Comparing circular and network buffers to examine the influence of land use on walking for leisure and errands
© Oliver et al; licensee BioMed Central Ltd. 2007
Received: 31 May 2007
Accepted: 20 September 2007
Published: 20 September 2007
There is increasing interest in examining the influence of the built environment on physical activity. High-resolution data in a geographic information system is increasingly being used to measure salient aspects of the built environment and studies often use circular or road network buffers to measure land use around an individual's home address. However, little research has examined the extent to which the selection of circular or road network buffers influences the results of analysis.
The objective of this study is to examine the influence of land use type (residential, commercial, recreational and park land and institutional land) on 'walking for leisure' and 'walking for errands' using 1 km circular and line-based road network buffers. Data on individual walking patterns is obtained from a survey of 1311 respondents in greater Vancouver and respondent's postal code centroids were used to construct the individual buffers. Logistic regression was used for statistical analysis.
Using line-based road network buffers, increasing proportion of institutional land significantly reduced the odds of 'walking for leisure 15 minutes or less per day' no significant results were found for circular buffers. A greater proportion of residential land significantly increased the odds of 'walking for errands less than 1 hour per week' for line-based road network buffer while no significant results for circular buffers. An increased proportion of commercial land significantly decreased the odds of 'walking for errands less than 1 hour per week' for both circular and line-based road network buffers.
The selection of network or circular buffers has a considerable influence on the results of analysis. Land use characteristics generally show greater associations with walking using line-based road network buffers than circular buffers. These results show that researchers need to carefully consider the most appropriate buffer with which to calculate land use characteristics.
Rising obesity rates in Canada is an important public health problem because of co-morbidities such as heart disease, diabetes and cancer . Increasing rates of obesity are linked to a changing social and physical environment rather than genetic factors  and research has begun to investigate the role of the built environment in influencing obesity by promoting or inhibiting physical activity [3–5]. Walking is important from a public health perspective because it is the most common physical activity among Canadians . While Canada's Physical Activity Guide  recommends Canadians accumulate at least 30 minutes of moderate physical activity (e.g. walking) per day a national survey has demonstrated that only 49% of Canadians meet this requirement .
An increasing number of studies demonstrate that aspects of the built environment can both promote and discourage walking [4, 5, 8–11]. Common measures of the built environment include land use type, density (e.g. residential density), land use mix and street connectivity (e.g. intersections per km2) .
A recent study of US metropolitan areas has demonstrated that a greater diversity of businesses in a neighbourhood increases walking . Greater proximity to shopping centres is associated with increased walking in older adults . Studies examining the influence of park land and green space on walking have demonstrated mixed results. In a study of 56 neighbourhoods in Portland a positive relationship between green space and physical activity among older adults was found . A study of women in Melbourne, Australia demonstrated that public open space (e.g. parks) did not influence walking for leisure or transport among women . Giles-Corti et al  examined 1,802 adults in Perth, Australia and results showed that open space was positively related to walking for transport but not walking for recreation. Residential density has been positively associated with walking in several studies [14, 17]. Measures of the built environment using indices (e.g. land use mix) have been associated with walking in several studies [14, 18]. Frank et al  in a study of adults in Atlanta has found more walkable neighbourhoods, measured using land use mix, residential density and intersection density, were associated with increased physical activity. More walkable neighbourhoods have also been associated with increased walking among older adults .
Research examining the influence of the built environment on physical activity and health has been limited by two factors. First, there is a need for theories or conceptual models articulating how aspects of the built environment may influence physical activity or other health outcomes [20–23]. Specifying conceptual models is important to identify and to understand how salient aspects of the built environment influence walking . Walking is influenced by physical characteristics such as the variety of potential destinations as well as aesthetics of the built environment . Second, limitations in spatial epidemiology, geographic information systems (GIS) and geographic data have made it difficult to empirically test models and hypotheses. Until recently most studies have relied on perceived rather than objective measures of the built environment . However, recent advances have resulted in the ability to identify the spatial location of respondents and use high-resolution spatial data to objectively measure their local environments [3, 5, 8, 23]. High-resolution road network data, land use data, cadastral data and census data are increasingly available to measure aspects of the built environment that may influence walking. This use of objective measures of the built environment is important to better understand how land use may influence walking behaviours and also to identify specific dimensions that could be used by planners and policy makers to increase walking . Increasing availability of high-resolution spatial data combined with the computational power of GIS means that many options are available to measure aspects of the built environment that influence walking; however, there has been relatively little focus on developing and testing methodologies to measure the built environment.
