Greenspace and obesity: a systematic review of the evidence
K Lachowycz, School of Environmental Sciences, University of East Anglia, Norwich, Norfolk, NR4 7TJ, UK. E-mail: firstname.lastname@example.org
Greenspace is theoretically a valuable resource for physical activity and hence has potential to contribute to reducing obesity and improving health. This paper reports on a systematic review of quantitative research examining the association between objectively measured access to greenspace and (i) Physical activity, (ii) Weight status and (iii) Health conditions related to elevated weight. Literature searches were conducted in SCOPUS, Medline, Embase and PYSCHINFO. Sixty studies met the inclusion criteria and were assessed for methodological quality and strength of the evidence. The majority (68%) of papers found a positive or weak association between greenspace and obesity-related health indicators, but findings were inconsistent and mixed across studies. Several studies found the relationship varied by factors such as age, socioeconomic status and greenspace measure. Developing a theoretical framework which considers the correlates and interactions between different types of greenspace and health would help study design and interpretation of reported findings, as would improvement in quality and consistency of greenspace access measures. Key areas for future research include investigating if and how people actually use greenspace and improving understanding of the mechanisms through which greenspace can improve health and, in particular, if physical activity is one such mechanism.
The rise in obesity is well documented (1) and research has recently expanded from a focus on individual determinants of obesity to investigating upstream influences, including how the environment in which people live influences their lifestyle and weight gain. Such socio-ecological approaches consider how individuals interact with their environments. One potentially important factor in a person's living environment is their access to greenspace, as greenspace is theoretically a valuable resource for physical activity (2) and hence could contribute to reducing obesity and improving health.
Recent socio-ecological model based reviews identified greenspace as one of a range of potential environmental determinants of obesity (3–7) and physical activity (8–13). They say little specifically about greenspace but conclude that environmental factors have potential to influence bodyweight, although findings are mixed and associations complex, particularly given inconsistencies in methodological approaches. One systematic review evaluated empirical evidence regarding the association between parks and recreation settings and physical activity (14). However, while there has been a recent proliferation of research in this field and reviews commissioned by Government departments and charitable organizations in the UK (15,16) and elsewhere (17,18), no systematic review has been published in a peer-reviewed journal which specifically evaluates the evidence for a relationship between greenspace and obesity. This paper addresses this by carrying out a systematic literature review of available studies which investigate the relationship between access to greenspace, obesity, and obesity-related health outcomes and behaviours.
The review focuses on three groups of health markers in relation to greenspace access in the home environment (i) Indicators of physical activity (ii) Weight status and (iii) Health conditions known to be related to elevated weight status. Home environment is defined as the geographic area surrounding the place of residence.
A literature search using four electronic databases (SCOPUS, Medline, Embase, PYSCHINFO) was conducted in February 2010. It was limited to peer-reviewed journal articles published between 2000 and 2010, representing a phase of research characterized by a focus on environmental determinants of health (19) and development of objective measures of living environments (20).
Search terms were based on greenspace definitions from planning and health research. Key relevant environmental terms such as ‘walkability’ were also included to identify papers where greenspace was analysed but not reported in abstract findings. The health search terms covered physical activity (e.g. exercise), weight status (e.g. body mass index [BMI]) and related health outcomes (e.g. diabetes). A full description of terms and search strategy is available in Table S1.
Resulting papers were screened against inclusion criteria outlined in Table 1. The primary author reviewed results of the initial search and selected potentially relevant papers from paper titles. A second stage reviewed abstracts and then full papers to select papers which met the inclusion criteria. A random selection of 20% of papers was screened by another author (A. J.) to confirm they were correctly selected.
