RESEARCH ARTICLE Built Environment, Natural Environment, and Mental

10.1029/2024GH001047
Health
Key Points:
Yehua Dennis Wei1 , Yu Wang1 , David S. Curtis2 , Sungeun Shin1 , and Ming Wen3,4
• Job sprawl, land mixed use, and green
space associate with lower depression 1
Department of Geography, University of Utah, Salt Lake City, UT, USA, 2Department of Family and Consumer Studies,
risk; the roles of air pollution and
University of Utah, Salt Lake City, UT, USA, 3Faculty of Social Sciences, University of Hong Kong, Hong Kong, China,
precipitation are opposite 4
• Association of population density and Department of Sociology, University of Utah, Salt Lake City, UT, USA
social capital with depression
prevalence exhibit significant spatial
diversity Abstract Mental health disorders have become a global problem, garnering considerable attention.
• Precipitation and socioeconomic status However, the root causes of deteriorating mental health remain poorly understood, with existing literature
mainly affect northwestern counties;
population demographics plays key predominantly concentrating on socioeconomic conditions and psychological factors. This study uses multi‐
role in southeastern counties linear and geographically weighted regressions (GWR) to examine the associations between built and natural
environmental attributes and the prevalence of depression in US counties. The findings reveal that job sprawl
Correspondence to: and land mixed use are highly correlated with a lower risk of depression. Additionally, the presence of green
Y. Wang, spaces, especially in urban area, is associated with improved mental health. Conversely, higher concentrations
[email protected] of air pollutants, such as PM2.5 and CO, along with increased precipitation, are linked to elevated depression
rates. When considering spatial correlation through GWR, the impact of population density and social capital on
Citation: mental health displays substantial spatial heterogeneity. Further analysis, focused on two high depression risk
Wei, Y. D., Wang, Y., Curtis, D. S., Shin, clustering regions (northwestern and southeastern counties), reveals nuanced determinants. In northwestern
S., & Wen, M. (2024). Built environment,
counties, depression rates are more influenced by factors like precipitation and socioeconomic conditions,
natural environment, and mental health.
GeoHealth, 8, e2024GH001047. https:// including unemployment and income segregation. In southeastern counties, population demographic
doi.org/10.1029/2024GH001047 characteristics, particularly racial composition, are associated with high depression prevalence, followed by
built environment factors. Interestingly, job growth and crime rates only emerge as significant factors in the
Received 15 MAR 2024
context of high depression risks in southeastern counties. This study underscores the robust linkages and spatial
Accepted 13 JUN 2024
variations between built and natural environments and mental health, emphasizing the need for effective
Author Contributions: depression treatment to incorporate these multifaceted factors.
Conceptualization: Yehua Dennis Wei,
Yu Wang Plain Language Summary This study examines the associations between built and natural
Formal analysis: Yehua Dennis Wei,
environmental attributes and the prevalence of depression in US counties. The findings reveal that greater job
Yu Wang, Sungeun Shin
Funding acquisition: Yehua Dennis Wei, sprawl and land mixed use are associated with lower depression prevalence. Additionally, the presence of green
David S. Curtis, Ming Wen spaces in urban area is associated with reduced depression prevalence. Conversely, higher concentrations of air
Methodology: Yehua Dennis Wei,
pollutants, such as PM2.5 and CO, along with increased precipitation, are linked to elevated depression rates.
Yu Wang, Sungeun Shin
Project administration: Yehua Associations between population density and social capital with depression prevalence display substantial
Dennis Wei spatial heterogeneity. Further analysis, focused on two high depression risk clustering regions (northwestern
Resources: Yehua Dennis Wei, David
and southeastern counties), reveals region‐specific determinants.
S. Curtis, Ming Wen
Supervision: Yehua Dennis Wei
Visualization: Yu Wang, Sungeun Shin
Writing – original draft: Yehua
Dennis Wei, Yu Wang 1. Introduction
Writing – review & editing: Yehua
Dennis Wei, Yu Wang, David S. Curtis, A significant number of US residents experience mental health problems, a burden that has intensified amidst the
Sungeun Shin, Ming Wen
socioeconomic strains and public health crisis brought about by COVID‐19. The impact is substantial, with nearly
10 percent of US adults experiencing major depression each year and a recent global rise in depression rates
(Moreno‐Agostino et al., 2021). Depression carries a range of socioeconomic and health repercussions, including
© 2024 The Author(s). GeoHealth the reduction of social capacity, a decline in labor productivity, and an increase in disability rates (Garrido‐
published by Wiley Periodicals LLC on Cumbrera et al., 2018). Notably, it stands as the most widespread psychiatric disorder associated with suicide
behalf of American Geophysical Union.
This is an open access article under the mortality (Hawton et al., 2013). Given this high prevalence and the substantial costs, identifying place‐based risk
terms of the Creative Commons factors of depression is imperative.
Attribution‐NonCommercial‐NoDerivs
License, which permits use and At individual‐level, risk of mental health problems is intricately linked to demographic characteristics, encom-
distribution in any medium, provided the passing factors like genes, gender, and age, etc. (Silva et al., 2016). Due to biological, familial and sociocultural
original work is properly cited, the use is
non‐commercial and no modifications or factors, women experience higher depression rates than men (Parker & Brotchie, 2010). Depressive symptoms in
adaptations are made. middle‐aged adults are significantly higher than in children and older adults (Cuijpers et al., 2020), because they

WEI ET AL. 1 of 20
 GeoHealth 10.1029/2024GH001047

usually have more anxiety and stress from financial status, work environment, and household relationship (Guan
et al., 2022). A higher level of education may expose individuals to greater academic, social, and financial stress,
which is linked to poor mental health (McCloud et al., 2023). Cvetkovski et al. (2019) reached the opposite
conclusion. Single parents usually have lower income levels and are more likely to experience prejudice in society
and mental health problems (Kim & Kim, 2020). Individual socio‐economic status can also significantly affect
mental health. Individuals with low income disproportionately encounter economic and social stressors and have
lower access to health and social resources (Marmot & Wilkinson, 2001). Unemployed people are linked to poor
mental health because of financial and mental stress (McKee‐Ryan et al., 2005). Black/African Americans report
greater psychological distress and more depressive symptoms (Barnes & Bates, 2017). Non‐Hispanic white
Americans generally experience higher rates of major depressive disorder than African and Mexican Americans
(Pamplin & Bates, 2021).

At area‐level, extensive research has shown that socioeconomic conditions impact mental well‐being (Lorant
et al., 2003; Richardson et al., 2015). Places with higher social capital generally have lower rates of mental health
disorders, due to emotional support and resources sharing (Thoits, 2011). However, close social ties can limit
individual freedom and welfare and lead individuals to engage in health‐compromising behaviors such as alcohol
consumption (Almedom, 2005). Limited health insurance coverage decreases access to mental health services and
elevates mental health problems (Finkelstein et al., 2012). Residing in areas with more crimes reduces physical
activity, decreases feelings of safety and increases risk of depression (Ross, 2017; Wei et al., 2021).
In recent years, the exploration of area‐level determinants has expanded to encompass physical and natural
features within the built environment or urban form due to the acceleration and expansion of urbanization. Ev-
idence indicates that urbanization contributes to psychological distress, and urban areas and larger cities tend to
have higher rates of mental disorders (J. Chen et al., 2015). On the other hand, better infrastructure and public
services in urban areas may increase residents' life satisfaction, thus improving their mental health (Stier
et al., 2021). With the rise of urban sprawl as a global phenomenon, the process and effects of urban sprawl have
drawn broad attention. Urban sprawl has been widely criticized for its association with various urban challenges,
including imbalanced jobs‐housing distribution, traffic congestion, air pollution, diminished social cohesion,
sedentary lifestyles, and adverse health outcomes (Wei & Ewing, 2018). Urban sprawl also contributes to resi-
dential segregation between different racial, ethnic, or socioeconomic groups, a spatial mismatch between low‐
income households and suitable employment opportunities, and higher poverty rates among Black/African
Americans. Consequently, urban sprawl is often considered a fundamental driver of urban problems in the US and
is believed to have negative implications for mental health. The diversity of land use and job opportunities can
enhance residents' access to resources, services, and employment opportunities, thereby meeting their daily needs
and reducing psychological stress (Van Hooff & Van Hooft, 2014; Wu et al., 2016).

