental health institutions per 100,000 population−0.060 (0.469)−0.077 (0.358)0.047 (0.575)0.226** (0.006)0.006 (0.925)No. of senior leisure welfare facilities per 1,000 elderly population−0.224** (0.006)0.110 (0.351)−0.008 (0.922)0.038 (0.646)0.452*** (< 0.001)0.001 (0.991)No. of in-home senior welfare facilities per 1,000 elderly population0.034 (0.685)0.095 (0.256)−0.006 (0.946)−0.103 (0.215)No. of social welfare volunteers per 1,000 population−0.082 (0.325)−0.208* (0.012)−0.093 (0.243)−0.069 (0.407)0.043 (0.602)Volunteer activity rate−0.223** (0.007)−0.216** (0.006)−0.199* (0.016)−0.117 (0.155)−0.185* (0.026)−0.143 (0.088)−0.171* (0.039)−0.115* (0.031)Green space ratio−0.131 (0.114)−0.123 (0.140)−0.120 (0.150)0.125 (0.131)No. of cultural facilities per 100,000 population−0.101 (0.225)−0.081 (0.328)0.054 (0.514)0.297*** (< 0.001)−0.027 (0.659)Air pollution level (PM2.5)0.444*** (< 0.001)0.161 (0.241)0.251** (0.002)0.264 (0.078)0.237** (0.004)0.277 (0.070)−0.101 (0.224)Air pollution level (PM10)0.435*** (< 0.001)0.037 (0.792)0.206* (0.013)−0.174 (0.265)0.178* (0.032)−0.231 (0.143)−0.157 (0.058)GRDP per capita0.094 (0.259)−0.032 (0.701)−0.008 (0.926)−0.126 (0.129)Fiscal independence rate0.346*** (< 0.001)−0.060 (0.560)0.090 (0.281)0.024 (0.776)−0.521*** (< 0.001)0.006 (0.955)Fiscal autonomy rate0.091 (0.272)0.064 (0.442)0.046 (0.585)0.021 (0.805)No. of businesses per 1,000 population0.004 (0.966)−0.034 (0.687)−0.003 (0.972)0.083 (0.319)Land price index for residential areas−0.267** (0.001)−0.034 (0.818)0.047 (0.576)0.046 (0.580)0.488*** (< 0.001)−0.078 (0.433)Land price change rate0.208* (0.012)−0.030 (0.828)−0.085 (0.308)−0.060 (0.475)−0.485*** (< 0.001)−0.087 (0.402)Housing supply rate−0.256** (0.002)−0.064 (0.545)−0.004 (0.962)−0.059 (0.477)0.366*** (< 0.001)0.001 (0.988)Values are presented as β (P).aSeoul metropolitan regions = 1, Non-metropolitan regions = 2.*P < 0.05, **P < 0.01, ***P < 0.001. Examining independent variables, using single and multiple linear regression analysis (urban areas) Values are presented as β (P). aSeoul metropolitan regions = 1, Non-metropolitan regions = 2. *P < 0.05, **P < 0.01, ***P < 0.001. For stress perception rate, the simple linear regression model showed significant positive correlations with the following variables: Seoul metropolitan area status (β = 0.513, P < 0.001), teacher-to-student ratio (β = 0.237, P = 0.004), healthy lifestyle practices rate (β = 0.245, P = 0.003), drinking rate (β = 0.185, P = 0.025), PM2.5 (β = 0.444, P < 0.001), PM10 (β = 0.435, P < 0.001), fiscal independence rate (β = 0.346, P < 0.001), and land price change rate (β = 0.208, P = 0.012). Negative correlations were observed between mean age (β = −0.337, P < 0.001), senior leisure welfare facilities per 1,000 older adults (β = −0.224, P = 0.006), volunteer activity rate (β = −0.223, P = 0.007), residential land price index (β = −0.267, P = 0.001), and housing supply rate (β = −0.256, P = 0.002). In the subsequent multiple linear regression analysis, Seoul metropolitan area status (β = 0.357, P = 0.001) and drinking (β = 0.201, P = 0.012) maintained significant positive correlations, while mean age (β = −0.212, P = 0.038) and volunteer activity rate (β = −0.216, P = 0.006) showed significant negative correlations. For depressive mood experience rate, the simple linear regression results showed positive correlations with Seoul metropolitan area status (β = 0.249, P = 0.002), sex ratio (β = 0.174, P = 0.035), crude divorce rate (β = 0.241, P = 0.001), obesity rate (β = 0.343, P < 0.001), PM2.5 (β = 0.251, P = 0.002), and PM10 (β = 0.206, P = 0.013). Negative correlations were found between depressive mood experience rate and the number of volunteers per 1,000 people (β = −0.208, P = 0.012), and volunteer activity rates (β = −0.199, P = 0.016). In the multiple linear regression analysis, Seoul metropolitan area status (β = 0.183, P = 0.042) and obesity rate (β = 0.317, P = 0.010) remained positively correlated with depressive mood experience rate. For depression symptom prevalence rate, significant positive correlations were found with Seoul metropolitan area status (β = 0.236, P = 0.004), sex ratio (β = 0.018, P = 0.003), student—teacher ratio (β = 0.163, P = 0.049), obesity rate (β = 0.282, P = 0.001), PM2.5 (β = 0.237, P = 0.004), and