To assess relations between the local built environment and physical activity it is necessary to define a spatial unit that best represents a respondent's local environment. The characteristics of this local environment, such as aesthetics and built environment factors are presumed to influence an individual's decision to engage in physical activity. There is a dearth of research examining the most appropriate method to define spatial units or assessing the sensitivity of results to the choice of spatial unit. Typically, one of three methods is used to define spatial units to measure aspects of the built environment. One method is to use pre-defined spatial units such as a census tracts or planning neighbourhoods to construct measures of the built environment [12, 23, 24]. A problem with using predefined areas is they do not necessarily correspond to areas an individual may walk. Built environment measures based on these units may have less error for individuals living in the centre of the unit than the edges potentially introducing bias into analysis. Given shortcomings with this technique, some studies have begun to use locational information (e.g. addresses, postal codes) to define unique areas for each individual. The most common method has been to establish a circular buffer around respondent's geocoded location at a given radius [19, 25–27]. While likely providing a more representative assessment of the built environment that may influence walking, a shortcoming is that a circle may not accurately represent the spatial area that influences walking. Circular buffers are likely to be inaccurate in areas with natural features such as rivers, lakes and cliffs or built features such as railways or suburbs with poor street connectivity. In such cases, areas within the buffer may be inaccessible by the respondent but still used to calculate built environment measures.
Because of the limitations of circular buffers, a few studies have used road network buffers to define areas within which an individual can walk as a base for measuring the built environment. Some studies have calculated distance to recreational facilities using ArcGIS network analyst to determine the shortest distance from the respondents' home by road [14, 28]. A polygon-based road network buffer has been used to define one-kilometre (km) areas around respondents' home locations . In this method, the endpoints of all possible journeys up to 1 km along the road network from the individual respondent are used to form the vertices of an irregular polygon that defines the traversable area within 1 km of the respondent's location. While this method may provide a more accurate assessment of the actual land area that influences walking, joining road vertices using straight lines may lead to inaccuracies in areas that do not have dense, regular grid street patterns.
The present study was designed to fill two gaps in this research area. The first goal of this paper is to present a new methodology to construct road-based network buffer in a geographic information system that may more accurately represent the 'walkable' neighbourhood of an individual and capture areas of an individual's local environment that may most directly influence walking. The second goal is to assess the extent to which the selection of circular or road-based network buffers affects the results of analysis examining the influence of land use on walking patterns. This paper is unique in that it specifically compares network and circular buffers and as such is an important addition to research examining the influence of the built environment on walking and physical activity.
The paper is divided into three remaining sections. The methods section discusses the individual data used, construction of circular and line-based road network buffers, measurement of the built environment and analysis methods. The results section presents the results of the analysis. The final section of the paper discusses the implications of the analysis and areas for future research.
Individual data was collected using a telephone survey in eight neighbourhood clusters in suburban municipalities (i.e. outside of the City of Vancouver) of the Greater Vancouver Regional District. The areas were created by clustering three to four census tracts, which are small geographic areas defined by Statistics Canada with a population between 2,500 and 8,000. The population of the neighbourhood clusters ranged from 11,000 to 18,000 (2001 Census of Canada). Areas were selected based on their median family income (2001 Census of Canada) and residential density (population per hectares of residential land). Residential density and median family income was determined for all Census Tracts in the Greater Vancouver Regional District. Residential density was selected because it is associated with walking and is correlated other dimensions of the built environment such as land use mix and connectivity . The aim was to select areas with equivalent income levels but differing residential densities. Four clusters with mid-high residential density (70 to 122 people per hectare of residential land) were selected, two of which with mid-high median family income ($53,000 – $77,000 CDN) and two of which with mid-low median family income ($32,000 – $44,000 CDN). Four clusters with mid-low residential density (29 to 61 people per hectare of residential land) were also selected, two of which with mid-high median family income and two of which with mid-low median family income.