However, a study by Garrido‐Cumbrera et al. (2018) found that sprawling areas usually have fewer mental health
problems, since households with more socioeconomic resources and lower risk of depression prefer less dense
suburban areas with more natural resources (Brueckner et al., 1999; Curtis et al., 2024). Similarly, more dense
neighborhoods tend to generate greater mental distress (Wang et al., 2021). Despite the importance of the built
environment in shaping mental health outcomes, there is still a limited body of empirical research examining its
impact.
Natural environmental conditions also influence mental health and are linked to the built environment. Air
pollution, a byproduct of urban sprawl and increased vehicle miles traveled, exacerbates psychological distress
(F. Chen et al., 2023; Piracha & Chaudhary, 2022), while green spaces and urban parks are stress‐relieving
remedies (Richardson et al., 2013; Sturm & Cohen, 2014). Interestingly, the quality of green spaces appears to
be more influential than the quantity in terms of their impact on mental health (Astell‐Burt et al., 2014). Moreover,
the impact of green spaces on mental health has been found to vary between urban and rural areas (Ryan
et al., 2024). Exposure to environmental toxins is another area‐level factor that heightens the risk of mental health
disorders (McLeod, 2017).
Previous studies have investigated the mental health implications of various factors, including demographic
characteristics, socioeconomic conditions, the built environment, and the natural environment (Dang et al., 2019;
Li & Wei, 2023; Wang et al., 2021). However, a gap persists in terms of an inclusive research framework that
integrates these influencing factors. Some argue that the contextual influence on mental health is primarily rooted
in area socioeconomic conditions, overshadowing the significance of the built environment (Garrido‐Cumbrera

WEI ET AL. 2 of 20
 GeoHealth 10.1029/2024GH001047

et al., 2018). The role of the urban context in mental health remains inadequately elucidated, and scholars disagree
about the definition of urban form or the built environment and their specific impacts on mental health (Clifton
et al., 2008). Furthermore, the impacts of the natural environment, including air pollution and green spaces, have
not been sufficiently explored (Mantler & Logan, 2015). Thus, we aim to advance the research on mental health
by conducting a comprehensive study of the role of the built and natural environments in shaping mental health
outcomes. Furthermore, our study contributes to existing knowledge by incorporating spatial perspectives to
better grasp the spatial heterogeneity of mental health problems.
This paper provides a comprehensive understanding of the diverse area‐level risk factors for depression preva-
lence rates across US counties by analyzing the effects of built and natural environmental attributes. We ask the
following research questions: (a) Where in the US is depression prevalence elevated? (b) How do built and natural
environments contribute to variation in depression prevalence rates? (c) Do the predictors of depression preva-
lence vary across regions? To examine these questions, we compile a large data set that incorporates county‐level
depression prevalence estimates for 2019 from the CDC Places, along with data from various publicly available
sources capturing an extensive array of built and natural environmental attributes. Our analysis aims to shed light
on the multifaceted causes of depression, providing valuable insights into the interplay between environmental
factors and mental health outcomes.

2. Analytical Framework and Methodology

2.1. Analytical Framework
Building upon the literature review, this study integrates built and natural environments with demographic factors
and socioeconomic conditions. The aim is to provide a more comprehensive understanding of the underlying
causes of depression across US counties. The analytical framework is presented in Figure 1.
Studies tend to use one or two simple indicators such as population density or urban sprawl index to study the
effects of built environment on health. Built environment covers multiple dimensions, and its comprehensive
effects on mental health have rarely been thoroughly examined. Research on the effects of cities on human
mobility and physical activities has predominantly focused on the built environment, mainly the three Ds (density,
distance, and diversity). More recently, two more Ds (destination and design) have been incorporated into
measures of urban form. These five Ds provide a comprehensive framework for understanding the influence of the
built environment on human mobility, physical activity, and potentially mental health. We thus use a series of
indices to represent built environment, including job sprawl index, population sprawl index, land mixed use
index, job mixed layout index, job‐housing balance index, population density and urban area. Green space,
pollutant PM2.5 and CO concentrations and precipitation are used to reflect the natural environment. Most studies
find exposure to air pollution and increased precipitation are linked to more mental health problems, while green
spaces can alleviate negative emotions (F. Chen et al., 2023; Kent et al., 2009; Leslie & Cerin, 2008; Sturm &
Cohen, 2014). Our analytical framework also includes those typical natural environmental attributes.

Our research framework also incorporates control variables to reduce inaccurate model estimation due to omitted
variables. Individual characteristics such as gender, aging, marital status, race and family structure affect mental
health (Cuijpers et al., 2020; Pamplin & Bates, 2021; Parker & Brotchie, 2010). Socioeconomic conditions also
play important roles in mental health. Usually, places with larger income segregation, lower job growth, higher
unemployment rates, limited access to stores, less social capital, low health insurance coverage, and more crimes
tend to have more mental health problems (McKee‐Ryan et al., 2005).

2.2. Data Sources

This study uses county level data for the continental US (i.e., excluding Alaska, Hawaii, and the US Territories).
The US is among the countries with the most serious mental health related problems, and its cities are charac-
terized by significant sprawling (Behnisch et al., 2022; Tikkanen et al., 2020). County diagnosed depression
prevalence among adults came from the Centers for Disease Control and Prevention (CDC) Places, 2021 data
release. CDC Places estimates of depression prevalence were derived from the Behavioral Risk Factor Sur-
veillance System 2019 data using small area estimation techniques. County population estimates for subgroups
were used to generate model‐based estimates of overall county depression prevalence among adults aged at least

WEI ET AL. 3 of 20
 GeoHealth 10.1029/2024GH001047

Figure 1. The influencing factors of depression.

18‐years‐old. The data and modeling approach are described in greater detail in Greenlund et al. (2022). In
sensitivity tests, we used an age‐adjusted estimate for depression prevalence among adults to test the robustness of
our results. Data representing built and natural environments, demographic factors, and socioeconomic conditions
come from multiple publicly available sources. Table 1 presents detailed information on variable definitions and
data sources. Additionally, we propose hypothesized associations for variables exhibiting relatively consistent
findings based on the literature review.

2.3. Hotspot Analysis

Hotspot analysis is a geospatial technique designed to identify clusters of spatial phenomena and their spatial
correlations. Hotspots refer to areas with a higher concentration of events compared to the expected events
occurring at random. There are many methods to identify the spatial pattern, such as the spatial autocorrelation

WEI ET AL. 4 of 20
 Table 1
Variable Selection and Data Sources
Type Abbreviation Indicator meaning Data sources

WEI ET AL.
Depression Dep Depression rate. Model‐based estimate for crude prevalence of Disease Control and
depression among adults aged 18–64 years (2019) Prevention (CDC)
Dep2 Age‐adjusted depression rate. Model‐based estimate for age‐
adjusted crude prevalence of depression among adults aged
18–64 years (2019)
Built Environment Sprawl_J Job sprawl. Centering_coefficient of variation in census block Smart Location Database (SLD)
group employment densities (2010). The more variation in Version 2.0
densities around the mean, the more centering and/or
subcentering exists within the county
Sprawl_P Population sprawl. Centering_coefficient of variation in census
block group population densities (2010). The more variation
in densities around the mean, the more centering and/or
subcentering exists within the county
Mixed_Land Land mixed use. 5‐tier employment and household entropy.
Values were weighted by the sum of block group's
population and employment as a percentage of the county
total (2010)
Mixed_Job Job mixed layout. 5‐tier employment entropy, weighted by the
sum of CBG population and employment as a percentage of
the county total (2010)
GeoHealth