Table 1. Inclusion criteria used to select studies
|1||The paper used empirical data to report analysis of obesity-related lifestyle and health outcomes in relation to access to greenspace.|
|2||The greenspace access measure was generated using objective methods, either by use of a Geographic Information System or an assessment by trained auditors using a consistent tool.|
|3||Greenspace access was assigned based on location of residence, e.g. (i) Distance to nearest greenspace or count of greenspaces within a certain distance and (ii) Amount of greenspace in the area. Experimental studies which looked at interactions with nature or simulated views of nature were not included.|
|4||Greenspace access was included as a separate variable within the analysis and results were reported specifically for greenspace, even if this was not the primary aim of the study. This excluded studies which only included greenspace as a potential confounder or as one variable in a composite environmental score.|
|5||One or more of the outcomes measure used in the study was an indicator of physical activity, weight status or health outcomes shown to be related to obesity. The outcome measure could be either self-reported or objectively measured.|
|6||There was adequate consideration of and adjustment for confounding factors, defined as including (where appropriate) adjustment for age, sex and a marker of socioeconomic status at a minimum.|
Methodological quality assessment
All included studies were assessed for methodological quality by the primary author and an independent reviewer using a 10-item scale (Table S2). Levels of agreement between reviewers were analysed using Cohen's Kappa for multiple raters, with agreement assessed on a dichotomous scale (‘Positive’ vs. ‘negative’ and ‘insufficiently described’[N]). In the case of disagreement, consensus was reached by discussion. There was no a priori reason for weighting the scores, so studies scored one point for each item and points were summed between 0–10. Studies were classified as high quality if they obtained a score of six or more.
Strength of the evidence
A formal meta-analysis approach was judged inappropriate because of heterogeneity of the greenspace access measures and outcomes. Studies were thus summarised according to greenspace and health measures, confounding factors, findings, and effect sizes (See Table S2), with the terms ‘association’ and ‘relationship’ used to describe a statistical, rather than necessarily causal, relationship. Each study was assigned by the primary author and independent reviewer to one of four levels describing the relationship between greenspace and health (i) Positive; (ii) Equivocal (weak/mixed); (iii) No relationship and (iv) Negative, with ‘positive’ defined as health promoting (e.g. increased walking) and ‘negative’ defined as health demoting (e.g. increased BMI). When summarizing findings, papers reporting results from the same study were covered individually.
The database search produced 2473 hits in SCOPUS and 601 in the Ovid databases. Screening by the primary author identified 219 papers in SCOPUS and 118 in Ovid databases of potential relevance (including duplicates across databases). Review of these papers against inclusion criteria produced a final list of 60 papers. A summary, ordered by location, is available (Table S3).
Papers failed the inclusion criteria for the following reasons: Not statistical analysis of obesity-related health markers in relation to greenspace (132 papers), greenspace not objectively measured or not based on residential location (63), results not specifically presented for greenspace (74), health marker not related to obesity (6) or insufficient adjustment for confounders (2).
Table 2 gives a count of papers according to strength of evidence and grouped by health outcome and study age-group.
Table 2. Count of papers by strength of the relationship between greenspace and obesity-related health indicators, by outcome measured and population age group
|Physical activity|| || || || |
| Objective||3 (3C)||2 (1C, 1A)||4 (3C, 1A)||0|
| Subjective||17 (3C, 10A, 4O)||11 (2C, 8A, 1O)||11 (2C, 9A)||2 (2A)|
|Weight status|| || || || |
| Objective||2 (2C)||3 (2C 1A)||2 (1C, 1A)||0|
| Subjective||1 (1A)||3 (1C, 2A)||2 (2A)||0|
|Obesity-related health outcomes||2 (2A)||1 (1C)||0||0|
Greenspace access measures
Studies were heterogeneous in the approaches and measures used. The most common measure was distance to nearest greenspace or count within a certain distance of home (27 studies), using either straight-line/Euclidean distances (13), network distances (14) or both (5). A further 15 studies calculated the percentage of greenspace within a certain distance or area. Two used an audit of greenness by trained assessors (21,22) and one derived scores of ‘recreational value’ for different greenspace types (23). Fifteen studies used multiple measures or more sophisticated approaches, including measures based on gravity models (24–27), quality of greenspace (24,27), type of greenspace (27–29), facilities available (30–32) and park service areas (33). A few studies focussed on greenspaces above a particular size (30,34–36) while one removed large parks (33).
Methodological quality assessment
There was 89.2% agreement on the 600 items scored during the quality assessment (kappa statistic 0.78, P < 001; substantial agreement) and full consensus was reached after discussion. Overall, 20 papers (33.3%) were rated as high methodological quality. The items where the majority of studies were judged negatively were: potential inclusion bias (77% of papers), use of subjective outcome measure (70%), no consideration of type or quality of greenspace (73%), no measure of greenspace use (83%) and testing of multiple variables (72%).
