-
A systematic review of the association between urban green space and health is presented.
-
The focus has been on socio-economic and socio-demographic confounders.
-
Results show some positive association between urban green space, mental health and cardiovascular diseases.
-
Evidence is rather weak, and socio-economic confounders have the highest impact on health outcome effects.
-
The mediating effect of urban green space to decrease health inequality among different socio-economic groups may be more important.
5.1 Introduction
5.2 Methods
5.3 Results
References | Country, city | Aim | HEALTH | Study design | Green space metrics and buffer | Confounder | Main results |
---|---|---|---|---|---|---|---|
Abelt and McLafferty (2017) | U.S., New York | Assessment of relationship between mothers’ exposure to green and blue spaces and adverse birth outcomes. | Birth outcome | Mixed-effects linear and logistic regression models; GIS analysis | NDVI in buffers of 100 m, 250 m and 500 m, NYC Street Tree Census | Neighborhood deprivation, education, race, marital status, population density, birth place | No consistent significant relationship between adverse birth outcomes and residential greenness in fully adjusted models. Deprived tracts were less “green” and closer to major roads and industrial land use. Significant association between street trees surrounding the home and reduced odds of preterm birth were identified |
Allan et al. (2017) | UK, England | Investigation of health variations by residential contexts, its causes and sensitivity to rural–urban classifications | General health (Limiting long-term illness) | Logistic regression models applied to UK census data (2001) | Rural-urban classification, no buffer used | Employment status, education, age, sex, marital status, ethnicity | A positive urban–rural health gradient was identified, with the exception of a protective “capital city” effect. |
Astell-Burt et al. (2014) | UK, country scale | Analysis of variation in minor psychiatric morbidity | Mental health | Longitudinal study with data from nine annual waves of the British Household Panel Survey (1996–2004) using multi-level linear regression | Green space, no buffer used | Employment status, income, education, tenure status, age, gender, marital status | Psychiatric morbidity appeared less prevalent among residents living in greener neighbourhoods. In those areas, residents had higher household income. The link between urban green space and health varied across the lifecourse |
Calogiuri et al. (2016) | Norway, country scale | Examination of green exercise as a physical activity across social groups. | Physical activity | Questionnaire survey (2168), reported weekly green exercise and other forms of physical activity, specific forms of GE, and perceived factors that promote GE | Green exercise (=physical activity taking place in natural environments), no buffer used | Income, education, sex, age | Green exercise was represented among the elderly. No associations were identified to sex, centrality, education level or household income |
Cohen-Cline et al. (2015) | U.S., Washington | Assessment of the association between green space access and self-reported depression, stress and anxiety in between twin pairs. | Mental health (self-reported depression, stress and anxiety) | Multi-level random intercept model of same-sex twin pairs (4338 individuals) from the community-based University of Washington Twin Registry | NDVI based, no buffer used | Income, neighborhood deprivation, sex, age, ethnicity, population density, physical activity | Only the association between green space and depression was significant in fully adjusted models. Association between green space and stress became non-significant when adjusted for confounders |
Cusack et al. (2017) | U.S., Portland and Austin | Identification of a relationship between multiple urban green space metrics and associations with term birth weight. | Birth outcome (term birth weight) | Vital Statistics data for a birth cohort (2005–2009). Correlation between green space metrics and mixed models | NDVI, % tree cover, % green space, % street tree buffering and access to parks (using US EPA EnviroAtlas Data), multiple buffer distances: 50 m, 250 m, 300 m, 500 m, 1000 m | Education, income, age, sex, ethnicity, population density | In fully adjusted significant associations disappeared. Maternal race, ethnicity and education as confounders were mostly reducing green space and birth weight associations |
Feda et al. (2015) | U.S., Buffalo | Assessment of the association between neighbourhood park area and perceived stress among adolescents, while controlling for physical activity. | Mental health (perceived stress) | 10-week Neighbourhood Environment study, GIS analysis, multivariate regression model | Parks, classification of high/low park access using a park access index, 800 m buffer distance | Socio-economic status (SES), race/ethnicity, housing density, physical activity | Park area was significantly associated to perceived stress also when controlling for socio-economic status. Physical activity and gender were not significant as confounder in fully adjusted models |