Balance_JH Jobs‐housing balance. Working population and actual jobs

equilibrium Index, weighted by the sum of block group's
population and employment as a percentage of the county
total (2010). The closer to one the more balanced the resident
workers and jobs in the Census Block Group
Density_P Population density. Gross population density (people/acre) on
unprotected land (2010)
Urban Urban area. 1 is urban area, and 0 is non‐urban area
Natural Environment Green_S Green space accessibility. Percentage of vegetated land within a Calculated using data from National
census tract, including developed open space, grass, shrub, Land Cover Database (NLCD)
and forest areas, and excluding those areas for intensive
agricultural uses (2001)
PM2.5 PM2.5 concentration. Average annual outdoor concentration of Center for Air, Climate, and Energy
PM2.5 (μg/m3, 2010) Solutions
CO CO concentration. Average annual outdoor concentration of CO
(ppb, 2010)
Precipitation Precipitation level. Average January precipitation at county National Centers for Environmental
level (inches, 2010) Information
Individual (or demographic) Gender Female rate. Percent population who are female (%, 2010) 2011 American Community
Factors Aging Aging population ratio. Percent population who are 67 years old Survey (ACS)
or older (%, 2010)
Divorce_R Divorce rate. Average divorce rate between 2010 and 2014 (%) 2014 American Community
Survey (ACS)
10.1029/2024GH001047

5 of 20
WEI ET AL.
Table 1
Continued
Type Abbreviation Indicator meaning Data sources
Racial_C Racial composition. Percent population who are non‐Hispanic 2011 American Community
whites along (%, 2010) Survey (ACS)
Family_S Family structure. The number of households with female heads Opportunity Atlas Database
(and no husband present) or male heads (and no wife
present) with own children under 18 years old present
divided by the total number of households with own children
present (%, 2010)
Socioeconomic Conditions Social_C Social capital index, created using principal component analysis Northeast Regional Center for Rural
using four factors (voter turnout, census response rate, Development
number of non‐profit organizations, and the aggregate of
GeoHealth

organization such as religious organizations, civic and social

associations) (2014)
Growth_J Job growth rate. Average annualized job growth rate (%) over Census Bureau
the time period 2004 to 2013
Rate_Unemp Unemployment rate. Average unemployment rate (%) between 2014 American Community
2010 and 2014 Survey (ACS).
Store_A Access to stores. Percentage of households with low income & Food Environment Atlas of USDA
more than 10 miles to store (2015). Low value means high Economic Research Service.
access
Uninsured_P Rate of uninsured people. Model‐based estimate for crude Disease Control and
prevalence of current lack of health insurance among adults Prevention (CDC)
aged 18–64 years (2019)
Income_S Income segregation. Average Gini coefficient of income from 2014 American Community
2010 to 2014 Survey (ACS)
Crime_R Crime rate. Violent crime rate 2010 (per 100,000) Uniform Crime Reports
10.1029/2024GH001047

6 of 20
 GeoHealth 10.1029/2024GH001047

analysis and the cluster analysis. For a better understanding of local conditions and spatial autocorrelation, local
indicators of spatial association (LISA), such as Gettis‐Ord Gi* statistic, is used to measure to what extent points
near a given point have similar values (Anselin, 1995). We also use the Gettis‐Ord Gi* statistic to figure out
whether urban sprawl and SWB are spatially autocorrelated.

2.4. Multiple Linear Regression

We use multiple linear regression to identify the association between built and natural environment attributes and
depression prevalence, with the following formula.

Dep = θ0 + θi BE + θj NE + θk ISECk + ε

where, Dep is our dependent variable, that is, county‐level crude prevalence of depression. BE is a vector of built
environment variables, and θi refers to the coefficients for each variable. NE represents the vector of natural
environment variables, and θj refers to the corresponding coefficients. The vector ISEC includes the demographic
and socioeconomic control variables. θ0 and ε are the intercept and residual term, respectively.

Given our primary interest in the effect of built and natural environments, built environment variables were
included in the initial model followed by adding natural environment, individual, and socioeconomic condition
variables (Model 1 to Model 4 in Table 2). Considering that green space is more suitable for urban areas
rather than rural areas, we also added the Urban*Greenspace interaction term in Model 5 to explore whether
the relationship between green space and depression is more significant in urban areas. Finally, we used the
age‐adjusted crude prevalence of depression (Dep2) to replace Dep in regressions to prove the robustness of
results.
The small area estimation techniques used to derive the CDC Places estimates of county‐level crude prevalence of
depression incorporate county‐level poverty data. We therefore do not include the poverty variable in the
regression. Other demographic and socio‐economic related data used by the CDC are based on surveys, which is
at the individual level rather than the county level. We use county‐level data, so the correlation should not be as
strong. Furthermore, we are careful with interpreting the effects of population characteristics. Given that the
determinants of depression may be highly correlated with each other and cause the results bias due to the mul-
ticollinearity in regression, we calculate the correlation coefficients of all variables. All correlation coefficients
are significantly less than 0.7, the minimum threshold for high correlation coefficients (Akoglu, 2018). Also, we
performed a VIF (variance inflation factor) test to detect potential problem of multicollinearity among inde-
pendent variables. VIFs in all regression results are less than 2.5, thus no obvious multicollinearity is presented in
our regression models.

2.5. Geographically Weighted Regression

Considering that there may be different relationships between variables across regions, a frequently occurring
case in geospatial data, we use the geographically weighted regression (GWR) to identify the spatial heteroge-
neity of determinants in depression. GWR processes operate within a small spatial scale rather than the global
scale. The formula for GWR is as follow.

Dep∼i = θ0∼ + θi∼ BE ∼ + θj∼ NE ∼ + θk∼ ISECk∼ + ε ∼

where ∼ represents counties in the small spatial scale. Depi∼ represents the crude prevalence of depression of
these counties. BE ∼ and NE ∼ are vectors of built and natural environment variables respectively. θi∼ and θj∼ are
their corresponding coefficients. The vector ISEC∼k indicates demographic and socioeconomic control variables.
θ0∼ and ε ∼ are the intercept and residual term, respectively. In building GWR model, we used adaptive bandwidths
so that roughly the same number of observations is used. Gaussian weights are chosen such that closer points have
more weights. We employ cross‐validation to test different bandwidth sizes or points.

WEI ET AL. 7 of 20
 GeoHealth 10.1029/2024GH001047

Table 2
The Effects of Built and Natural Environments on Depression
Model 1 Model 2 Model 3 Model 4 Model 5
Variables Dep Dep Dep Dep Dep
Built Environment Sprawl_J 0.633*** (0.119) 0.824*** (0.114) 0.584*** (0.0954) 0.447*** (0.107) 0.444*** (0.108)
Sprawl_P 0.0590 (0.0913) 0.0434 (0.0837) 0.0913 (0.0680) 0.0310 (0.0862) 0.0324 (0.0863)
Mixed_Land 15.44*** (1.181) 12.08*** (1.095) 8.123*** (1.032) 3.632*** (1.113) 3.698*** (1.113)
Mixed_Job 2.625*** (0.790) 0.239 (0.771) 4.415*** (0.777) 2.721*** (0.818) 2.769*** (0.815)
Balance_JH 1.413* (0.725) 0.239 (0.685) 0.251 (0.646) 0.550 (0.631) 0.588 (0.630)
Density_P 0.0971** (0.0484) 0.138** (0.0598) 0.0643* (0.0346) 0.291 (0.183) 0.287 (0.182)
Urban 0.146 (0.124) 0.557*** (0.116) 0.662*** (0.104) 0.398*** (0.108) 0.0378 (0.200)
Natural Environment Green_s 1.849*** (0.302) 0.336 (0.275) 0.644** (0.298) 0.259 (0.350)
PM2.5 0.254*** (0.0332) 0.167*** (0.0311) 0.150*** (0.0336) 0.145*** (0.0338)
CO 11.99*** (1.390) 6.679*** (1.167) 5.716*** (1.245) 5.662*** (1.247)
Precipitation 0.346*** (0.0261) 0.332*** (0.0267) 0.225*** (0.0283) 0.230*** (0.0285)
Individual (or demographic) Factors Gender 16.04*** (2.075) 21.55*** (2.209) 21.66*** (2.218)
Aging 17.20*** (1.480) 17.90*** (1.951) 17.78*** (1.955)
Divorce_R 0.320*** (0.0242) 0.250*** (0.0251) 0.248*** (0.0252)
Race_C 7.700*** (0.388) 10.47*** (0.528) 10.47*** (0.527)
Family_S 4.631*** (0.729) 1.905** (0.838) 1.831** (0.837)
Socioeconomic Conditions Social_C 0.433*** (0.105) 0.430*** (0.105)
Growth_J 5.753 (3.546) 5.681 (3.542)
Rate_Unemp 0.160*** (0.0197) 0.161*** (0.0197)
Store_A 3.898*** (0.775) 3.830*** (0.772)
Uninsured_P 0.0510*** (0.0119) 0.0512*** (0.0119)
Income_S 10.74*** (1.634) 10.91*** (1.640)
Crime_R 0.000150 (0.000282) 0.000139 (0.000284)
Urban*Green_S 0.772** (0.358)
Constant 31.14*** (0.640) 21.93*** (0.803) 7.244*** (1.199) 5.758*** (1.529) 5.925*** (1.538)
Observations 3,043 3,041 3,040 2,559 2,559
R‐squared 0.177 0.287 0.470 0.527 0.528
VIF 1.49 1.65 1.75 1.99 ‐
Note. Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.

