To investigate associations between neighbourhood greenspace and weight status, and to explore the contribution of physical activity to these associations.
Cross-sectional observational study over two time-periods.
Participants were adults (aged 18 years+) in from a nationally representative sample of the English population for the time periods 2000–2003 (n=42 177) and 2004–2007 (n=36 959).
Weight status was defined as body mass index (BMI) category according to WHO classification. Neighbourhood greenspace was measured using the Generalised Land use Database for England that defines greenspace as parks, open spaces and agricultural land, excluding domestic gardens. Multinomial logistic regression models were used to estimate associations between neighbourhood greenspace and BMI and, in eligible sub-samples, to investigate the contribution of total physical activity to these. All models were adjusted for age, sex, social class, economic activity, neighbourhood income deprivation and urban/rural status.
In 2000–2003 there was a counterintuitive association between greenspace and BMI. Residence in the greenest areas was significantly associated with increases in overweight (12%) and obesity (23%). In 2004–2007, there was a small protective effect of greenspace for those living in the greenest areas, but this was not statistically significant. Markers of total physical activity did not attenuate associations. Tests for interactions with urban/rural status confirmed that significant associations between neighbourhood greenspace and obesity were only present in urban areas in 2000–2003.
Better evidence for the utility of greenspace in the prevention of weight gain is required before greenspace interventions are developed.
Research investigating the determinants of weight status has recently expanded to include an ecological approach in order to help explain the recent rapid increases in the population prevalence of overweight and obesity.1 This ecological approach focuses on elements of the social, built and natural environments in which people live as risk factors for weight gain. One hypothesised pathway by which broader features of the natural environment may operate on weight status is by varying individual exposure to green environments that are supportive or unsupportive of physical activity. Recent research has suggested that greenspaces, such as parks, woodland, playing fields and other open public spaces may be important sites of formal- and leisure-time physical activity, such as formal sports participation, and recreational walking and cycling.2, 3 This may be particularly true in urban areas where access to greenspace may be more limited compared with rural and semi-rural areas.4, 5
Two recent international reviews have found that, in general, proximity to parks and recreation settings increase physical activity,6 particularly walking.7 For example, in Australia, proximity to greenspace and parkland has been found to be positively associated with walking3 and meeting recommended levels of physical activity.8 However, in the United Kingdom, the evidence base is more equivocal with no clear relationships between greenspace and adult leisure-time physical activity.9 Conflicting findings have emerged when directly investigating relationships with overweight and obesity. A cross-national European study found that respondents in areas with higher levels of greenery had 40% less risk of being overweight or obese.10 In Denmark, Nielsen and Hansen11 found an inverse association between distance to greenspaces and obesity in a sample of adults. However, a similar study in the United Kingdom found a positive relationship between distance to total greenspace and overweight—poorer access to total greenspace was associated with a lower likelihood of being overweight or obese.12
Many of these studies have limitations that may explain the mixed, positive and negative findings found in the literature. Research has tended to only have been undertaken in urban rather than rural areas, has used a variety of differing measures of exposure to greenspace, and are not nationally representative population studies in that they only use data from one city or region. Studies investigating the relationship between greenspace and obesity have also tended not to investigate the contribution of physical activity behaviours in explaining observed associations. For example, it may be that reported null or counter-intuitive associations are due to physical activity being lower or higher in greener areas. In this paper, we investigate whether (i) local access to greenspace is associated with weight status and (ii) whether markers of total physical activity mediate associations between greenspace and weight status, using data from a large, nationally representative population survey in England over two separate time periods.
Materials and methods
Data used in this study were drawn from the publically available Health Survey for England (HSE), an annual nationally representative cross-sectional survey of individuals in England. In order to keep available neighbourhood greenspace variables contemporaneous with individual outcomes, data were pooled into two time periods, 2000–2003 and 2004–2007. Overall we analysed data on 79 136 adult men and women aged 18 years and over (n=42 177 in 2000–2003; n=36 959 in 2004–2007).
Body mass index (BMI; weight (kg)/height2 (m2)) was categorised using World Health Organization definitions13 of underweight (>18.5 kg m−2), normal (18.5–25 kg m−2), overweight (25–30 kg m−2) and obese (30 kg m−2+). Age, sex, social class (as measured by head of household occupational social class) and economic activity (employed, unemployed, retired or other inactive) were also obtained.