Flouri et al. (2014) | UK, England | Analysis of the effect of urban neighbourhoodgreen space on young children’s emotional and behavioural adjustment and resilience. | Mental health (emotional and behavioural problems from early to middle childhood) | 2000–2007 Millenium Cohort Study data on families in England whose children had emotional and behavioural problems living in urban neighbourhoods | Neighbourhood green space was measured using the 2001 Generalised Land Use Database, excluding domestic gardens, no buffer used | Tenure status, income support, poverty, education, sex, ethnicity, age, household composition | Green space appeared unrelated to child adjustment on the whole, but it predicted emotional resilience. Poor children with more green space in their neighbourhood had fewer emotional problems from age 3 to 5, relative to those living in less green neighbourhoods |
Kihal-Talantikite et al. (2013) | France, Lyon | Exploration of a risk cluster of infant mortality according to neighbourhood characteristics | Infant mortality | Cluster analysis assessing spatial aggregation. 715 cases of infant deaths (2000–2009) obtained by city halls in Lyon | Green space index, calculated by using land cover datasets, no buffer used | Neighbourhood deprivation | Spatial distribution of infant mortality resulted to be not random but the high-risk cluster disappeared after controlling for greenness level and socio-economic deprivation. Both deprivation and greenness levels seem to have an independent effect |
Jenkin et al. (2015) | New Zealand, country scale | Identification of neighbourhood characteristics associated with children’s unhealthy behaviour. | Obesity (BMI), and weight-related behaviours (physical activity, diet, etc.) | Logistic regression models with individual-level data for children from 2006/7 New Zealand Health Survey | Green space (measured in distance and proportion in neighbourhood), no buffer used | Neighbourhood deprivation, income, education, employment status, tenure status, age, sex, ethnicity, household composition | Greater access to green space was significantly associated with lower sugar-sweetened beverage consumption next to neighbourhood deprivation, which was also significantly positively associated with other negative health behaviours |
McMorris et al. (2015) | Canada, country scale | Analysis of associations between residential greenness and physical activity. | Physical activity | 2001 Canadian Community Health Survey data, logistic regression | NDVI around home, buffer distances: 30 m, 500 m | Income, sex, age, marital status | Association was most significant for those in the higher income groups but positive associations were observed between greenness and physical activity in all income groupings |
Higher income households lived in areas with higher greenness | |||||||
Michael et al. (2014) | U.S., Portland | Examination of the effect of a neighborhood-changing intervention on changes in obesity in older women | Obesity (BMI) | Retrospective cohort design, change in BMI and neighborhood built environment over 18-year period (1986–2004) among older women; structured interviews and clinical examinations | Green space proximity using Euclidian distance from participant’s residence to closest edge of the nearest public park or green space. no buffer used | Occupational manual labor = employment status?, socio-economic status (SES), age, education | No significant association between changes in neighborhood walkability or parks and green spaces and variation in BMI over time was identified in fully adjusted models. SES, education, age among others appeared to be significant in adjusted models |
Mitchell et al. (2015) | Europe, urban areas | Identification of potential neighbourhood characteristics associated with narrower socio-economic inequalities in mental well-being. | Mental health | Multi-level regression models using data of the 2012 European Quality of Life Survey | recreational/green areas non-specific, no threshold, participants were asked about accessibility (with great difficulty, with some difficulty, etc.) | Education, employment status, sex | Socio-economic inequality in mental well-being was significantly narrower among respondents reporting good access to green/recreational areas, compared with those with poorer access |
Mukherjee et al. (2017) | India, Delhi | Analysis of the association between park availability and major depression. | Mental health (depression) | Questionnaire survey data from the baseline cross-sectional survey of the Centre for Cardio-Metabolic Risk Reduction in South Asia (CARRS); mixed-effects logistic regression models; development of park availability index with GIS | Park availability index with a buffer of 1-km distance | Education, employment status, income, age, gender, marital status, household composition | In adjusted models a significant association between area of nearest park and major depression remained and smaller park areas were associated with higher odds of depression |