3. Results
3.1. Spatial Distribution and Clustering of Depression Prevalence and Predictors
We first conducted the hot spot analysis, which shows that depression prevalence forms geographic clustering
across US counties (Figure 2). High depression rates are mainly found in two regions. One is the northwestern
region, mainly counties in Washington, Oregon, northern Idaho, and Montana (Figure 3). The other covers parts
of the southeastern and midwestern regions, including Alabama (AL), Arkansas (AR), Kentucky (KY), Louisiana
(LA), Michigan (MI), Missouri (MO), Oklahoma (OK), Tennessee (TN), and West Virginia. However, low
depression rates are found in central United States.
Population density varies widely among US counties (Figure 4). Except for some counties in New York, Chicago,
and Los Angeles, etc., the vast majority of counties have relatively low population density. Population clustering
and job clustering occur more frequently in the eastern US with limited land rather than the western portion.
Levels of land mixed use and job mixed layout are high in north‐central US, which tends to have lower depression
rates. The distribution of jobs‐housing balance is very similar to that of land mixed use and job mixed layout.

WEI ET AL. 8 of 20
 GeoHealth 10.1029/2024GH001047

Figure 2. Spatial clustering of depression prevalence.

More urbanization, dense population and developed industries fill much of the eastern US, which has very limited
green space and higher PM2.5 concentrations (Figure 5). It should be noted that increasing awareness of the effects
of green spaces on mental health stems from concerns about declining green space due to urbanization (Byomkesh
et al., 2012). Thus, green space is usually defined as urban vegetated space such as parks, grasslands, sports and
playing fields and near‐road trees, rather than agricultural land (Vanaken & Danckaerts, 2018). Thus, the green
space index in our study refers to the percentage of vegetated land within a census tract, including developed open
space, grass, shrub, and forest areas, and excluding those areas for intensive agricultural uses. Thus, it more
appropriately represents green space in urban areas rather than rural areas. Counties along the Appalachian
Mountains, the west bank of the Mississippi River, and the western region of the US have higher CO concen-
trations. Precipitation is significantly higher in counties along the Pacific coast and in some southeastern states
such as Florida and Alabama.

Figure 3. Spatial distribution of the prevalence of depression by county.

WEI ET AL. 9 of 20
 GeoHealth 10.1029/2024GH001047

Figure 4. Spatial distribution of built environment factors.

3.2. Built and Natural Environmental Attributes as Predictors of Depression Prevalence

The above mapping analysis provides basis for conducting more rigorous modeling, and we ran a series of
regression models to identify place‐based predictors of county depression prevalence rates. Model 1 to Model 4 in
Table 2 shows that built environment accounts for 17.7% of the variance in depression rates, and the addition of
natural environment increases the R2 to 28.7%. The built and natural environments explain between one‐fourth
and one‐third of depression.
When all control variables are added, job sprawl, land mixed use, job mixed layout and urban areas are signif-
icantly associated with depression prevalence (Model 4). Lower values for job sprawl indices indicate greater
sprawl or less compact development, thus job sprawl is linked to decreased depression prevalence. Counties with
high level of land mixed use and job mixed layout, as well as urban areas are associated to lower depression
prevalence. For natural environment, all variables show the same association as our hypotheses. More green
spaces, lower level of air pollution, and less precipitation are linked to lower depression prevalence.
For control variables, high proportion of female, divorced people, and non‐Hispanic whites, as well as more
single‐parents families share, are associated with high depression rates, while large share of aging people is
usually tied to lower risk of being depression. Counties with high unemployment rate, poor access to store, more
uninsured population share, and severe income segmentation are connected to more depression prevalence, while
increased social capital is linked to better mental health.

WEI ET AL. 10 of 20
 GeoHealth 10.1029/2024GH001047

Figure 5. Spatial distribution of natural environmental factors.

We maintain that the relationship between natural environment and mental health also depends on the charac-
teristics of built environment. Therefore, we also examine its interaction with the urban context. The coefficient of
Urban*Greenspace in Model 5 is negatively significant, implying that the negative correlation between green
space and depression prevalence mainly concentrates in urban areas.

3.3. Spatial Heterogeneity of Determinants of Depression

We explore the spatial heterogeneity of the relationship between depression and independent variables by using
GWR. The positive and negative ratios are used to measure how these factors have different effects on depression
across counties (Table 3). The positive ratio is the proportion of counties with a positive impact on depression
(i.e., increasing the depression rate), while the negative ratio is the share of the counties with a negative effect on
depression (i.e., decreasing the depression rate). The largest positive ratio for built and natural environment
variables is 63.76% (Sprawl_P), followed by 62.66% (Sprawl_J), meaning that urban sprawl increases depression
rates in more than 60% of counties. The largest negative ratio is Mixed_Land (62.89%), followed by Green_S
(61.54%), indicating that mixed land use and green space decrease depression risks in more than 60% of counties.
Thus, the associations between built and natural environments and depression are unevenly distributed across
counties. Conversely, there are some demographic and SES factors are spatially stationary. Specifically, for more
than 80% counties, high female share, more non‐Hispanic white population, less aging people, single‐parent
family structure and income segregation are associated with higher depression prevalence.
We further the above analysis by mapping the correlations of population density and social capital and depression
rates, which shows more significant spatial patterns than other variables. Although high population density is
linked to less depression prevalence in the vast majority of counties, it shows positive ties with depression risk in a
few northeastern metropolitan areas such as the New York and Philadelphia‐Camden‐Wilmington Metropolitan
Areas (Figure 6). Although the positive coefficients are also concentrated in Minneapolis–Saint Paul and Salt
Lake City metropolitan areas, the P‐value in these regions are larger than 0.1, that is, the coefficients are
insignificant in these regions. Thus, we cannot conclude the relationship between population density and
depression rates is positive in these regions. Different from the majority of counties, high level of social capital is
associated with high depression prevalence in northwestern counties and some counties in the northeastern region
(Figure 7). Although some north‐central counties show the positive relationship between social capital and
depression rates, only coefficients for counties in south Minnesota are significant.

WEI ET AL. 11 of 20
 GeoHealth 10.1029/2024GH001047

Table 3
Summary of GWR Coefficient Estimates at County Level
Min. Median Max. Global Positive ratio (%) Negative ratio (%)
Sprawl_J 1.983 0.161 1.565 0.316 62.66 37.34
Sprawl_P 1.321 0.075 2.023 0.074 63.76 36.24
Mixed_Land 55.202 1.521 8.444 4.707 37.11 62.89
Mixed_Job 14.087 0.610 32.266 2.587 43.64 56.36
Balance_JH 9.318 0.581 42.082 0.260 59.64 40.36
Density_P 2.472 0.030 3.737 0.055 47.83 52.17
Urban 2.882 0.056 3.552 0.529 52.40 47.60
Green_S 14.633 0.638 10.298 0.022 38.46 61.54
PM2.5 1.997 0.027 1.060 0.196 43.42 56.58
CO 42.116 0.829 32.446 2.462 57.52 42.48
Precipitation 2.600 0.139 3.870 0.251 63.69 36.31
Gender 194.910 10.621 75.944 12.971 89.25 10.75
Aging 52.404 9.846 22.827 18.382 15.96 84.04
Divorce_R 148.120 7.199 83.707 23.034 65.59 34.41
Racial_C 1.955 0.141 1.819 0.273 39.81 60.19
Family_S 8.951 9.246 45.233 15.178 94.53 5.47
Social_C 3.940 2.858 21.289 6.329 82.75 17.25
Growth_J 84.316 5.869 197.080 3.052 33.22 66.78
Rate_Unemp 1.064 0.062 0.538 0.157 73.87 26.13
Store_A 16.051 1.212 114.100 4.292 63.82 36.18
Uninsured_P 0.329 0.080 1.263 0.173 68.97 31.03
Income_S 21.860 7.069 34.576 8.387 84.62 15.38