Physical activity was assessed as total self-reported physical activity undertaken as part of work, domestic labour, sport and leisure in the last 4 weeks. This summary measure was calculated by the UK National Centre for Social Research. The low, medium and high categories were derived as a combination of both the type of exercise undertaken in each domain and the relative intensity of the activities undertaken. For further details, see Sproston and Mindell.14
Neighbourhoods were characterised using Middle Super Output Areas (MSOAs), an administrative geographic unit. Within England, there are 6780 MSOAs with a mean population of 7247 (range 2153–15 331), and within our sample individuals resided in n=4100 MSOAs in 2000–2003; and n=4180 in 2004–2007. The mean number of HSE respondents per MSOA with complete data on BMI, age, sex, social class and economic activity was similar (n=16) in 2000–2003 and (n=15) 2004–2007.
To each eligible respondent, we attached a neighbourhood-level measure of greenspace, income deprivation and urban/rural status. Data on the quantity of local greenspace was drawn from the 2001 and 2005 Generalised Land Use Database (GLUD) for England.15, 16 Generalised Land Use Database defines greenspace as inclusive of parks, open spaces and agricultural land, but excludes domestic gardens. Areas of greenspace that are less than 5 m2 were not included in the data set, and data were precise to 10 m2. We calculated the percentage of each MSOA classified as greenspace, and divided them into five equal groups (0 to ⩽20%, >20 to ⩽40%, >40 to ⩽60%, >60 to ⩽80%, >80 to ⩽100%). For presentational purposes, these percent greenspace exposure groups are reported in the tables as 0–20, 20–40, 40–60, 60–80, 80–100. Data for 2001 were attached to the 2000–2003 time period; data for 2005 were attached to the 2004–2007 time period.
Deprivation was measured using the income deprivation domain of the 2004 and 2007 English Index of Multiple Deprivation (IMD). This measure is the best available measure of low-income, and represents the proportion of low-income households in each MSOA defined as in receipt of various forms of state income support, means-tested tax credits or asylum seeker support.17, 18 The income deprivation sub-domain was used instead of the global measure of deprivation (IMD), as the global measure included data on a number of area-wide health indicators. Data for 2004 were attached to the 2000–2003 time period and data for 2007 were attached to the 2004–2007 time period.
Urban/rural status was categorised using a national Rural-Urban Classification System developed by a UK government initiative.19 Using this, each respondent's MSOA was classified into three groups: urban; town and fringe; village, hamlet or isolated dwelling.
In order to investigate associations between greenspace and obesity, we analysed data using a multinomial logistic regression model. This allowed us to estimate relative risk ratios (RRRs) for adults who were overweight or obese, relative to those whose BMI was normal. RRRs were estimated and were judged to be statistically significant if 95% confidence intervals (CIs) did not span unity. CIs were adjusted for the clustering of respondents within MSOAs. Confounding variables were added to this basic model, and in this analysis we controlled for age, sex, social class, economic activity, neighbourhood income deprivation and urban/rural status in turn. We then investigated whether interactions existed between greenspace and age, sex, social class, economic activity, income deprivation and urban/rural status. This was done by first stratifying models by levels of the potential modifying variable, and secondly, by testing interaction terms in the regression models. Model fit was evaluated using differences in the Bayesian Information Criterion between nested models. Underweight respondents were excluded from final analyses (BMI<18.5 kg m−2) due to small numbers. All models were fitted using Stata 10.1 (StataCorp LP, College Station, TX, USA).
In order to analyse the relative contribution of total physical activity to the estimated relationship between greenspace and obesity, we fitted the model specified above in a further model for physical activity. This model tested whether total physical activity mediated the relationship between greenspace and weight status. Total physical activity variables were not available over all HSE survey years. Therefore, analyses incorporating total physical activity data were limited to 2002–2003. The estimated RRRs between greenspace and weight status were then compared before and after inclusion of total physical activity to establish whether the estimates of the relationship changed after inclusion in the model.
Table 1 summarises the key descriptive characteristics of the sample. Women are overrepresented, and respondents primarily resided in suburban (2000–2003) and urban areas (2004–2007). The average age of participants was 45.5 and 48.6 years across the two time periods. In 2000–2003, 59.7% of the sample was either overweight or obese, rising to 62.3% in 2004–2007. Respondents classified as obese were around 4 years older than the sample average. In 2000–2003 the proportion of all obese persons residing in the least green neighbourhood was the smallest (12.6%); in 2004–2007 it was the second smallest (16.8%).
