Association of Redlining and Natural Environment with Depressive Symptoms in Women in the Sister Study

Mental health disorders are a major contributor to the global burden of disease. In 2019, mental health disorders were the seventh and second leading cause of global disability-adjusted life years and years lived with disability, respectively.¹ The prevalence of mental health disorders has steadily increased over the past two decades, with almost a 50% increase from 2009 to 2019.² Globally, $\sim 970\text{\hspace{0.17em}million}$ people, equivalent to 12.6% of the world population, experienced a mental health disorder in 2019.¹ Among all the mental health conditions, depression is the most common, with higher prevalence for women than for men in almost all the countries worldwide.¹ The impact of mental health conditions on the global economy is substantial, with costs estimated to be around $\text{USD\hspace{0.17em}}\$2.5$ trillion per year in 2010 and projected to reach $\text{USD\hspace{0.17em}}\$6$ trillion by 2030.³ Recognizing this issue, the United Nations (UN) 2030 Agenda for Sustainable Development Goals (SDGs) identifies mental health as a key factor for achieving global development goals. Mental health has taken on even greater significance in recent years under the threat and disruption of the COVID-19 pandemic, with a reported 25% increase in anxiety and depression rates worldwide.⁴ In the United States, the prevalence of depressive symptoms in low-income groups has almost tripled in comparison with the rates in high-income groups during this period.⁵

Along with the awareness of mental health issues, there has been growing interest in the potentially beneficial effects of the natural environment on human health and well-being. Even small exposures to nature, such as window views,^6–9 may serve as an effective public health strategy for improving human health and well-being, which would go a long way in achieving the UN SDGs.¹⁰ The economic value of visiting nature for mental health improvement alone is estimated to be $\sim 4\%$ of the global ecosystem services value (estimated at $\text{USD\hspace{0.17em}}\$150$ trillion for the year 2018) per year.¹¹ The beneficial effects of natural environment on mental health are derived from myriad ecosystem services,¹² including cultural services, such as esthetic enjoyment of landscape,^13,14 and regulating services, such as pollutant removal and heat mitigation.^15–17 Although the definitions of natural environment vary across studies from a single type of vegetation (e.g., trees, grass alone) to a composite index of all natural elements [e.g., Normalized Difference Vegetation Index (NDVI)], studies are consistent in finding that exposure to the natural environment is associated with better mental health.^18–22 NDVI is calculated as the difference between near-infrared and red reflectance divided by their sum. The values range from $-1$ to 1, wherein zero and negative values generally indicate barren surfaces (e.g., rock and soil) and nonvegetated land (e.g., water bodies and ice), and greater positive values indicate denser green vegetation.²³ Previous research¹⁸ using NDVI to estimate overall natural environment reported that a 0.1 increase in NDVI within $500\mathrm{m}$ of residences was found to be associated with a respective odds ratio (OR) of 0.85 [95% confidence interval (CI): 0.76, 0.95] and 0.81 (95% CI: 0.70, 0.93) of having depression and anxiety in adults. With a focus on women, another study²⁰ reported that participants who lived in a neighborhood with the highest quartile of NDVI within $250\mathrm{m}$ of their residence had an associated 13% reduction in depression risk in comparison with those living in a neighborhood with the lowest quartile of NDVI. Although natural environment generally benefits mental health, ecosystem services derived from certain types of natural environment may be more relevant to mental health. For instance, many of the psychological benefits from nature are supported by the Attention Restoration Theory^24,25 and the Stress Reduction Theory,^6,26 which suggest that spending time in or simply looking at nature helps with recovery from mental fatigue, alleviation of anxiety, and stress reduction. These beneficial effects of natural environment may be largely determined by human perception or landscape preference. An experimental study²⁷ reported that people had a preference for higher density of street trees but no preference for herbaceous cover. Other empirical research²² observed a positive association of better mental health with tree cover but not herbaceous cover.