Greenspace and physical activity
The search identified 50 studies examining the relationship between greenspace and physical activity. The majority (41) used self-reported measures, nine used accelerometers. These studies were conducted in USA (28), England (6), Australia (7), the Netherlands (4), Canada (2), New Zealand (1), Portugal (1), Sweden (1) and Europe-wide (1).
Twenty studies (40%) reported a positive association between greenspace and physical activity. They included six among children/teenagers (21,29,37–41), within which there was some evidence of interactions with sex (37) and ethnicity (37,41). Fourteen studies reported evidence of a relationship among adults (22–24,26,30,32,42–45), including four looking at older people living in Oregon (46–49). There were 28 studies which found no evidence of a relationship (15 studies) or results were weak or mixed (13). Two studies found negative relationships (50,51) and some negative findings were found in two studies for those with access to high quality large greenspaces (27) and in sunnier weather (36).
Several studies examined how relationships might vary with the measure of greenspace access. Two Australian studies found no relationship between physical activity and parks, but found an association with distance to coastal environments (52,53). Research in Perth found that accessibly of public open space was not associated with overall activity, but those with very good access to attractive, large spaces were more likely to achieve high levels of walking (24). Jones et al.'s study in Bristol measured greenspace type (formal, sports, natural, etc.) but reported no significant relationships with physical activity (28). Cohen et al. found that particular park amenities, e.g. shaded areas, were associated with higher activity (29). Two studies (38,42) used both counts of greenspaces and percentage area within various distances and found the number of greenspaces within a certain distance was more important than size in relation to physical activity (38).
Six studies measured the relationship between greenspace access and utilization. Cohen et al. found living within 1 mile of a park was positively associated with park use and frequency of leisure exercise (32). Three studies (24,28,45) found that residents living closer to parks visited them more frequently and higher utilization was associated with higher activity levels; however the direct relationship between park access and physical activity was statistically insignificant (28,45) or significant only for those with access to attractive and large spaces (24). Mowen et al.'s analysis among older adults found that park visitation frequency mediated the relationship between proximity and daily physical activity (54). A study in Baltimore (55) found no association between park access and use of parks for physical activity but a marginally significant association between access and total physical activity. Five studies surveyed if physical activity actually took place in the local neighbourhood or in greenspace. There was evidence of an association between access to greenspace and activity in the local neighbourhood (30,47–49,56) but Canadian research found mixed evidence for a relationship between access to parks and activity undertaken within them (30,42).
Overall, the evidence for an association between access to greenspace and physical activity is mixed. The majority of studies (66%) found some evidence of a positive association, although only 40% found an association that appeared unambiguous.
Greenspace and weight status
Thirteen studies investigated the relationship with weight status, all using BMI as the marker, with seven using self-/parent-reported BMI and six using objective measures. Studies were from USA (10), Canada (2) and Europe-wide (1).
Three studies (23%) reported a positive (i.e. reduced BMI) relationship between greenspace and BMI. Liu et al. found that increased vegetation was associated with reduced weight among young people living in high population densities (57) while Bell et al. reported increased greenspace was associated with less weight gain over 2 years (58). Across eight European cities, people were 40% less likely to be obese in the greenest areas (22).
Six studies found mixed or weak evidence of a relationship between greenspace and BMI, and four found none. A study of adults living in Seattle (59) examined both access to communal greenspace and vegetation indexes derived from remote sensing (Normalized Difference vegetation Index [NDVI]), finding a negative relationship between access to greenspaces and BMI in low NDVI (low amounts of natural vegetation) areas, and a slight positive relationship in high NDVI areas. Several studies found slight evidence of a relationship between greenspace and BMI, which was either marginally significant (60), heavily attenuated by adjustment for socioeconomic status (33) or only in some ethnic groups (61). There was also variation by greenspace type, with relationships found only for access to beaches in New Zealand (62) and park playgrounds among children in Canada (31).
Overall, the majority of studies found some evidence of a relationship with BMI, or report mixed results across subgroups and according to the greenspace measure used.