PM10 (β = 0.178, P = 0.032). A negative correlation was found with volunteer activity rate (β = −0.185, P = 0.026). In the multiple linear regression model, Seoul metropolitan area status (β = 0.193, P = 0.036) and obesity rate (β = 0.231, P = 0.030) remained significant positive predictors of depression symptom prevalence. For suicide rate, the simple linear regression showed positive correlations with the following variables: Seoul metropolitan area status (β = −0.478, P < 0.001), mean age (β = 0.602, P < 0.001), sex ratio (β = 0.283, P = 0.001), proportion of single-person households (β = 0.388, P < 0.001), crude divorce rate (β = 0.577, P < 0.001), teacher-to-student ratio (β = −0.174, P = 0.036), healthy lifestyle practices rate (β = −0.546, P < 0.001), high-risk drinking rate (β = 0.573, P < 0.001), obesity rate (β = 0.433, P < 0.001), drinking rate (β = −0.256, P = 0.002), social welfare budget share (β = −0.315, P < 0.001), hospital beds per 1,000 people (β = 0.321, P < 0.001), mental health institutions per 100,000 people (β = 0.226, P = 0.006), older adult leisure facilities per 1,000 older adults (β = 0.452, P < 0.001), volunteer activity rate (β = −0.171, P = 0.039), cultural facilities per 100,000 people (β = 0.297, P < 0.001), fiscal independence rate (β = −0.521, P < 0.001), residential land price index (β = 0.488, P < 0.001), land price change rate (β = −0.485, P < 0.001), and housing supply rate (β = 0.366, P < 0.001). In the multiple regression analysis, Seoul metropolitan area status (β = −0.176, P = 0.039), mean age (β = 0.387, P < 0.001), proportion of single-person households (β = 0.167, P = 0.031), crude divorce rate (β = 0.384, P < 0.001), healthy lifestyle practice rate (β = −0.202, P = 0.018), and volunteer activity rate (β = −0.115, P < 0.031) were found to have significant correlations with suicide rate. The correlation analysis between the dependent and independent variables for rural areas is summarized as follows (Table 7). Table 7Examining independent variables, using single and multiple linear regression (rural areas)VariablesStressDepressive moodDepressive symptomsSuicideSingleMultipleSingleMultipleSingleMultipleSingleMultipleSeoul metropolitan area statusa 0.110 (0.325)0.064 (0.567)0.018 (0.869)−0.093 (0.405)Mean age−0.348** (0.001)0.169 (0.358)−0.161 (0.149)−0.025 (0.826)0.131 (0.242)Sex ratio0.209 (0.060)0.047 (0.674)−0.034 (0.763)−0.012 (0.918)Proportion of single-person households−0.220* (0.047)−0.047 (0.734)−0.247* (0.026)0.078 (0.557)−0.062 (0.583)0.053 (0.637)Crude divorce rate0.166 (0.137)0.031 (0.783)0.013 (0.905)−0.032 (0.776)Teacher-to-student ratio0.375** (0.001)0.300* (0.024)0.251* (0.023)−0.088 (0.489)0.169 (0.130)−0.024 (0.829)Rate of healthy lifestyle practices−0.282 (0.282)−0.173 (0.120)−0.351** (0.001)−0.381*** (< 0.001)−0.249* (0.024)−0.061 (0.565)High-risk drinking rate0.261* (0.018)0.103 (0.414)0.110 (0.327)0.092 (0.413)0.227* (0.041)0.011 (0.920)Obesity prevalence rate0.431*** (< 0.001)0.438** (0.005)0.220* (0.047)0.152 (0.147)0.111 (0.320)0.359** (0.001)0.325** (0.005)Drinking rate0.339** (0.002)0.155 (0.273)0.174 (0.118)0.117 (0.296)0.183 (0.100)Proportion of social welfare budget in general accounting0.259* (0.019)0.159 (0.353)0.334** (0.002)0.265 (0.059)0.203 (0.067)−0.107 (0.337)No. of medical institutions per 1,000 population0.153 (0.170)0.210 (0.058)0.191 (0.085)−0.102 (0.360)No. of hospital beds per 1,000 population−0.004 (0.973)0.042 (0.713)−0.024 (0.835)−0.205 (0.068)No. of mental health institutions per 100,000 population−0.040 (0.719)0.030 (0.786)0.041 (0.713)0.036 (0.745)No. of senior leisure welfare facilities per 1,000 elderly population−0.221* (0.048)0.225 (0.120)−0.101 (0.370)0.020 (0.858)−0.156 (0.166)No. of in-home senior welfare facilities per 1,000 elderly population0.019 (0.864)0.123 (0.271)0.126 (0.259)−0.048 (0.670)No. of social welfare volunteers per 1,000 population−0.041 (0.712)−0.007 (0.953)−0.019 (0.863)−0.073 (0.514)Volunteer activity rate−0.065 (0.563)−0.043 (0.702)0.109 (0.330)−0.004 (0.973)Green space ratio−0.151 (0.174)−0.005 (0.968)0.058 (0.607)−0.001 (0.994)No. of cultural facilities per 100,000 population−0.048 (0.667)−0.163 (0.143)−0.123 (0.272)0.113 (0.311)Air pollution level (PM2.5)0.473*** (< 0.001)0.091 (0.651)0.495*** (< 0.001)0.497* (0.014)0.305** (0.005)0.519* (0.011)0.175 (0.116)Air pollution level (PM10)0.456*** (< 0.001)0.122 (0.534)0.431*** (< 0.001)−0.077 (0.713)0.226* (0.041)−0.219 (0.271)0.147 (0.187)GRDP per capita0.226* (0.041)0.014 (0.915)0.042 (0.708)−0.035 (0.757)0.044 (0.694)Fiscal independence rate0.265* (0.016)0.022 (0.919)0.182 (0.102)0.123 (0.272)−0.069 (0.539)Fiscal autonomy rate0.169 (0.129)0.078 (0.489)0.039 (0.726)0.339** (0.002)0.139 (0.192)No. of businesses per 1,000 population0.027 (0.810)−0.100 (0.372)−0.029 (0.799)0.258* (0.019)0.136 (0.184)Land price index for residential areas0.189 (0.092)0.092 (0.416)0.079 (0.483)0.239* (0.032)−0.169 (0.328)Land price change rate−0.244* (0.027)−0.099 (0.362)−0.172 (0.122)−0.188 (0.092)−0.325** (0.003)−0.445* (0.013)Housing supply rate0.018 (0.874)0.072 (0.522)0.103 (0.355)0.214 (0.053)Values are presented as β (P).aSeoul metropolitan regions = 1, Non-metropolitan regions = 2.*P < 0.05, **P < 0.01, ***P < 0.001. Examining independent variables, using single and multiple linear regression (rural areas) Values are presented as β (P). aSeoul metropolitan regions = 1, Non-metropolitan regions = 2. *P < 0.05, **P < 0.01, ***P < 0.001. For stress perception rate, the simple linear regression analysis revealed significant positive correlations with the following variables: teacher-to-student ratio (β = 0.375, P = 0.001), high-risk drinking rate (β = 0.261, P = 0.018), obesity rate (β = 0.431, P < 0.001), drinking rate (β = 0.339, P = 0.002), share of social welfare budget (β = 0.259, P = 0.019), PM2.5 (β = 0.473, P < 0.001), PM10 (β = 0.456, P < 0.001), GRDP per capita (β = 0.226, P = 0.041), and fiscal autonomy rate (β = 0.265, P = 0.016). Negative correlations were observed with mean age (β = −0.348, P = 0.001), proportion of single-person households (β = −0.220, P = 0.047), number of older adult leisure facilities per 1,000 older adults (β = −0.221, P = 0.048), and land price change rate (β = −0.244, P = 0.027). In the multiple linear regression analysis, teacher-to-student ratio (β = 0.300, P = 0.024) and obesity rate (β = 0.438, P = 0.005) maintained significant positive correlations. For depressive mood experience rate, the simple linear regression analysis showed significant positive correlations with teacher-to-student ratio (β = 0.251, P = 0.023), obesity rate (β = 0.220, P = 0.047), share of social welfare budget (β = 0.334, P = 0.002), PM2.5 (β = 0.495, P < 0.001), and PM10 (β = 0.431, P < 0.001). The proportion of single-person households (β = −0.247, P = 0.026) showed a significant negative correlation. In the multiple linear regression analysis, only PM2.5 (β = 0.497, P = 0.014) remained statistically significant. For depression symptom prevalence rate, the Simple linear regression analysis showed a negative correlation with healthy lifestyle practice rate (β = −0.351, P = 0.001), and positive correlations with PM2.5 (β = 0.305, P = 0.005) and PM10 (β = 0.226, P = 0.041). In the multiple linear regression analysis, healthy lifestyle practice rate (β = −0.381, P < 0.001) remained negatively correlated, while PM2.5 (β = 0.519, P = 0.011) remained positively correlated. For suicide rate, the simple linear regression analysis revealed positive correlations with high-risk drinking rate (β = 0.227, P = 0.041), obesity rate (β = 0.359, P = 0.001), fiscal autonomy rate (β = 0.339, P = 0.002), number of businesses per 1,000 people (β = 0.258, P = 0.019), and residential land price index (β = 0.239, P = 0.032). Negative correlations were found with healthy lifestyle practice rate (β = −0.249, P = 0.024) and land price change rate (β = −0.325, P = 0.003). In the multiple linear regression analysis, obesity rate (β = 0.325, P = 0.005) remained positively correlated with suicide rate, while land price change rate (β = −0.445, P = 0.0013) showed a significant negative correlation. During each multiple linear regression analysis, multicollinearity among the variables was assessed using VIF value. No variable exhibited VIF value greater than 10; therefore, no significant multicollinearity issues were identified (Supplementary Tables 1 and 2). DISCUSSION This study analyzed the sociodemographic, health and welfare, environmental, and economic factors in urban and rural areas that affect the mental health and suicide rates of residents in each