Nichani et al. (2017) | New Zealand, country scale | Analysis of association between green space exposure during pregnancy and depression | Mental health (antenatal depression) | Adjusted logistic mixed effect models; data from the Growing Up in New Zealand study, a longitudinal pre-birth cohort study; GIS analysis | Proportion of green space within Census Area Unit including parks, beaches, urban parklands/open spaces, forests, grasslands, and croplands, but excluded private gardens, no buffer used | Education, employment status, neighbourhood deprivation, age, ethnicity, marital status, socio-economic status | No associations between green space and antenatal depression could be found. Even when adjusted for relevant factors such as deprivation |
Padilla et al. (2016) | France, Nice | Assessment of diverse socio-economic, health, accessibility, and exposure factors to explain how they are linked to environmental health inequalities related to infant and neonatal mortality | Infant and neonatal mortality | Creation of environmental indicators with GIS. Multiple Component Analysis, Hierarchical Clustering and standard Poisson regression models using official statistics collected from death certificates in the city halls of Nice municipalities | Green space provided by Corine Land Cover data including natural area but not agricultural areas. Proportion of green area in census block with 10 m2 of green space per habitant | Neighbourhood deprivation | Identified a significant relationship between infant and neonatal mortality risk and level of deprivation, but the link to urban green space and other exposure variables was not entirely clear. Deprivation was significantly positively correlated to proximity to high-traffic roads and negatively correlated to the proportion of green space as well as to the distance to respective health-care facilities |
Paquet et al. (2013) | Australia, Adelaide | Assessment of associations between accessibility, greenness, size, and type of public open spaces and clinical risk markers for cardiometabolic diseases and mediating factors. | Cardiometabolic risk | Data from the 2000–2003 North West Adelaide Health Study, a longitudinal cohort; Poisson regression models; GIS analysis | NDVI, accessibility of public open spaces within 1000 m road-distance from participant’s residence buffer | Education, income, socio-economic status (SES), gender, age | The number and proportion of public open spaces were not statistically significantly related to cardiometabolic health but greenness, size, and type were inversely related to cardiometabolic health |
Pearson et al. (2014) | New Zealand, country scale | Assessment of environmental factors in reducing obesity and promoting physical activity. | Weight (overweight, obesity, both), weight-related behaviours (physical activity, vegetable and fruit consumption) | Logistic regression models using data from the 2006/2007 New Zealand Health Survey based on face-to-face interviews | environmental characteristics (e.g. food outlets and green space [...], Accessibility of useable green space, Proportion of meshblock consisting of useable green space), no buffer used | Neighbourhood deprivation, age, gender, ethnicity | Increased neighbourhood deprivation and decreased access to neighbourhood green space were significantly associated with risk of overweight and/or obesity. Increased access to green space was associated with high levels of walking. Significant trend for low levels of walking was additionally positively associated with neighbourhood deprivation |
Richardson et al. (2017a) | U.S., Pittsburgh | Examination of influence of neighbourhood green space, walkability and crime on physical activity in low-income African American adults. | Physical activity | Self-reported data on demographics, functional limitations, objective measures of physical activity using accelerometry, neighbourhood green space using GIS, and walkability via street audits in 791 predominantly African-American neighbourhoods | Calculation of green space area surrounding each participant’s household within 1-km network buffer | Income, education, number of crimes reported, age, marital status | No significant association was found between neighbourhood green space, walkability, incidents of crime and physical activity |