Greenspace and obesity-related health outcomes
Just three studies examined the association between greenspace and markers of obesity-related health outcomes. Maas et al.'s study in the Netherlands found a lower prevalence of diseases in areas with more greenspace, including coronary heart disease and diabetes (63). An England-wide study found an association between greenspace exposure and lower premature mortality from circulatory disease (64). A study of adolescents in Minnesota measured metabolic syndrome scores (MetS), a cluster of risk factors associated with cardiovascular disease and diabetes, finding lower scores in greener areas although this result was marginally significant (65).
Nineteen studies presented results as odds ratios of the binary health marker, mostly using least access to greenspace as the reference group. A European-wide study calculated that adults in the highest quintile of greenery were more than three times likely to report they were physically active (OR 3.32, 2.46–4.50) compared with those in least green areas (22). Most studies had more modest estimates of effect. Some reported different effect sizes for subgroups, e.g. boys aged 5–18 in Atlanta (37) were 2.3 (1.7–3.2) times more likely to have walked recently if they had access to at least one greenspace, whereas the odds ratios for girls was 1.7 (1.2–2.4).
This is a relatively new field of research and only 60 papers were identified by this review, of which almost half (28) were published in the last 2 years (2008 and 2009). Around two-thirds (33 out of 50 papers) found a positive relationship or some weak or mixed evidence of an association between greenspace and physical activity, 9 out of 13 reported a positive or equivocal relationship with BMI and three papers found some association with obesity-related health outcomes. However, around a third of studies found no relationship, two found a negative relationship and results were equivocal across many papers.
Given the large-range factors which affect weight status and potential time-lags between exposure and change in bodyweight, the lack of a strong association with weight outcomes found in these cross-sectional studies is unsurprising. Several studies found evidence that relationships varied by factors such as age and socioeconomic status and also by the measure of greenspace used. Improvement in the theoretical understanding of the mechanisms through which greenspace may influence health would help study design and interpretation of reported findings. Advances could include identifying which factors within the social-ecological model of health are specifically important for the relationship between greenspace and obesity. In other words, when, how and for whom is access to greenspace associated with obesity? A recent review of recreation settings and physical activity (14) also notes the need for more specific models and calls for improved measures of greenspace. It is noteworthy that most reviewed studies used crude measures of greenspace, with no consideration of quality or other environmental features. More sophisticated approaches are needed (20), particularly as several studies showed size and attractiveness to be associated with utilization frequency.
All studies were cross-sectional and therefore suffer from widely acknowledged methodological limitations. Most importantly, it is not possible to determine if an observed relationship between greenspace and health is causal. There is the possibility of selection effects where more active people choose to live in greener environments (66). The studies also varied hugely in choice of confounding variables and therefore some positive results could be because of residual confounding. Particularly problematic may be inadequate adjustment for socioeconomic factors given the well documented association between deprivation and obesity (67). Furthermore, greenspace was just one of many exposures being tested in several studies so statistically significant findings were more likely to arise because of multiple tests. The majority of studies (44 out of 60) relied on self-reported physical activity or BMI, which is prone to recall bias. Nevertheless, this did not obviously lead to a bias in results. Few studies measured actual use of greenspace. The employment of new technologies such as global positioning systems to record where people are active will help address this.
This review has a number of strengths and limitations. Weaknesses include that the search was restricted to English-language articles and just four databases were searched, although these were judged to best capture relevant studies. The search focussed on peer-reviewed literature but relevant studies may be reported elsewhere. However, limiting inclusion to peer-reviewed studies ensured a high quality of papers. Several papers were based on related populations and these were counted individually within the summary, which may overestimate counts of particular findings. Strengths include the wide set of search terms used and assessment of study quality.
There is some evidence for an association between greenspace and obesity-related health indicators, but findings were inconsistent and mixed across the studies. Developing a more solid theoretical socio-ecological framework which considers the various correlates and interactions between different types of greenspace and health would help both formulation and interpretation of the body of research.
Conflict of Interest Statement
No conflict of interest was declared.
Thanks to Emma Coombes, Flo Harrison, Cam Lugton, Esther Rind and Pauline Vissers for help reviewing the studies. K. L. is a PhD Candidate funded by a joint MRC/ESRC Interdisciplinary Studentship. The work was supported by the Centre for Diet and Activity Research, a UK Clinical Research Collaboration Public Health Research Centre of Excellence. Funding from the British Heart Foundation, Department of Health, Economic and Social Research Council, Medical Research Council, and the Wellcome Trust, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged.