Recently, several reviews concluded that contact with green space can benefit health. Strong evidence was found for the short-term restorative effects of contact with nature, suggesting potential health benefits from reductions in chronic stress [1
], and for a negative (protective) association between the amount of green space and all-cause mortality [2
]. Evidence for the beneficial impact of green on general health, wellbeing, and physical activity was weaker. For mental health, the conclusions differed between reviews, with one concluding that there was strong evidence for a positive relation between the amount of green space and mental health [2
] and another concluding there is only weak evidence for this relation [3
Differences between population groups in their response to green space remain underexplored and have been suggested as an important focus for future research [1
], especially because the few studies that considered subgroups within the population reported mixed findings [2
]. Studying the relation between neighbourhood green space and health for specific population groups is important, because it provides information about who might benefit most from green interventions. Policy makers can use this knowledge to improve the effectiveness of their public health interventions.
Several studies reported a stronger, positive relationship between green space and health for specific population groups, such as children, older people, housewives, and people with a lower socioeconomic status (SES) [4
]. It has been suggested that these population groups profit more from a green living environment because they spend more time in their neighbourhood [9
] and, therefore, are more exposed to the neighbourhood green space than their counterparts [4
]. In the case of low SES groups, additional mechanisms have been suggested. Low SES groups might depend more on local green facilities for physical exercise and other activities than higher SES groups, who, because of their better financial situation and higher mobility, are less dependent on the residential proximity to green spaces for the use of these spaces [4
]. Furthermore, low SES groups might benefit more from local green space because their generally poorer health offers more opportunity for health improvement [10
As stated, the evidence for the beneficial health impacts of green space for specific population groups is not consistent. The results do not consistently pinpoint the same population groups as those that profit the most and they are inconsistent with regard to different health outcomes. Studies from the United Kingdom, for example, reported that more green space was associated with lower rates of cardiovascular and respiratory diseases in men, but not in women [11
]. Moreover, the presence of more green space was associated with better mental health for men at any age, but only for older women [12
]. In the Netherlands, associations between neighbourhood green space and health among housewives and the elderly were found for self-reported health symptoms, but not for general and mental health [4
]. Another Dutch study reported green space having beneficial (general) health effects in children and the elderly, especially in the most urbanised areas [7
]. Furthermore, the health benefits of green space for low SES groups are also not consistently reported, since no SES moderation was found in studies from Spain and the United States [13
The inconsistencies in the literature might reflect that the health benefits of neighbourhood green space not only differ between population groups, but could also depend on specific neighbourhood circumstances, such as the urbanisation level of the area. Therefore, different geographical settings should be investigated using identical measures for green space, health, and demographic characteristics. In the current study, we examined the relationship between green space and health among subpopulations in four European cities: Stoke-on-Trent, United Kingdom; Doetinchem, The Netherlands; Barcelona, Spain; and Kaunas, Lithuania. These four cities differ in size and urbanisation level (see Table 1
), but also in climate and societal characteristics (e.g., the employment rate for women, country’s prosperity level). By studying the relationship between neighbourhood green space and health for different subpopulations in these four cities, we wanted to explore the robustness of the associations and whether we can identify circumstances in which green space is associated with the health of different population groups. We thus investigated whether the association between neighbourhood green space and general and mental health differed by the age, gender, educational level, and employment status of residents and if this diversity was consistent across the four European cities.
2. Methods and Materials
Data were collected as part of the PHENOTYPE study, an EU-funded study that examined the health effects of the natural environment and its underlying mechanisms. A survey was carried out from May to October 2013 in four cities. The overall PHENOTYPE project design is reported elsewhere [15
All subjects gave their informed consent for inclusion before they participated in the study. Ethical approvals were obtained from each of the relevant bodies: Clinical Research Ethics Committee of the Municipal Health Care (CEIC PS-MAR), Barcelona, Spain (2012/4978/I); Staffordshire University Faculty of Health Science ethics committee, United Kingdom; Medical Ethical Committee of the University Medical Centre Utrecht, Netherlands; Lithuanian Bioethics Committee, Lithuania (2012-04-30 Nr.6B-12-147).