Richardson et al. (2017b) | UK, Scotland | Assessment of a potential relationship between neighbourhood natural space and private garden access and children’s developmental change over time | Mental health (emotional and behavioural difficulties) | Longitudinal data of 4- to 6-year-old urban children from the Growing Up in Scotland (GUS) survey; random-intercept repeated-measures linear models | Area (%) of total natural space and parks within 500 m of child’s home using Scotland’s Greenspace Map, buffer of 500 m | Education, income, neighbourhood deprivation, age, sex | Groups with more park or total natural space close to their homes had slightly better social, emotional and behavioural health outcomes. Having access to a garden was related to sizeable mental health benefits (children from degree-educated households over those from households with no educational qualifications) |
Roe et al. (2016) | UK, London, West Midlands, Greater Manchester | Exploration of relationship between general health and individual, social and physical environmental predictors in deprived white British and black and minority ethnic groups. | Self-reported general health, physical activity | Household questionnaires (face-to-face interviews); Chi-Squared Automatic Interaction Detection (CHAID) segmentation analyses, Correlated Component Regression analyses | Wards identified via land use classifications, perception and use of local green space asked in interviews, no buffer used | Income, employment status, age, ethnicity, gender | General health in the “worst” health group (i.e. ethnicities within the Mixed group) was significantly more likely to be predicted by urban green space usage and perception variables than in any other ethnic group |
Roe et al. (2017) | UK, two areas in Central Scotland | Analysis of the effect of urban nature on stress reduction in different segments of deprived urban communities. | Mental health (stress reduction) | A cross-sectional household questionnaire administered by a survey company, using a face-to-face, computer-assisted interviews; Latent Class Analysis | Green space includes parks, woodlands, scrub and other natural environments. Perception of green space access, quantity and quality and motivation for visiting green space was asked in survey, quantity of green space through a datazone green space measure, no buffer used | Education, neighbourhood deprivation, age, gender | Our study found that the opportunities in the immediate neighbourhood for stress reduction vary by age. Stress coping in youth is likely supported by being social and keeping physically active outdoors, including local green space visits. By contrast, local green space appears not to support stress regulation in young–middle-aged and older adults, who choose to stay at home |
Thompson et al. (2016) | Scotland, Edinburgh and Dundee | Assessment if the use of local green space is associated with stress in deprived urban communities. | Mental health (stress) | Cross-sectional survey of 406 adults in four communities of high urban deprivation in Scotland, UK; Correlated Component Regression | Self-reported access to green space asked in survey. Green space based on reclassifications of the Ordnance Survey MasterMap and a city-wide audit of green space for Edinburgh and cross-referencing, no buffer used | Education, neighbourhood deprivation, employment status, age, sex, marital status | Quantity of green space in the neighbourhood, and access to garden or allotment, were significantly predicting of stress in deprived communities |
Triguero-Mas et al. (2015) | Spain, Catalonia | Assessment of the association between natural outdoor environments and general and mental health. | Self-reported general and mental health | Cross-sectional data from adults interviewed in Catalonia (Spain) between 2010 and 2012 as part of the Catalonia Health Survey; logistic regression and negative binominal models | NDVI, access to green spaces, access to blue spaces, exposure to natural outdoor environments within 300 m from residence; sensitivity analyses used 100-m, 500-m, and 1- km buffers | Education, socio-economic status (SES), gender, age, marital status, population density | Green spaces were significantly associated with better self-perceived general health and better mental health, particularly. Results remained significant for different buffers, and when stratifying for socio-economic status. No association between green spaces and social contacts and physical activity. The results for blue spaces were not conclusive |
Xu et al. (2017) | China, Hong Kong | Analysis of the relationship between green space and mortality. | Mortality (cardiovascular, diabetes, chronic respiratory, lung cancer mortality) | Mortality data for every registered death during the period of 2006–2011 from the Hong Kong Census and Statistics Department; Negative binomial regression models; Generalised Estimating Equations; GIS analysis | NDVI, no buffer used | Income, education, employment status, population density, sex | Higher NDVI values were significantly associated with lower cardiovascular and diabetes mortality, and non-significantly associated with lower chronic respiratory mortality. Associations were stronger for males and low-income area residents |