Table 2 shows RRRs and 95% CI for independent relationships between greenspace, overweight and obesity, and other variables, after adjustment for all covariates. For 2000–2003, a counterintuitive association was observed, with an increase in the relative risk of overweight and obesity as neighbourhoods became greener (test for trend: overweight P=0.024; obese P=0.000). When compared with the ‘normal’ BMI range, there was a significant increase in the relative risk of being overweight and obese in the greenest compared with the least green neighbourhoods (overweight: RRR=1.12, 95% CI: 1.03, 1.22; obese: RRR=1.23, 95% CI: 1.11, 1.37). In 2004–2007, there was a small decrease in the relative risk of overweight (RRR=0.95, 95% CI: 0.87, 1.04) and obesity (RRR=0.92, 95% CI: 0.82, 1.02) in the greenest areas. However, these associations did not reach statistical significance (except for the relative risk of obesity in areas with 60–80% greenspace). Residing in town and fringe areas in 2000–2003 was independently associated with a significant increased risk of overweight (RRR=1.11, 95% CI: 1.02, 1.20) and obesity (RRR=1.21, 95% CI: 1.10, 1.34). However, associations were inconsistent and not statistically significant in the 2004–2007 time period.
Stratification of models by age, sex, social class, economic activity and neighbourhood income deprivation indicated that these variables did not moderate relationships between greenspace and BMI. All interaction terms in these regression models did not reach significance at the 95% confidence level (P>0.05). However, stratification by urban/rural status indicated that an interaction plausibly existed with obesity (BMI 30 kg m−2+) and this was confirmed when interactions were formally tested (P<0.05). Figure 1 shows the stratified RRRs for obesity for each of the three categories of urban/rural status, and their 95% CIs. Exposure to greener areas in urban, and town and fringe locations was associated with an increased risk of obesity, whereas exposure to greener neighbourhoods in village and isolated hamlets was associated with a lower relative risk of obesity. CIs in pooled and stratified models showed that associations between greenspace and obesity were only statistically significant for those residing in urban areas.
In order to determine the contribution of total physical activity in the observed associations, we assessed the relative contribution of a marker of total physical activity in eligible sub-samples of the current data set after controlling for age, sex, social class, economic activity, neighbourhood income deprivation and urban/rural status. Table 3 shows the RRRs and 95% CIs for the adjusted relationship between greenspace, and overweight and obesity with and without additional adjustment for total physical activity. The data available for this analysis included 15 653 HSE respondents spanning 2002–2003. There was an increased risk of overweight and obesity in the most green, compared with the least green neighbourhoods (RRR=1.20, 95% CI: 1.06, 1.37; RRR=1.21, 95% CI: 1.02, 1.44). However, associations, although reporting increases in relative risks, were inconsistent across quintiles of greenspace and 95% CIs spanned unity. Associations between greenspace and weight status were only marginally attenuated in the greenest neighbourhoods after adjustment for total physical activity.
A recent systematic review of the greenspace and obesity literature20 highlighted that this is a relatively new field and evidence is only just emerging. This systematic review revealed that there are only 13 studies that directly investigate the links between greenspace and weight status and that the majority has been undertaken in the United States, with only one study undertaken outside of North America. Three of these studies have found positive relationships between greenspace and BMI and reduced weight-gain over time, particularly for children and young people.10, 21 However the remaining 10 studies found mixed or weak evidence for a relationship or no evidence of a relationship at all.20 In this large, nationally representative study of English adults with robust outcome and comprehensive exposure measures replicated over two time periods, we find mixed results. In 2000–2003, there was a counterintuitive association between greenspace and BMI, with residence in the greenest areas associated with a 12% increase in risk for overweight, and a 23% increase in risk for obesity. In 2004–2007, a small protective effect of greenspace was observed for those living in the greenest areas, with a (non-statistically significant at the 95% level) reduction in the risk of overweight (5%), and obesity (8%). The presence of significant interactions with urban/rural status confirms that the statistically significant counter-intuitive observed associations are only present in urban areas in 2000–2003.
We attempted to explain these relationships by hypothesising that associations may be mediated by relevant individual weight-related behaviours—it may be that residents of greener areas were less physically active. In many similar studies an assessment of the role of individual weight-related behaviours in the relationship between environment and obesity is not often undertaken. Here, after separately controlling for a measure of total physical activity, effect estimates were virtually unchanged. This suggests that the observed relationship between greenspace and weight status was not mediated by this behavioural measure and implies that the primary behavioural pathway (physical activity) by which total greenspace is hypothesised to impact on weight may be less important than previously thought. Such findings suggest that the selection pathways hypothesised in the literature, whereby more physically active people move to greener areas, are not supported by these data.