3. Study Population and Data Collection
In each city, approximately 30 neighbourhoods varying in the amount of neighbourhood green space and SES were selected (see Table 1
for a description of the neighbourhoods). Neighbourhood green space was defined using Urban Atlas. We extracted the land cover categories from Urban Atlas that seemed most relevant in relation to the aims of our study (i.e., green urban areas; agricultural + semi-natural areas + wetlands; forests; wetlands; water bodies). For Doetinchem, a Dutch database (Top10NL) was used. For neighbourhood SES, country-specific data were used to divide the neighbourhoods into SES categories. See Supplement A
for further details on the selection of neighbourhoods based on the Urban Atlas and Top10NL green data and SES data. Survey data were collected using face-to-face interviews, with the exception of Lithuania, where data were collected using a postal questionnaire. Per city, approximately 1000 adults, aged 18–75 years, were interviewed (n = 3947, overall response rate 20%). For further details on the data collection, see Supplementary Materials Supplement A
Additionally, in each neighbourhood, an audit was carried out to assess the amount and quality of the green spaces and of other characteristics of the neighbourhood. For each neighbourhood, a fixed sample of streets was selected, ensuring that unique, but important, features of the neighbourhood were included. The streets were combined into a route that was systematically inspected by two trained auditors (in a small number of cases, by one auditor). The auditors received a short training and a fieldwork training guide, providing them with specific instructions and example photographs per item. The training also included a shared observation of a neighbourhood where discrepancies between assessments were discussed and resolved before the start of the actual audit. During the audit, the auditors filled out a form containing closed questions about, for instance, the number of trees, the architectural character, maintenance of the area, or the general impression of the neighbourhood. With this approach, we attempted to obtain more objective information about the amount and quality of neighbourhood green. Audit data are considered more objective than the perceptions of residents, but are not as objective as registry or land-use data, since they remain an assessment.
For the purposes of our study, we selected respondents with complete data for the indicators of interest, providing us with a sample of 3771 respondents in 124 neighbourhoods (96% of the study population).
6.1. Does the Association between Neighbourhood Green Space and Health Differ by Educational Level?
In Barcelona, with the exception of the audit quality of green space, the association between neighbourhood-level green space and general health differed significantly between low and high-educated groups (Table 3
). Low-educated residents reported good general health significantly more often when the amount or quality of neighbourhood-level green was higher. Mental health showed similar patterns, albeit statistically significant differences between the low- and high-educated residents only occurred for the audit amount of neighbourhood green.
In Doetinchem, there were significant differences between the intermediate- and high-educated residents with regards to the association between the perceived quality of neighbourhood green space and general and mental health. Intermediate-educated residents tended to report better mental and general health when the perceived quality of green space was better, whereas the high-educated residents tended to report poorer health, though neither of these associations were significant. There were no differences between the educational groups in the relation between the amount of green space (neither audit nor perceived) and health.
In Kaunas, there were significant differences between educational groups in terms of the association between the perceived amount of neighbourhood green and general health. Low-educated residents reported having poor general health significantly more often when the perceived amount of neighbourhood green space was higher, while no association was found for the other education groups. In the case of mental health and audit amount of green, there were significant differences between the intermediate- and high-educated residents, in line with Doetinchem and Barcelona. Intermediate-educated residents tended to report better mental health when neighbourhoods were greener, while high-educated residents tended to report poorer mental health (associations not significant).
In Stoke-on-Trent, the relation between the audit quality of green and general health differed significantly between low- and intermediate-educated residents, with the low-educated residents reporting poor general health significantly more often when living in neighbourhoods with a better quality of green space. No differences were found between the educational groups for the other green measures or for mental health.
6.2. Does the Association between Neighbourhood Green Space and Health Differ by Age?
In Barcelona and Kaunas, the associations between green space and health differed by age. In Kaunas, the impact of living in greener neighbourhoods (perceived) on general health differed significantly between the age groups in such a way that residents under the age of 65 reported poor
general health significantly more often in greener neighbourhoods (Table 4
). There were also significant differences between the age groups in the relation between green space (audit quality and perceived amount) and mental health, again with residents under the age of 65 tending to report poor health more often, albeit these associations were not significant.
In Barcelona, there was a significant positive association between the perceived amount of neighbourhood green space and general health among the elderly, which was not found in the younger population. No significant differences by age were found for the audit neighbourhood green indicators or for mental health.
In the other two cities, no differences were found in the relation between neighbourhood-level green space and health between the elderly and the younger age group.
6.3. Does the Association between Neighbourhood Green Space and Health Differ by Employment?
Only in Barcelona and only in the case of the audit amount of green space was the hypothesis supported, with a positive association between neighbourhood greenness and mental health among residents who were not employed full-time, and no association among residents with a fulltime job (Table 5
). No differences by employment status were found for general health.
In Kaunas, the pattern was reversed, e.g., residents with a full-time job tended to report better mental and general health when the quality of (perceived) green space was better, while those who were not employed in a full-time job tended to report poorer health when the quality of neighbourhood green increased, though neither of these associations were significant.
In the two other cities, there were no differences in the association between neighbourhood green space and health by employment status.
6.4. Does the Association between Neighbourhood Green Space and Health Differ by Gender?
There were no significant differences in the association between neighbourhood green space on the mental or general health of men or women in the four cities (data not shown).