However, exposure to the natural environment is not equitably distributed across the population, with socioeconomic status as a primary basis of inequity and environmental injustice.^28,29 Urban greenery is found to be positively associated with income and educational attainment (i.e., undergraduate or graduate degree) and negatively associated with percentage of minoritized racial and ethnic groups and percentage of people living in poverty across neighborhoods in the United States.^30–32 Likewise, groups experiencing greater disadvantage tend to live in areas with more impervious surfaces and therefore are more susceptible to increased temperature or urban heat island effects.³³ Moreover, minoritized and low-income populations tend to live in areas that are subject to more changes in land use/land cover for development and therefore are more susceptible to loss of nature and to deprivation of benefits from ecosystem services.³⁴

The causes of inequitable distribution of access or exposure to natural environment are complex.^35,36 Historical housing policies implemented at local, state, and/or federal levels are potential factors driving inequitable access to nature. For instance, the U.S. Federal Home Owners’ Loan Corporation (HOLC) established a policy in the 1930s to identify certain neighborhoods as too risky for mortgage investment. This policy, referred to as redlining, is considered to be a major contributor to persistent racial segregation in the United States.^37,38 The HOLC was tasked to create residential security maps to evaluate the risk of housing foreclosures for mortgages in 239 U.S. cities. The HOLC maps were created based on many factors, including housing quality, income, and racial profiles.^39,40 Neighborhoods were mapped and assigned a letter grade and a color code: A (best, green), B (still desirable, blue), C (declining, yellow), and D (hazardous, red - thus the term “redlined”). The neighborhoods assigned a grade of D were typically characterized by poor housing quality, predominately non-White and immigrant residents, and/or presence of hazardous environmental exposure sources. Residents who lived in the neighborhoods receiving the worst grade were often declined for loans or favorable mortgage terms,⁴¹ resulting in long-term disinvestment in the development and growth of these neighborhoods.⁴²

This historical pattern of racial discrimination driven by redlining has shaped contemporary disparities in environmental exposure,^43–46 climate adaptation,^33,47 and health conditions⁴⁸ between urban neighborhoods. Nardone et al.⁴⁴ observed that both present-day (2010) and estimated historical (1930) NDVI values consistently decreased across HOLC Grades A to D. With a specific focus on trees, Locke et al.⁴⁵ reported that neighborhoods formerly redlined had, on average, 20% less tree cover in comparison with neighborhoods formerly assigned a green or A grade in 37 HOLC cities. A later study⁴⁶ found similar results when including all cities with digital HOLC data available from the Mapping Inequality Project,⁴¹ with $\sim 20\%$ less tree cover in Grade D neighborhoods in comparison with Grade A neighborhoods. In addition, the estimated difference in ecosystem service values derived from tree cover in Grade A vs. D neighborhoods was $\sim \text{USD\hspace{0.17em}}\$112\text{\hspace{0.17em}million}$ annually.⁴⁶

The relative lack of ecosystem services derived from tree cover in redlined neighborhoods may play a role in the disproportionate health burden in these communities. There is a higher prevalence of poor mental health observed in redlined neighborhoods vs. nonredlined neighborhoods.^49,50 Although the influence of redlining on the distributions of natural environment^44–46 and on health conditions has been studied,^48,51–53 the influence of redlining on the relationships between natural environment and mental health has not been previously investigated. This study attempts to address this research gap by examining associations in the variability of tree cover, depressive symptoms, and HOLC grade using data from a nationwide U.S. cohort of women.

We observed a statistically significant association between current neighborhood greenery (tree cover) and women’s depressive symptoms. This association and neighborhood greenery independently varied by HOLC grade. The importance of the natural environment in promoting human health and well-being is well-recognized.^84,85 Just as adverse effects of the environment have been shown to disproportionately impact socioeconomically disadvantaged communities,^28,29 there is concern that this burden may be compounded by the unequal distribution of benefits derived from the natural environment. To achieve the goals for healthy and sustainable communities, it is critical to understand the disparities in environmental exposure, health, and their relationships across communities. Recognizing the public health importance of mental health with a consideration of historical economic practices of redlining, we examined the present-day associations between neighborhood greenery and depressive symptoms across neighborhoods with historically defined HOLC grades.