3.4. Determinants of Depression in Northwestern and Southeastern Counties

There are two main high depression prevalence regions in US, that is, northwestern and southeastern counties
(Figure 2), which need further examination. Model 1 to Model4 in Table 4 show the multiple linear regression
results for northwestern counties. Correlation analysis for northwestern counties shows that there is a high
correlation between land mixed use, job mixed layout and jobs‐housing balance, which differs from findings
based on counties nationwide. Considering high land mixed use tends to have more job mixed layout and high
level of jobs‐housing balance, we only include land mixed use in our regression. To avoid inaccurate estimation
caused by omitted variables, we also included the mean value of three variables (Mixed index) in the regression in
Model 5. Table 5 shows the regression results for southeastern counties.
After adding all variables, we find that there are fewer variables associated with depression prevalence in
northwestern counties. No built environment variable shows significant correlation with depression prevalence
(Model 4). Among other variables, socioeconomic variables have made the greatest contribution to R2, followed
by natural environment. This means socioeconomic conditions and natural environment are the main de-
terminants of depression risks in this region. For southeastern counties, demographic variables are most relevant
to the high depression rates, with the highest increased R2. The built environment variables also make a significant
contribution to R2, thus ranking as the second most important predictors.

4. Discussion
There is the high depression prevalence clustering in US northwestern region and US southeastern and midwestern
regions, while low depression prevalence mainly clustering in central United States. Compared with other regions,
the northwestern regions usually have longer‐term gray and rainy weather. Less exposure to sunlight can lead to
seasonal affective disorder, a subtype of depression. Southeastern regions like in the Appalachian and Southern

WEI ET AL. 12 of 20
 GeoHealth 10.1029/2024GH001047

Figure 6. Effects of population density on depression.

Mississippi Valley regions, face many problems such as high poverty rates, limited employment opportunities, and
generally more challenging socioeconomic conditions. People in the Great Lakes region and Maine, with cold
winter temperature, also suffer more severe depression. Counties in central United States have dry and mild
climate, and the adequate sunlight exposure may reduce the risk of depression. There are more rural counties in the
central region. Compared to urban counties, rural counties experience fewer social disparities, less pollution, and
more connection with the nature, which contribute to higher levels of mental health (Ventriglio et al., 2021).

Job sprawl may result in more open space in urban centers and more job opportunities in suburbs, which can relax
people's commute and improve mental health (Garrido‐Cumbrera et al., 2018). Land mixed use areas have more
access to resources, services and entertainments, which can meet residents' needs and promote more outdoor
activities and social interactions (Wu et al., 2016), thereby improving mental health. Job mixed layout enables
people to find jobs that interest them, preventing the sense of monotony and boredom and thus reducing risks of
being depressed. The more advanced infrastructure and comprehensive healthcare in urban areas enhance people's
life satisfaction, therefore urban areas are significantly correlated with good mental health. For natural envi-
ronment factors, green space is linked to lower depression prevalence by absorbing airborne pollutants and
making people feel relaxed (Leslie & Cerin, 2008). Due to the higher population density, more crowded public
space, and more serious air pollution caused by transportation and/or industrial production in urban areas, the role
of green space is more evident in urban areas. Exposure to PM2.5 and CO pollution increases the risk of depression
(F. Chen et al., 2023; Liu et al., 2018; Von Lindern et al., 2016). Also, precipitation increases the potential for
depression, since it reduces the time people can expose to sunlight and outdoor activities, thereby inhibiting
serotonin production (Kent et al., 2009). All significant control variables support our hypotheses in Table 1,
except for racial composition. Although some argue that Black/African Americans suffer more mental problems

WEI ET AL. 13 of 20
 GeoHealth 10.1029/2024GH001047

Figure 7. Effects of social capital on depression.

(Barnes & Bates, 2017), LaVeist et al. (2014) find that non‐Hispanic whites experience depression more than
other racial/ethnic groups, which is supported by our result.

Our study also reveals significant spatial heterogeneity in the association between population density and
depression, as well as social capital and depression. Different from the majority of counties, high population
density is associated to higher depression prevalence in some northeastern metropolitan areas such as the New
York and Philadelphia‐Camden‐Wilmington Metropolitan Areas. These metropolitan areas are characterized by
high population density and limited urban space, which leads to the diseconomies of agglomeration, such as
traffic jam, high housing cost, and high crime rates. Northeastern counties and some counties in the northwestern
region, as well as counties in south Minnesota, show a different correlation between social capital levels and
depression prevalence compared to the majority of counties. Residents in northeastern counties witness large gaps
in income levels and employment opportunities, with high concentration of highly paid financial and high‐tech
jobs. Dense population and intense competition may cause social stress, resulting negative effect of social capital
on mental health. For example, the busy lifestyle and high competition in the New York metropolitan area are
very likely to make socializing a form of pressure for some people, thus higher levels of social capital are
connected with more depression risks. Counties in the northwestern region may have relatively less dense
population distribution and have more significant agriculture and manufacturing industries. Increasing social
capital may threaten their rural life style and likely causes more depression. Another possible reason is that since
the northwest regions have higher depression rates, the depressed people may increase their social interactions to
cope with their problem (De Silva et al., 2007). Counties in south Minnesota show the positive relationship

WEI ET AL. 14 of 20
 GeoHealth 10.1029/2024GH001047

Table 4
Determinants of Depression in Northwestern Counties
Model 1 Model 2 Model 3 Model 4 Model 5
Variables Dep Dep Dep Dep Dep
Built Environment Sprawl_J 0.411* (0.212) 0.279 (0.200) 0.216 (0.189) 0.168 (0.176) 0.166 (0.178)
Sprawl_P 0.120 (0.199) 0.0439 (0.184) 0.116 (0.173) 0.0506 (0.165) 0.0408 (0.170)
Mixed_Land 8.496*** (2.264) 6.755*** (2.214) 5.602** (2.192) 0.0512 (1.875)
Mixed Index 0.380 (2.230)
Density_P 0.171 (0.217) 0.0522 (0.236) 0.134 (0.228) 0.343 (0.274) 0.354 (0.277)
Urban 0.136 (0.258) 0.0366 (0.256) 0.0384 (0.258) 0.182 (0.227) 0.193 (0.217)
Natural Environment Green_S 1.270 (1.037) 1.223 (0.992) 1.579* (0.805) 1.595** (0.805)
PM2.5 0.270* (0.139) 0.171 (0.143) 0.117 (0.145) 0.115 (0.145)
CO 7.333* (4.015) 5.789 (3.930) 3.171 (3.378) 3.172 (3.376)
Precipitation 0.121*** (0.0258) 0.0923*** (0.0272) 0.0567* (0.0300) 0.0569* (0.0295)
Individual (or demographic) Factors Gender 28.32*** (7.326) 14.78** (6.404) 14.76** (6.330)
Aging 5.479 (3.756) 5.558 (4.831) 5.586 (4.780)
Divorce_R 0.0307*** (0.0111) 0.0328*** (0.0107) 0.0332*** (0.0106)
Race_C 0.856 (1.286) 0.709 (1.264) 0.737 (1.313)
Family_S 4.907*** (1.859) 3.228* (1.875) 3.279* (1.854)
Socioeconomic Conditions Social_C 0.312** (0.148) 0.324** (0.148)
Growth_J 0.986 (9.705) 0.820 (9.676)
Rate_Unemp 0.118** (0.0453) 0.118** (0.0453)
Store_A 1.796 (1.306) 1.798 (1.305)
Uninsured_P 0.186*** (0.0317) 0.188*** (0.0326)
Income_S 6.210* (3.611) 6.251* (3.575)
Crime_R 2.64e 05 (0.00111) 4.82e 05 (0.00110)
Constant 28.59*** (1.585) 24.84*** (2.007) 9.929** (4.357) 15.61*** (4.547) 15.41*** (4.529)
Observations 167 167 167 162 162
R‐squared 0.170 0.263 0.352 0.557 0.557
VIF 1.51 1.6 1.56 2.16 2.19
Note. Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.