region. The findings revealed significant differences between urban and rural areas in South Korea concerning these factors. When comparing demographic factors between the two groups, previous studies reported that the suicide rate among the elderly is higher than that among other age groups, with more men than women dying from suicide.3233 Additionally, studies in Germany and Belgium found a positive correlation between the proportion of single-person households and suicide rates in these regions.343536 Marriage is generally considered a protective factor against suicide.3738 Consequently, rural areas—with a higher mean age, sex ratio (higher male population), lower proportion of single-person households, and lower marriage rates—may be more vulnerable to suicide in terms of demographic factors. In the health and welfare sector, rural areas have lower rates of healthy lifestyle practices,394041 higher rates of high-risk drinking4243 and obesity,44 lower proportion of social welfare budget,45 fewer medical institutions,46 fewer elderly home-based welfare facilities,47 and lower volunteer activity rate48 than urban areas. In contrast, they also had protective factors, such as lower rates of alcohol consumption and higher numbers of mental health institutions4649 and senior leisure welfare facilities50 than urban areas. Regarding environmental factors, rural areas exhibited a more favorable green space ratio,51 greater availability of cultural facilities,52 and lower air pollution levels (PM2.5, PM10)53 than urban areas. In economic factors, the results showed that urban areas have a higher degree of fiscal independence and are better able to collect taxes, whereas rural areas have a higher degree of fiscal autonomy. According to a previous study in South Korea, high financial independence can increase the cost of social welfare provided to local residents, which can contribute to an increase in the suicide rate.54 Therefore, rural areas demonstrated more strengths than the urban areas in this respect. On the other hand, the residential land price index has been identified as a factor that can increase suicide risk when it is unequal.55 Previous studies have suggested that rising land prices reduce depression56; fluctuations in land prices increase suicide rates5758; and homeownership decreases suicide rates.59 The results of this study showed that in rural areas, an increase in land prices tends to raise suicide rates, while a decrease in land prices also leads to an increase in suicide rates. This can be interpreted as an extension of previous studies. In short, urban and rural areas in South Korea exhibit significant differences in demographics, health, welfare, environmental, and economic factors influencing residents’ mental health and suicide rates. These findings support the justification for building separate models for each area. Positive correlations were found between stress perception, depression, and its prevalence in both urban and rural areas, which is consistent with the results of previous studies reporting that depression and stress are closely related.60 However, stress and suicide rates were negatively correlated in urban areas and suicide rates were positively correlated in rural areas, which is inconsistent with the findings of previous studies.61 These inconsistent results appear to be affected by the mean age of the region. Previous studies have reported that the younger the age, the higher the stress perception,62 and the older the age, the higher the suicide rate.3233 As a result, the difference in mean age among urban municipalities is considered to have distorted the correlation between stress perception and suicide rate, showing a negative correlation between these two factors; whereas mean age has a limited impact on rural areas, showing a positive correlation between stress and suicide rate, as in previous studies. Stress, depression, and depressive symptoms were higher in metropolitan cities, whereas suicide rates were higher in non-metropolitan areas. This is consistent with the results of a study conducted in South Korea,63 which found that metropolitan cities have active interregional exchanges, whereas non-metropolitan regions have an unfavorable impact on suicide rates due to regional deprivation. The number of single-person households and divorce rates were also positively correlated with the suicide rate, which is consistent with Durkheim’s (2001) theory that isolation or a lack of a support system can increase the risk of suicide. The practice of healthy behavioral habits acts as a protective factor in suicide rates, which is consistent with the results of previous studies.6465 The prevalence of obesity was positively correlated with depression and depressive symptoms.66 Drinking rates are positively correlated with stress.67 Volunteer activity rate was negatively correlated with stress perception and suicide rates. The volunteer activity rate is an indicator of the cohesion of a community.68 It may not have a direct impact on mental health or suicide prevention but is thought to show a correlation, as in this study. In rural areas, the number of students per teacher was positively correlated with community stress. The high teacher-to-student ratio can have an indirect impact on the community because it can affect students’ academic achievement69 and teachers’ workload.70 In this study, these results were only found in rural areas. According to a study conducted in South Korea, rural areas have a higher workload for teachers than urban areas; and they also have poor local educational support environments, fewer private education opportunities, and lower household education spending,71 making rural areas more susceptible to high teacher-to-student ratio. This study showed that air pollution in rural areas affects depression. This is consistent with previous research findings that air pollution affects mental health, especially depression.72 However, in this study, PM2.5, showed a correlation with the depression not in urban areas, but only in rural areas. A study conducted in China found that residents in rural and fishing villages are more vulnerable to air pollution due to their low self-protection awareness of air pollution, many outdoor activities, few medical facilities, and low economic status.73 In addition, a study conducted in the United States found that PM2.5 increased the suicide rate in rural and fishing villages, while there was no correlation in urban areas.74 This study analyzed factors affecting the mental health of local residents by classifying all regions of the country into urban and rural areas. We were able to increase the reliability of the data by using official statistics from government agencies and public institutions and minimize temporary outliers by converting five years of data into an average. In addition, among studies conducted overseas, there were few studies that analyzed the factors that affected suicide by collecting data on the ecological factors of various communities. However, this study has several limitations. First, since the regression analysis was conducted using the average value of five years, the effect of time could not be identified, and the causal relationship could not be explained. Second, the study divided the local area into urban and rural areas based on administrative classification, which resulted in a dichotomous division into two categories, urban and rural, which couldn’t reflect the actual environment of local residents. Third, data from 2017 to 2021 were used; therefore, they did not reflect the changes caused by the coronavirus disease 2019 pandemic that occurred at the end of 2019. Additionally, the restricted timeframe of data collection raises the possibility of bias related to temporal variations. Fourth, the variables used in this study might not precisely reflect actual conditions. Thus, discrepancies could exist between the measured values of the variables and reality. Through this study, we found that the factors affecting stress perception, depressive mood experience, depressive symptoms, and suicide rates differ depending on regional characteristics, and extracted the risk and protective factors affecting mental health and suicide in urban and rural areas. It is necessary to promote mental health and suicide prevention strategies tailored to the regional characteristics of cities and rural areas. In the future, we expect to see additional research on how each risk and protective factor affects stress and suicide risk in the long term through a prospective analysis of long-term data.