Our findings on the depressive symptoms are consistent with previous research reporting that redlined neighborhoods had higher prevalence of poor mental health.⁵⁰ In this analysis, CES-D scores and rates of having depressive symptoms tended to be higher in neighborhoods with the less-desirable HOLC grades. Consistent with Nardone et al.,⁴⁴ we observed that exposure to the natural environment as estimated by NDVI is significantly lower in historically redlined neighborhoods in comparison with those that were not. In addition, the observed differences of tree cover in Grade D and A neighborhoods were not only one-time phenomenon but consistent throughout the survey period, with findings similar to those observed in previous studies.^45,46 At the address level, neighborhoods formerly assigned as Grade D had on average 20% and 13% less tree cover at the $500\text{-m}$ and $\mathrm{2,000}\text{-m}$ buffer, respectively, in comparison with neighborhoods formerly assigned as Grade A. The greater difference observed at the smaller buffer size is likely attributable to a smaller classification error because there is greater probability that the larger buffer area will stray into adjacent areas that were assigned other HOLC grades. Because environmental exposures are not restricted by administrative boundaries, further investigation is needed to understand the effects of redlining in differently graded adjacent neighborhoods.

Although the contemporary disparities in the distribution of neighborhood greenery and health conditions are both associated with the historical patterns of racial discrimination, the effects of neighborhood greenery on health outcomes can be complex. Although some studies suggest that there is a minimum threshold of exposure to natural environment for meeting better health outcomes,^22,86 small amounts of neighborhood greenery, such as those observed in the redline neighborhoods, may not show evident effects on health outcomes. However, stratified analysis by HOLC grade reveals that more tree cover is associated with reduced odds of having depressive symptoms across all the HOLC grades. The benefits of tree cover on depressive symptoms appear not to be linear across the HOLC grades but rather much more potent in redlined neighborhoods, where a wide range of ecosystem services could be missing because of the scarcity of tree cover.^44,46 Disadvantaged groups are likely to suffer precipitous and cumulative impacts associated with poverty and race, including diminished access to health care,^87–89 air pollution,^43,90,91 and noise,^92,93 which may lead to a greater gap for improvement in health conditions and environmental exposures (e.g., reduction in air pollution and increase in neighborhood greenery) in comparison with those in nondisadvantaged groups.⁹⁴ Thus, even with only 13% of the participants residing in redlined neighborhoods, we observed much stronger effects of neighborhood greenery on depressive symptoms.

Our findings suggest increased tree cover as a strategy for addressing environmental justice issues. Regardless of the small amount of tree cover in redlined neighborhoods, an increase of the same percentage of tree cover shows greater effects on improving mental health in redlined neighborhoods in comparison with nonredlined neighborhoods. Given that redlined neighborhoods tend to have greater risks for climate impacts such as heat exposure³³ and air pollution,⁴³ which are found to associate negatively with many health outcomes,^95–99 increasing tree cover not only has the potential to reduce the prevalence of depression but also to enhance overall resilience to climate impacts. In addition, increases in tree cover may bring co-benefits for health promotion and the environment through creating opportunities for physical activity and social interactions and through reducing air pollution. An investment of tree planting in vulnerable communities may offer an effective solution for reducing disparities in both environmental exposures and health conditions. The beneficial effects from tree cover may even be more pronounced during public health emergencies, such as the COVID-19 pandemic, whereby many studies reported that disadvantaged communities have greater risks of adverse health outcomes.^100–102 Although there are complex social and environmental factors involved, our findings support that contact with natural environment may help alleviate mental health issues, especially during public health crises.^9,103,104 However, we caution that efforts are needed to ensure that communities are part of decision-making processes when addressing disparities. Factors such as location and types of trees planted, clear descriptions of benefits conveyed, and potential impacts on housing prices should be carefully considered.

The strengths of this analysis include the availability of individual-level data on health outcome and covariates from a large U.S. national cohort, as well as environmental exposures at multiple buffer sizes. The Sister Study collects a wealth of information on biological, sociodemographic, lifestyle, and health-related factors from participants across the United States⁵⁴ that allowed us to build a model accounting for multiple potential covariates. The broad geographic distribution of Sister Study participants provided us an opportunity to systematically examine the effects of historical HOLC grades on the associations between neighborhood greenery and depressive symptoms. The greenery measures used in this analysis represent the proportion of tree canopy within each $30\text{-m}$ pixel, which provide sufficient accuracy for estimating tree canopy across the United States.¹⁰⁵ Because the distribution of vegetation is greatly affected by climate, the use of a generalized linear mixed effect model accounts for potential clusters by climate zone.

