Greenspace and depression incidence in the US-based nationwide Nurses' Health Study II: A deep learning analysis of street-view imagery
Does having more visible nature, like trees and grass, around your home, as seen from street level, affect your risk of becoming depressed?
College of Health researcher(s)
Highlights
- We examined the association between greenspace and depression risk over two decades (2001–2017).
- Deep learning models applied to derive street view image-based greenspace exposure and linked to a US nationwide cohort.
- Street-view trees were associated with a lower risk of depression, particularly clinician-diagnosed depression.
- Street-view grass was associated with a lower risk of depression in areas with lower air pollutant levels.
- Our results support the potential role of urban tree planting interventions in reducing the risk of depression.
Abstract
Background
Greenspace exposure is associated with lower depression risk. However, most studies have measured greenspace exposure using satellite-based vegetation indices, leading to potential exposure misclassification and limited policy relevance. We examined the association of street-view greenspace measures with incident depression in a prospective cohort of US women.
Methods
We applied deep learning segmentation models to 350 million US street-view images nationwide (2007–2020) to derive ground-level greenspace metrics, including percentage of trees, grass, and other greenspace (plants/flowers/fields), and linked metrics to Nurses’ Health Study II participants’ residences (N = 33,490) within 500 m each year. Cox proportional hazards models estimated the relationship between street-view greenspace metrics and incident depression, assessed through self-report of clinician-diagnosed depression or regular antidepressant use and adjusted for individual- and area-level factors.
Findings
In adjusted models, higher percentages of street-view trees were inversely associated with incident depression (HR per IQR, 0.98; 95%CI: 0.94–1.01) and specifically clinician-diagnosed depression (HR per IQR, 0.94; 95%CI: 0.90–0.99). Higher percentages of street-view grass were also inversely associated with incident depression, but only in areas with low particulate matter (PM2.5) levels (HR per IQR, 0.79; 95%CI: 0.71–0.86). Results were consistent after adjusting for additional spatial and behavioral factors, and persisted after adjusting for traditional satellite-based vegetation indices.
Conclusion and relevance
We observed participants who lived in areas with more trees visible in street-view images had a lower risk of depression. Our findings suggest tree-planting interventions may reduce depression risk.
Frequently Asked Questions about Greenspace and Depression
How did this study investigate the relationship between greenspace and depression?
This study utilized deep learning models to analyze over 350 million U.S. street-view images collected between 2007 and 2020. These analyses quantified the percentage of different types of ground-level greenspace (trees, grass, and other green vegetation) visible within 500 meters of the residences of 33,490 participants in the Nurses' Health Study II (NHSII) cohort. The researchers then used statistical models to examine the association between these street-view greenspace metrics and the subsequent incidence of depression, assessed through self-reported clinician-diagnosed depression or regular antidepressant use, over a 16-year follow-up period (2001-2017).
What were the main findings regarding street-view greenspace and the risk of depression?
The study found that a higher percentage of trees visible in street-view images within 500 meters of participants' homes was associated with a lower risk of incident depression, particularly clinician-diagnosed depression. Specifically, an increase of 13.60% (the interquartile range) in street-view trees was associated with a 6% lower risk of clinician-diagnosed depression. Higher percentages of street-view grass were also linked to a lower risk of incident depression, but this association was only observed in areas with low levels of fine particulate matter (PM2.5) air pollution. No significant association was found between the percentage of other visible greenspace (plants, flowers, or fields) and the risk of depression.
How does this study's method of measuring greenspace differ from previous research?
Most prior studies examining the link between greenspace and depression have relied on satellite-based vegetation indices like the Normalized Difference Vegetation Index (NDVI). This study breaks from that by using street-view imagery and deep learning to assess ground-level exposure to specific types of greenspace as seen from the street. The authors argue that satellite measures may not accurately capture the greenspace features most relevant to mental health, such as street trees and aesthetic qualities of landscaping, nor do they differentiate between types of greenspace like trees and grass, which may have different impacts.
Why did the researchers focus on trees and grass as specific types of greenspace?
The researchers focused on trees and grass because they are prominent types of greenspace in residential environments that might influence health through different pathways. Trees, for example, are hypothesized to reduce stress and restore attention, encourage physical activity and social interaction, and mitigate environmental hazards like noise and air pollution. Grass areas, particularly in open spaces, may also promote physical activity and social engagement. By analyzing these separately, the study aimed to provide more specific insights into which types of ground-level greenspace might be most beneficial for mental health.
What role did air pollution (PM2.5) seem to play in the association between greenspace and depression?
The study found evidence of effect modification by PM2.5 levels on the association between street-view grass and incident depression. Specifically, higher percentages of street-view grass were associated with a lower risk of depression only in areas with low PM2.5 levels. In areas with medium to high PM2.5, this protective association was not observed. The researchers suggest that in cleaner environments, people may be more likely to utilize grassy areas for outdoor activities, leading to mental health benefits. However, in more polluted areas, increased outdoor activity in grassy areas might lead to higher exposure to air pollutants, potentially counteracting some of the mental health benefits. This interaction was not found for street-view trees, possibly because trees have a greater capacity to filter air pollutants.
How robust were the findings, and what sensitivity analyses were conducted?
The inverse association between street-view trees and incident depression was generally robust across various sensitivity analyses. These included adjusting for potential mediators like air pollution and physical activity, accounting for geographic region and precipitation, and even when also adjusting for traditional satellite-based greenness (NDVI), where the association with NDVI was nullified while the street-view tree association persisted. The findings were also consistent with less restrictive criteria for defining depression cases and were strengthened when the analysis was limited to data from 2007 onwards, when street-view imagery became available.
What are the potential implications of these findings for public health and urban planning?
The finding that more street-view trees are associated with a lower risk of depression suggests that urban greening initiatives, particularly those focused on tree planting in residential areas, could be a valuable strategy for promoting mental health in communities. The differential effect of grass depending on air quality highlights the importance of considering the broader environmental context when planning green spaces. The study also suggests that ground-level, visible greenspace, as captured by street-view imagery, may be a more relevant metric for mental health than broader satellite-based measures.
What are some limitations of the study that should be considered when interpreting the results?
The study has several limitations. The most appropriate geographical scale for measuring greenspace exposure is still uncertain. The study only assessed greenspace around residential addresses and lacked information on participants' daily movements and interactions with greenspaces elsewhere. Street-view images may not capture all relevant greenspaces, such as private backyards. The use of 2007 street-view data for earlier residential addresses (2001-2005) could introduce measurement error. Finally, the study population consisted primarily of white female nurses living across the US, which may limit the generalizability of the findings to other populations.