between social capital and depression rates. People in these counties face more crime rates and higher income
segregation, and higher social capital levels may expose them to more crimes and result in more mental stress.
When the determinants of depression in the northwestern and southeastern counties were further examined, so-
cioeconomic conditions, followed by natural environment factors, play important roles in depression prevalence in
northwestern counties, while the role of built environment is insignificant. Northwestern counties, which may be
because low density rural counties account for a large proportion of northwestern counties, thus depression
prevalence is less associated with built environment. Large income segregation is significant and increases people's
psychological stress and mental problems. Although Washington and Oregon are known for their high‐tech in-
dustry, there are still many counties focusing on agriculture and/or forestry. Counties in some states such as Idaho
and Montana are predominantly agricultural, and seasonal unemployment may be a key factor of high depression
prevalence. For natural environment, precipitation is associated with high depression rates. Located in a high
latitude region, northwestern counties usually go through longer overcast and rainy weather than counties in other
regions. High precipitation reduces sunlight and outdoor activities and is a main determinant of depression. US
Census Bureau's Household Pulse Surveys in December 2020 and January 2023 report that Seattle is among the 15
largest metro areas with the highest share of adults feeling down, depressed or hopeless.

For southeastern US counties, the high depression prevalence may be derived from demographic factors, then
followed by built environment factors. Southeastern US counties have high racial disparities, unemployment rates

WEI ET AL. 15 of 20
 GeoHealth 10.1029/2024GH001047

Table 5
Determinants of Depression in Southeastern Counties
Model 1 Model 2 Model 3 Model 4
Variables Dep Dep Dep Dep
Built Environment Sprawl_J 0.624*** (0.140) 0.575*** (0.129) 0.370*** (0.105) 0.174 (0.116)
Sprawl_P 0.256** (0.106) 0.102 (0.0922) 0.147* (0.0816) 0.0141 (0.101)
Mixed_Land 14.37*** (1.313) 12.07*** (1.160) 8.164*** (1.035) 4.126*** (1.325)
Mixed_Job 0.406 (0.841) 0.625 (0.729) 2.472*** (0.768) 0.439 (0.628)
Balance_JH 1.814** (0.915) 2.135** (0.831) 0.766 (0.644) 0.556 (0.824)
Density_P 0.479*** (0.0857) 0.677*** (0.109) 0.373*** (0.0809) 1.660*** (0.393)
Urban 0.532*** (0.134) 0.939*** (0.123) 0.865*** (0.106) 0.546*** (0.108)
Natural Environment Green_S 3.666*** (0.304) 0.955*** (0.288) 0.693** (0.319)
PM2.5 0.321*** (0.0381) 0.0608* (0.0354) 0.0713* (0.0367)
CO 13.07*** (1.296) 5.852*** (1.218) 3.626** (1.419)
Precipitation 0.235*** (0.0301) 0.316*** (0.0282) 0.155*** (0.0368)
Individual (or demographic) Factors Gender 11.56*** (2.052) 14.85*** (2.295)
Aging 14.93*** (1.788) 13.81*** (2.954)
Divorce_R 0.00825 (0.0219) 0.0132 (0.0139)
Race_C 9.554*** (0.424) 9.695*** (0.521)
Family_S 6.664*** (0.771) 1.212 (0.817)
Socioeconomic Conditions Social_C 0.663*** (0.216)
Growth_J 7.096* (4.113)
Rate_Unemp 0.102*** (0.0192)
Store_A 3.457*** (0.827)
Uninsured_P 0.0349** (0.0154)
Income_S 15.00*** (1.662)
Crime_R 0.000769*** (0.000246)
Constant 28.11*** (0.735) 21.36*** (0.818) 11.55*** (1.159) 2.331 (1.634)
Observations 2,043 2,042 2,042 1,672
R‐squared 0.167 0.322 0.509 0.587
VIF 1.42 1.47 1.60 1.89
Note. Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.

and poverty segregation, which have brought socioeconomic problems to the region. Study finds that the
hopefulness for the future can serve as a safeguard against people from the detrimental impact of adverse ones
(Lankarani & Assari, 2017). Compared with African American, non‐Hispanic whites usually have less confidence
in the future (Bailey et al., 2019). Thus, counties with high proportion of non‐Hispanic whites are positively
linked to high depression rates. The same as the regression based on all US counties, high female and less aging
people proportions are positively connected to high depression risk. The role of built environment is also
important. The efficient and proper utilization of land is conducive to leaving more space for leisure and
relaxation and relieving the stress of residents. Counties with more land mixed use, higher population density and
more urban area are associated with less depression disorders.
Different from the regression result based on all US counties (Model 4 in Table 2), we find some socioeconomic
conditions, that is, job growth and crime, only affect depression prevalence in the southeastern counties. As the
region with prominent socioeconomic problems, people living there may suffer from high social stress, violence
and crime, which increasing their risk of being depressed. The increased job growth rates provide employment
opportunities for the unemployed and alleviate socioeconomic problems such as poverty and income segmen-
tation, which reduces people's life pressure. Thus, both high job growth rates and decreased crime rates reduce
depression rates. Last, all other significant variables have the same effect with the regression results based on all

WEI ET AL. 16 of 20
 GeoHealth 10.1029/2024GH001047

US counties, except for the proportion of uninsured people. The high association between high proportion of
uninsured people and less depression may be because southeastern counties concentrate the majority of African
Americans with high share of adults without health insurance (Sood & Sood, 2021).

5. Conclusion and Limitation

5.1. Conclusion
Our study identifies important variables that may impact mental health across US counties, with the built
environment emerging as a significant factor. Urban sprawl is a prominent phenomenon in the US and is widely
recognized as a contributor to various urban problems. It has long been associated with increased anxiety and
depression, leading to negative impact on mental health (Garrido‐Cumbrera et al., 2018). However, our study
reveals that job sprawl is linked to lower depression prevalence, potentially due to increased access to job op-
portunities and open space. Land mixed use reduces commuting and life stress. Additionally, job mixed layouts
offer a more diverse range of job opportunities, increasing the likelihood of securing satisfying employment and
lowering the risk of long‐term unemployment. Thus, these factors exhibit positive associations with overall
mental well‐being.
It is important to acknowledge that urban sprawl may still contribute to anxiety and depression such as increasing
commuting distances (Garrido‐Cumbrera et al., 2018). It is crucial to determine the extent to which job sprawl can
effectively reduce depression prevalence. Implementing measures such as promoting mixed land use and diverse
job layouts can help alleviate the prevalence of depression.
The natural environment is also connected with mental health. We found that the presence of green space is
associated with lower depression prevalence, and this connection is particularly significant in urban areas.
Increasing green space within urban areas, fostering social connections, expanding employment opportunities,
and improving social security and welfare systems play significant roles in mitigating depression prevalence. As
expected, PM2.5 and CO pollution worsen mental health problems. Air pollution not only induces negative moods
but also poses a threat to physical health. Consequently, exposure to PM2.5 and CO is negatively correlated with
overall mental well‐being. Furthermore, our findings indicate that increased precipitation is linked to a higher
prevalence of depression.

The presence of social capital, which provides individuals with additional resources and robust social support
(Thoits, 2011), is linked to a lower prevalence of mental disorders. However, increasing social capital may also
increase peoples' stress in certain geographical contexts. Unemployment, high income segregation, limited store
access and insufficient insurance coverage, which have plagued poorer neighborhoods for a long time, tend to
exacerbate mental health problems.
Further analysis of influencing factors for the two high depression prevalence areas shows that socioeconomic
conditions and natural environment play important roles in changing depression prevalence in northwestern
counties. Unemployment, income segregation and precipitation are significantly associated with high depression
rates. Depression tends to be further exacerbated by precipitation and seasonal unemployment in these regions.
Implementing strategies such as expanding employment opportunities and offering unemployment subsidies can
play a crucial role in preventing and alleviating depression. Our examination of the effects of January temperature
and its interaction with precipitation on mental health revealed insignificant relationships. Consequently, the cold
winter itself does not significantly contribute to increased depression, and winter activities such as skiing are
popular and enjoyed in this region.

For southeastern counties, demographic factors, especially the racial composition, play significant roles in
influencing depression prevalence. The built environment is also important. Land mixed use, high population
density and more urban areas are linked to less mental disorders. It needs to be noted that some socioeconomic
factors, mainly job growth and crime, only affect the depression risk in southeastern counties.

5.2. Strengths and Limitations

Our study constructs a more comprehensive analytical framework to explore the correlations between built and
natural environments and the prevalence of depression, as well as their spatial heterogeneity. We have identified
the various effects of multidimensional urban sprawl on mental health, which has rarely been studied previously.

WEI ET AL. 17 of 20
 GeoHealth 10.1029/2024GH001047

Moreover, our study has shown that not all dimensions of urban sprawl is bad to mental health. We have provided
more nuanced analysis of the effect of natural environment on mental health, and that such effects also vary
between urban and rural areas. We also demonstrate that the effect of social capital on mental health depends on
places or local context. Further identification was conducted on the determinants of two areas (northwestern and
southeastern counties) with high depression prevalence, indicating again the significance of places in the
determination of mental health.
Our study also has limitations. In the aftermath of the COVID‐19 and associated lockdowns, concerns about
mental health have heightened significantly, which has prompted a surge in public awareness and engagement,
with more individuals actively participating in outdoor activities as a means to alleviate mental stress (Li &
Wei, 2023). The pandemic has thus raised the importance of built and natural environment on mental health.
Future research should consider the moderating effect of COVID‐19 epidemic on the relationship between built
and natural environments and depression. Also, more efforts are need to interpret the locally varied relationships
across different places.

Conflict of Interest
The authors declare no conflicts of interest relevant to this study.

Data Availability Statement

All of the data we use is publicly available, see Table 1 Data Sources.

Acknowledgments References
Research reported in this publication was
supported by the National Institute on Akoglu, H. (2018). User's guide to correlation coefficients. Turkish Journal of Emergency Medicine, 18(3), 91–93. https://doi.org/10.1016/j.tjem.
Aging of the National Institutes of Health 2018.08.001
under Award Number R01AG080440 and Almedom, A. M. (2005). Social capital and mental health: An interdisciplinary review of primary evidence. Social Science & Medicine, 61(5),
the Wasatch Environmental Observatory, 943–964. https://doi.org/10.1016/j.socscimed.2004.12.025
University of Utah. The content is solely Anselin, L. (1995). Local indicators of spatial association—LISA. Geographical Analysis, 27(2), 93–115. https://doi.org/10.1111/j.1538‐4632.
the responsibility of the authors and does 1995.tb00338.x
not necessarily represent the official views Astell‐Burt, T., Mitchell, R., & Hartig, T. (2014). The association between green space and mental health varies across the lifecourse. A lon-
of the National Institutes of Health. gitudinal study. Journal of Epidemiology & Community Health, 68(6), 578–583. https://doi.org/10.1136/jech‐2013‐203767
Bailey, R. K., Mokonogho, J., & Kumar, A. (2019). Racial and ethnic differences in depression: Current perspectives. Neuropsychiatric Disease
and Treatment, 15, 603–609. https://doi.org/10.2147/ndt.s128584
Barnes, D. M., & Bates, L. M. (2017). Do racial patterns in psychological distress shed light on the Black–White depression paradox? A sys-
tematic review. Social Psychiatry and Psychiatric Epidemiology, 52(8), 913–928. https://doi.org/10.1007/s00127‐017‐1394‐9
Behnisch, M., Krüger, T., & Jaeger, J. A. (2022). Rapid rise in urban sprawl: Global hotspots and trends since 1990. PLOS Sustainability and
Transformation, 1(11), e0000034. https://doi.org/10.1371/journal.pstr.0000034
Brueckner, J. K., Thisse, J. F., & Zenou, Y. (1999). Why is central Paris rich and downtown Detroit poor? An amenity‐based theory. European
Economic Review, 43(1), 91–107. https://doi.org/10.1016/s0014‐2921(98)00019‐1
Byomkesh, T., Nakagoshi, N., & Dewan, A. M. (2012). Urbanization and green space dynamics in Greater Dhaka, Bangladesh. Landscape and
Ecological Engineering, 8(1), 45–58. https://doi.org/10.1007/s11355‐010‐0147‐7
Chen, F., Zhang, X., & Chen, Z. (2023). Air pollution and mental health: Evidence from China Health and Nutrition Survey. Journal of Asian
Economics, 86, 101611. https://doi.org/10.1016/j.asieco.2023.101611
Chen, J., Chen, S., & Landry, P. F. (2015). Urbanization and mental health in China: Linking the 2010 population census with a cross‐sectional
survey. International Journal of Environmental Research and Public Health, 12(8), 9012–9024. https://doi.org/10.3390/ijerph120809012
Clifton, K., Ewing, R., Knaap, G. J., & Song, Y. (2008). Quantitative analysis of urban form: A multidisciplinary review. Journal of Urbanism,
1(1), 17–45. https://doi.org/10.1080/17549170801903496
Cuijpers, P., Karyotaki, E., Eckshtain, D., Ng, M. Y., Corteselli, K. A., Noma, H., et al. (2020). Psychotherapy for depression across different age
groups: A systematic review and meta‐analysis. JAMA Psychiatry, 77(7), 694–702. https://doi.org/10.1001/jamapsychiatry.2020.0164
Curtis, D. S., Kole, K., Brown, B. B., Smith, K. R., Meeks, H. D., & Kowaleski‐Jones, L. (2024). Social inequities in neighborhood health
amenities over time in the Wasatch Front Region of Utah: Historical inequities, population selection, or differential investment? Cities, 145,
104687. https://doi.org/10.1016/j.cities.2023.104687
Cvetkovski, S., Jorm, A. F., & Mackinnon, A. J. (2019). An analysis of the mental health trajectories of university students compared to their
community peers using a national longitudinal survey. Studies in Higher Education, 44(1), 185–200. https://doi.org/10.1080/03075079.2017.
1356281
Dang, Y., Dong, G., Chen, Y., Jones, K., & Zhang, W. (2019). Residential environment and subjective well‐being in Beijing. Environment and
Planning B: Urban Analytics and City Science, 46(4), 648–667. https://doi.org/10.1177/2399808317723012
De Silva, M. J., Huttly, S. R., Harpham, T., & Kenward, M. G. (2007). Social capital and mental health: A comparative analysis of four low income
countries. Social Science & Medicine, 64(1), 5–20. https://doi.org/10.1016/j.socscimed.2006.08.044
Finkelstein, A., Taubman, S., Wright, B., Bernstein, M., Gruber, J., Newhouse, J. P., et al. (2012). The Oregon health insurance experiment:
Evidence from the first year. Quarterly Journal of Economics, 127(3), 1057–1106. https://doi.org/10.1093/qje/qjs020
Garrido‐Cumbrera, M., Ruiz, D. G., Braçe, O., & Lara, E. L. (2018). Exploring the association between urban sprawl and mental health. Journal of
Transport & Health, 10, 381–390. https://doi.org/10.1016/j.jth.2018.06.006

WEI ET AL. 18 of 20
 GeoHealth 10.1029/2024GH001047

Greenlund, K. J., Lu, H., Wang, Y., Matthews, K. A., LeClercq, J. M., Lee, B., & Carlson, S. A. (2022). PLACES: Local data for better health.
Preventing Chronic Disease, 19, 210459. https://doi.org/10.5888/pcd19.210459
Guan, N., Guariglia, A., Moore, P., Xu, F., & Al‐Janabi, H. (2022). Financial stress and depression in adults: A systematic review. PLoS One,
17(2), e0264041. https://doi.org/10.1371/journal.pone.0264041
Hawton, K., i Comabella, C. C., Haw, C., & Saunders, K. (2013). Risk factors for suicide in individuals with depression: A systematic review.
Journal of Affective Disorders, 147(1–3), 17–28. https://doi.org/10.1016/j.jad.2013.01.004
Kent, S. T., McClure, L. A., Crosson, W. L., Arnett, D. K., Wadley, V. G., & Sathiakumar, N. (2009). Effect of sunlight exposure on cognitive
function among depressed and non‐depressed participants. Environmental Health, 8(1), 1–14.
Kim, G. E., & Kim, E. J. (2020). Factors affecting the quality of life of single mothers compared to married mothers. BMC Psychiatry, 20, 1–10.
https://doi.org/10.1186/s12888‐020‐02586‐0
Lankarani, M. M., & Assari, S. (2017). Positive and negative affect more concurrent among Blacks than Whites. Behavioral Sciences, 7(3), 48.
https://doi.org/10.3390/bs7030048
LaVeist, T. A., Thorpe Jr, R. J., Pierre, G., Mance, G. A., & Williams, D. R. (2014). The relationships among vigilant coping style, race, and
depression. Journal of Social Issues, 70(2), 241–255. https://doi.org/10.1111/josi.12058
Leslie, E., & Cerin, E. (2008). Are perceptions of the local environment related to neighbourhood satisfaction and mental health in adults?
Preventive Medicine, 47(3), 273–278. https://doi.org/10.1016/j.ypmed.2008.01.014
Li, H., & Wei, Y. D. (2023). COVID‐19, cities and inequality. Applied Geography, 160, 103059. https://doi.org/10.1016/j.apgeog.2023.103059
Liu, S., Ouyang, Z., Chong, A. M., & Wang, H. (2018). Neighborhood environment, residential satisfaction, and depressive symptoms among
older adults in residential care homes. The International Journal of Aging and Human Development, 87(3), 268–288. https://doi.org/10.1177/
0091415017730812
Lorant, V., Deliège, D., Eaton, W., Robert, A., Philippot, P., & Ansseau, M. (2003). Socioeconomic inequalities in depression: A meta‐analysis.
American Journal of Epidemiology, 157(2), 98–112. https://doi.org/10.1093/aje/kwf182
Mantler, A., & Logan, A. C. (2015). Natural environments and mental health. Advances in Integrative Medicine, 2(1), 5–12. https://doi.org/10.
1016/j.aimed.2015.03.002
Marmot, M., & Wilkinson, R. G. (2001). Psychosocial and material pathways in the relation between income and health. BMJ, 322(7296), 1233–
1236. https://doi.org/10.1136/bmj.322.7296.1233
McCloud, T., Kamenov, S., Callender, C., Lewis, G., & Lewis, G. (2023). The association between higher education attendance and common
mental health problems among young people in England: Evidence from two population‐based cohorts. The Lancet Public Health, 8(10), e811–
e819. https://doi.org/10.1016/s2468‐2667(23)00188‐3
McKee‐Ryan, F., Song, Z., Wanberg, C. R., & Kinicki, A. J. (2005). Psychological and physical well‐being during unemployment: A meta‐
analytic study. Journal of Applied Psychology, 90(1), 53–76. https://doi.org/10.1037/0021‐9010.90.1.53
McLeod, D. L. (2017). Children’s exposure to environmental toxins: Socioeconomic factors and subsequent effects on mental health and function.
Center for the Human Rights of Children, 13.
Moreno‐Agostino, D., Wu, Y. T., Daskalopoulou, C., Hasan, M. T., Huisman, M., & Prina, M. (2021). Global trends in the prevalence and
incidence of depression: A systematic review and meta‐analysis. Journal of Affective Disorders, 281, 235–243. https://doi.org/10.1016/j.jad.
2020.12.035
Pamplin, II, J. R., & Bates, L. M. (2021). Evaluating hypothesized explanations for the Black‐White depression paradox: A critical review of the
extant evidence. Social Science & Medicine, 281, 114085. https://doi.org/10.1016/j.socscimed.2021.114085
Parker, G., & Brotchie, H. (2010). Gender differences in depression. International Review of Psychiatry, 22(5), 429–436. https://doi.org/10.3109/
09540261.2010.492391
Piracha, A., & Chaudhary, M. T. (2022). Urban air pollution, urban heat island and human health: A review of the literature. Sustainability, 14(15),
9234. https://doi.org/10.3390/su14159234
Richardson, E. A., Pearce, J., Mitchell, R., & Kingham, S. (2013). Role of physical activity in the relationship between urban green space and
health. Public Health, 127(4), 318–324. https://doi.org/10.1016/j.puhe.2013.01.004
Richardson, R., Westley, T., Gariépy, G., Austin, N., & Nandi, A. (2015). Neighborhood socioeconomic conditions and depression: A systematic
review and meta‐analysis. Social Psychiatry and Psychiatric Epidemiology, 50(11), 1641–1656. https://doi.org/10.1007/s00127‐015‐1092‐4
Ross, C. E. (2017). Social causes of psychological distress. Routledge.
Ryan, S. C., Sugg, M. M., Runkle, J. D., & Thapa, B. (2024). Advancing understanding on greenspace and mental health in young people.
GeoHealth, 8(3), e2023GH000959. https://doi.org/10.1029/2023gh000959
Silva, M., Loureiro, A., & Cardoso, G. (2016). Social determinants of mental health: A review of the evidence. The European Journal of Psy-
chiatry, 30(4), 259–292.
Sood, L., & Sood, V. (2021). Being African American and rural: A double jeopardy from Covid‐19. The Journal of Rural Health, 37(1), 217–221.
https://doi.org/10.1111/jrh.12459
Stier, A. J., Schertz, K. E., Rim, N. W., Cardenas‐Iniguez, C., Lahey, B. B., Bettencourt, L. M., & Berman, M. G. (2021). Evidence and theory for
lower rates of depression in larger US urban areas. Proceedings of the National Academy of Sciences, 118(31), e2022472118. https://doi.org/10.
1073/pnas.2022472118
Sturm, R., & Cohen, D. (2014). Proximity to urban parks and mental health. The Journal of Mental Health Policy and Economics, 17(1), 19.
Thoits, P. A. (2011). Mechanisms linking social ties and support to physical and mental health. Journal of Health and Social Behavior, 52(2), 145–
161. https://doi.org/10.1177/0022146510395592
Tikkanen, R., Fields, K., Williams, R. D., & Abrams, M. K. (2020). Mental health conditions and substance use: Comparing US needs and
treatment capacity with those in other high‐income countries. The Commonwealth Fund.
Vanaken, G. J., & Danckaerts, M. (2018). Impact of green space exposure on children’s and adolescents’ mental health: A systematic review.
International Journal of Environmental Research and Public Health, 15(12), 2668. https://doi.org/10.3390/ijerph15122668
Van Hooff, M. L., & Van Hooft, E. A. (2014). Boredom at work: Proximal and distal consequences of affective work‐related boredom. Journal of
Occupational Health Psychology, 19(3), 348–359. https://doi.org/10.1037/a0036821
Ventriglio, A., Torales, J., Castaldelli‐Maia, J. M., De Berardis, D., & Bhugra, D. (2021). Urbanization and emerging mental health issues. CNS
Spectrums, 26(1), 43–50. https://doi.org/10.1017/s1092852920001236
Von Lindern, E., Hartig, T., & Lercher, P. (2016). Traffic‐related exposures, constrained restoration, and health in the residential context. Health
& Place, 39, 92–100. https://doi.org/10.1016/j.healthplace.2015.12.003
Wang, L., Zhou, Y., Wang, F., Ding, L., Love, P. E., & Li, S. (2021). The influence of the built environment on people's mental health. Sustainable
Cities and Society, 74, 103185. https://doi.org/10.1016/j.scs.2021.103185

WEI ET AL. 19 of 20
 GeoHealth 10.1029/2024GH001047

Wei, Y. D., & Ewing, R. (2018). Urban expansion, sprawl and inequality. Landscape and Urban Planning, 177, 259–265. https://doi.org/10.1016/
j.landurbplan.2018.05.021
Wei, Y. D., Xiao, W., Medina, R., & Tian, G. (2021). Effects of neighborhood environment, safety, and urban amenities on origins and desti-
nations of walking behavior. Urban Geography, 42(2), 120–140. https://doi.org/10.1080/02723638.2019.1699731
Wu, Y. T., Prina, A. M., Jones, A., Barnes, L. E., Matthews, F. E., Brayne, C., & CFAS, M. (2016). Land use mix and five‐year mortality in later
life: Results from the Cognitive Function and Ageing Study. Health & Place, 38, 54–60. https://doi.org/10.1016/j.healthplace.2015.12.002

WEI ET AL. 20 of 20