There are also study limitations to consider in the interpretation of these results. First, study findings are based on associations derived from cross-sectional analysis limiting causal inference. Second, although there are aspects of the Sister Study cohort that are nationally representative,⁵⁴ the cohort consists only of women and is biased in some important ways, including a bias toward non-Hispanic White, well-educated, women over 55 y of age and with a sister diagnosed with breast cancer. Furthermore, by focusing on HOLC cities that included a greater percentage of participants who were non-White, these findings may not generalize to the entire Sister Study cohort. In addition, by focusing only on cities subject to HOLC practices, we eliminated most of the participants in the cohort and therefore limited the power of our statistical analysis. To address this limitation, we analyzed the Sister Study full cohort and the subset sample of participants who were not residing in areas subject to HOLC practices. The results were consistent with the observation in the main analysis that increases in tree cover were associated with reduced ORs of having depressive symptoms. However, the effects were smaller and less evident.

We were able to account for many individual-level covariates but not all. Although redlined neighborhoods are more subject to gentrification, we do not have data to account for this factor. Previous research found that not everyone prefers to have trees in their neighborhoods.^106,107 We do not have information about participants’ preferences for landscaping. Likewise, we do not have information to address potential residential self-selection bias. Another limitation related to model adjustment is that we did not account for potential seasonal effects. Others have shown that season has an influence on depressive symptoms.^108–110 We examined season of CES-D-10 collection as an additional covariate. However, we did not observe differences in results for the models with vs. without season as a covariate. Because the participants in this analysis are mostly over 55 y of age, these findings are consistent with previous research¹¹¹ reporting that the effects of seasonal patterns on mental health were reduced for people age 55 y and older.

Harmful environmental factors, such as air pollution, are inversely associated with better mental health^112,113 but can be mitigated by tree cover.¹¹⁴ As with tree cover, exposure to air pollution is not equitably distributed, where redlined neighborhoods tend to have higher air pollution levels.⁴³ Although we did not account for harmful environmental factors such as air pollution, future research should consider including both beneficial and detrimental environmental factors to understand the compounded disparities in the effects of environmental exposures on health outcomes.

The distribution of vegetation is greatly affected by climate^79,80 and may subsequently influence the relationships between neighborhood greenery and health outcomes. Previous studies^115,116 observed stronger associations between neighborhood greenery and health outcomes for people living in arid climates, where the vegetation is scarce, in comparison with continental or temperate climates. People who resided in historically redlined neighborhoods in an arid climate may be even more susceptible to potential risks of climate impacts. Due to our sample size, we were not able to evaluate whether the relationships between neighborhood greenery and depressive symptoms vary across climate zones for people who resided in historically redlined neighborhoods. However, our model approach accounts for the clustering by climate zone so that our results are generalizable across climate zones. We acknowledge that the effects of greenery on depressive symptoms by climate zone is of interest and see an important need to build on this line of research.

Relationships between environments, human health, and socioeconomic factors are complex. There are increasing efforts to address environmental and social justice issues across the United States, including President Biden’s Executive Order 13990, titled “Protecting Public Health and the Environment and Restoring Science to Tackle the Climate Crisis.”¹¹⁷ We build on the existing literature by identifying the effects of neighborhood greenery on depressive symptoms and the contemporary disparities in environmental exposures created by the housing policy developed almost a century ago. We further extend current knowledge to understand the variations in the relationships between neighborhood greenery and depressive symptoms in the historical HOLC cities. Findings from this analysis not only provide additional evidence on the benefits of neighborhood greenery on human health but also identify the communities at risk and suggest a remediation strategy focused on neighborhood greenery for those communities continuing to experience legacy environmental injustice because of the economic policies of the 1930s.

The authors would like to thank T. Luben and J. Woo for their time and thoughtfulness in providing valuable feedback. The authors would also like to thank the Sister Study Epidemiology Team for all their support and thank our former U.S. EPA ORISE colleagues, R. Silva and F. Cochran for their efforts in establishing this collaborative effort between the U.S. EPA and the NIEHS. Finally, the authors are indebted to our former EPA colleague, L. Jackson, for her dedication and remarkable contributions in fostering the eco-health research field.

This work was funded, in part by the Intramural Research Program of the NIH, NIEHS (ZO1 ES-044005 to D.P.S.).

This paper has been reviewed by the Center for Public Health and Environmental Assessment in the Office of Research and Development at the U.S. EPA and approved for publication. Approval does not signify that the contents reflect the views and policies of the Agency, nor does the mention of trade names of commercial products constitute endorsement or recommendation for use.

The authors declare they have nothing to disclose. They have no conflict of interest and no actual or potential competing financial interests.