So what may explain the associations reported here? First, Coombes et al.,12 suggest that the type and quality of available greenspace may be a critical factor in determining behavioural pathways to weight status. In their study in Bristol, UK, they found that access to ‘formal’ greenspaces (areas characterised by an organised layout and structured path network, which are generally well maintained) was associated with physical activity and a reduced BMI. For other type of greenspace, there were mixed results, with greenspace positively and negatively associated with overweight and obesity. In our national study, we were unable to disaggregate our broad greenspace exposure measure to this level of detail and were, therefore, unable to assess how different types of greenspace may be associated with weight.
Second, we were unable to explore the role of non-exercise activity thermogenesis in our analyses.23 Non-exercise activity thermogenesis is the energy expenditure of all physical activities excluding volitional sporting-like exercise of which sedentary behaviours, particularly sitting (at work, in motorised transport, watching television and playing computer games) are a major component. Sedentary behaviour has been proposed as a major independent risk factor for societal weight gain24, 25, 26 and is sometimes conceptualised as the ‘third’ behaviour related to energy balance. In this study, sedentary behaviour may partly explain the observed associations if sample respondents who reside in greener areas are more sedentary than those in less green areas. Many studies investigating associations between greenspace and weight status, do not explore the role of sedentariness that may partly explain the inconsistent findings reported in many existing studies.
Several limitations of this study must be noted. First, our exposure measure could not be disaggregated to assess specific types and aesthetic quality of greenspace. This does not allow us to explore how different types of greenspace may impact on weight and weight-related behaviours differently. Second, our exposure measure only excludes greenspace less than 5 m2. Inevitably, this means that some very small areas of greenspace are still included in the data set; these small patches of greenery may serve no useful purpose for physical activity. A related issue is that information on the actual size and configuration of areas of greenspace in each neighbourhood was not available. Knowing the configuration and maximum size of greenspace in each neighbourhood would allow us to better model areas of greenspace that are most likely to be useful for physical activity. Finally, no data on respondent utilisation of greenspace were available so we were unable to assess who uses greenspace, and how frequently, in their local environment.
In conclusion, this study—the largest nationally representative study to date—finds no good evidence to suggest a protective effect of greenspace on overweight and obesity, and that markers of total physical activity do not mediate observed relationships. The strongest findings from the analyses presented here are counterintuitive, and suggest that the risk of being overweight and obese is highest in the greenest areas and in urban locations. However, failing to adequately take into account the type of greenspace and the role of sedentary behaviours in the analyses of these associations may explain the direction of the relationships observed here. Researchers and policy-makers alike have called for environmental interventions focused on improving access to, and use of, green and natural environments in population obesity prevention strategies.27, 28 However, better evidence for the utility of greenspace for the prevention of weight gain through weight-related behavioural pathways is required before large-scale investment in greenspace policies and programmes are undertaken.
Egger G, Swinburn B . An ecological approach to the obesity pandemic. Br med J 1997; 315: 477–480.
Macintyre S, Macdonald L, Ellaway A . Do poorer people have poorer access to local resources and facilities? The distribution of local resources by area deprivation in Glasgow, Scotland. Soc Sci Med 2008; 67: 900–914.
Giles-Corti B, Broomhall MH, Knuiman M, Collins C, Douglas K, Ng K et al. Increasing walking—How important is distance to, attractiveness, and size of public open space? Am J Prev Med 2005; 28: 169–176.
Maas J, Verheij RA, Groenewegen PP, de Vries S, Spreeuwenberg P . Green space, urbanity, and health: how strong is the relation? J Epidemiol Community Health 2006; 60: 587–592.
Maas J, Verheij RA, Spreeuwenberg P, Groenewegen PP . Physical activity as a possible mechanism behind the relationship between green space and health: a multilevel analysis. BMC Public Health 2008; 8: 206.
Kaczynski AT, Henderson KA . Environmental correlates of physical activity: a review of evidence about parks and recreation. Leisure Sci 2007; 29: 315–354.
Kaczynski AT, Potwarka LR, Saelens BE . Association of park size, distance, and features with physical activity in neighborhood parks. Am J Public Health 2008; 98: 1451–1456.
Duncan M, Mummery K . Psychosocial and environmental factors associated with physical activity among city dwellers in regional Queensland. Prev Med 2005; 40: 363–372.
Hillsdon M, Panter J, Foster C, Jones A . The relationship between access and quality of urban green space with population physical activity. Public Health 2006; 120: 1127–1132.
Ellaway A, Macintyre S, Bonnefoy X . Graffiti, greenery, and obesity in adults: secondary analysis of European cross sectional survey. Br Med J 2005; 331: 611–612.
Nielsen TS, Hansen KB . Do green areas affect health? Results from a Danish survey on the use of green areas and health indicators. Health Place 2007; 13: 839–850.
Coombes E, Jones AP, Hillsdon M . The relationship of physical activity and overweight to objectively measured green space accessibility and use. Soc Sci Med 2010; 70: 816–822.
World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Report of a WHO Consultation. Geneva, 2000.
Sproston D, Mindell J . Health Survey for England 2004. The Health of Minority Ethnic Groups. The Information Centre: London, 2006.
Office of the Deputy Prime Minister. Generalised Landuse Database for England 2001. Office of the Deputy Prime Minister: London, 2005.
Department for Communities and Local Government. Generalised Landuse Database Statistics for England 2005. London, 2007.
Noble M, Wright G, Dibben C, Smith GAN, McLennan D, Anttila C et al. Indices of Deprivation 2004. Office of the Deputy Prime Minister, Neighbourhood Renewal Unit: London, 2004.
Noble M, McLennan D, Wilkinson K, Whitworth A, Barnes H, Dibben C . The English Indices of Deprivation 2007. Department for Communities and Local Government: London, 2007.
Countryside Agency. Rural and Urban Area Classification 2004. An Introductory Guide. Countryside Agency: Cheltenham, 2004.
Lachowycz K, Jones A . Greenspace and obesity: a systematic review of the evidence. Obes Rev 2011; 12: 183–189.
Liu GC, Wilson JS, Qi R, Ying J . Green neighborhoods, food retail and childhood overweight: differences by population density. Am J Health Promot 2007; 21: 317–325.
Bell JF, Wilson JS, Liu GC . Neighborhood greenness and 2-year changes in body mass index of children and youth. Am J Prev Med 2008; 35: 547–553.
Levine JA . Non-exercise activity thermogenesis (NEAT). Best Pract Res Cl En 2002; 16: 679–702.
Levine JA, Weg MWV, Hill JO, Klesges RC . Non-exercise activity thermogenesis—the crouching tiger hidden dragon of societal weight gain. Arterioscl Throm Vas 2006; 26: 729–736.
Owen N, Ekelund U, Hamilton M, Gardiner P, Dunstan DW . Sedentary behavior in adults: longitudinal, experimental, and intervention evidence. J Phys Act Health 2010; 7: S334–S336.
Owen N, Mitas J, Sarmiento OL, Reis R, Sugiyama T . Identifying the built environment determinants of physical activity and sedentary behavior: emerging international evidence. J Phys Act Health 2010; 7: S364–S366.
Faculty of Public Health Medicine. Great outdoors: how our natural health service uses greenspace to improve wellbeing: an action report. London, 2010.
Marmot M. . Fair society, healthy lives. Strategic review of health inequalities in England Post 2010. The Marmot Review: London, 2010.
This work was supported by the award of a Philip Leverhulme Prize to Professor Steven Cummins. He is also supported by a UK National Institute of Health Research Senior Fellowship. The views expressed in this publication are those of the authors and not necessarily those of the UK National Health Service, the UK National Institute of Health Research or the UK Department of Health.
The authors declare no conflict of interest.
About this article
Cite this article
Cummins, S., Fagg, J. Does greener mean thinner? Associations between neighbourhood greenspace and weight status among adults in England. Int J Obes 36, 1108–1113 (2012). https://doi.org/10.1038/ijo.2011.195
- physical activity
Establishing associations between residential greenness and markers of adiposity among middle-aged and older Chinese adults through multilevel structural equation models
International Journal of Hygiene and Environmental Health (2020)
Associations of neighborhood socioeconomic conditions with self-rated health, mental distress, and health behaviors: A nationwide cross-sectional study in Japan
Preventive Medicine Reports (2020)
Science of The Total Environment (2020)
Frontiers in Public Health (2020)
BMC Public Health (2020)