A sampling frame of households for each neighbourhood was generated from the local telephone provider and numbers were de-duped to remove multiple (i.e. 2 lines per household), ineligible (e.g. fax, business) or invalid numbers. Random Digit Dialling (RDD) was used to select a household from the sampling frame and a minimum of five call-backs were made to reduce bias due to non-response. Interviews were conducted by experienced telephone interviewers using Computer Assisted Telephone Interviewing (CATI). The survey was piloted in January 2006 and the full survey was conducted over two-weeks in February 2006. Data was collected for 1935 adults and the survey achieved a response rate of 29% calculated as the percent of co-operative contacts divided by the number of contacts. In this current study we use data for respondents between 20 and 60 years of age. There were 1529 respondents in this age range and 218 were excluded due to missing variables or an invalid postal code resulting in an analytic sample of 1311.
The survey assessed the amount of time respondents spend walking both for errands and for leisure. These were selected as dependent variables because they may be associated with different land uses (e.g. walking for leisure may be associated with park land and walking for errands may be associated with commercial land) and as such are a useful way to examine associations with measures of the built environment. Walking for errands was assessed using the item "In a typical week in the past few months how many hours did you spend walking from home to grocery stores, banks, or to do other errands? none, less than 1 hour, from 1 to 5 hours, from 6 to 10 hours, from 11 to 20 hours, greater than 20". For analysis a dichotomous variable was constructed from the responses (less than one hour = 1, one hour or greater = 0). Walking for leisure was assessed used the item "On a typical day in the past 3 months, how much time did you spend walking for leisure? 0 minutes, 15 minutes or less, 16 – 30 minutes, 31 minutes to one hour, over an hour." The responses were dichotomized for analysis (15 minutes or less = 1, greater than 15 minutes = 0). Canada's Physical Activity Guide recommends adults engage in at least 30 minutes of physical activity (e.g. walking, house cleaning, weightlifting, running) to be accumulated in 10 minute intervals . The cut-off of 'greater than 15 minutes' was selected because individuals walking this amount would meet at least half of their daily physical activity requirements through walking for leisure.
Five demographic variables were included to control for potential confounding with the outcome variables. Gender (female = 1, male = 0) and age (continuous) were included in all models. Household income was based on respondent self-report and three categories were created: low income (less than $40,000 CDN) middle income ($40,000 to $80,000 CDN) and high income ($80,000 CDN and over). Dummy variables were created and middle income was the reference category. Respondent's marital status was categorized into single, married/common law and divorced/widowed. Dummy variables were constructed and married/common law was the reference category. Body Mass Index (BMI, weight (kg)/height (m)2) was calculated based on self-reported heights and weights. This variable was included because a higher BMI may be independently associated with lower levels of physical activity. The presence of a self-reported chronic condition (yes = 1, no = 0) was included because such conditions may limit respondents ability to engage in walking activities.
Creation of circular and line-based road network buffers
Four steps were involved in the creation of the circular and line-based road network buffers. First, survey respondents were geocoded using the Statistics Canada Postal Code Conversion File to assign a latitude and longitude co-ordinate to the centroid of respondent's self-reported six digit postal code. Postal codes were used a proxy for respondents actual street address location. A study has found 88% of Canadians in urban areas have a postal code within 200 metres of their true address location .
The second step was the creation of both circular and line-based road network buffers around each respondent's postal code centroid. A circular buffer of 1 km radius was constructed around the centroid of each survey respondent's postal code in ArcGIS 9 from Environmental Systems Research Institute (ESRI). A buffered line-based road network buffer was created using the British Columbia Road Network file, from GIS Innovations, which includes all roads in the study area. ArcGIS network analyst was used to calculate a 950 metre line-based buffer along the road network from each respondent's postal code centroid. A 50 metre simple buffer was then constructed around this line-based buffer, resulting in a 1000 metre buffer (950 m along the road plus 50 m away from the road) around each individual constrained to contiguous roadway. Only the portion of parcels that were within 50 m of the roadway were included in calculations. This may represent a better approximation of potential destinations locally accessible to the individual respondent.
A 50 metre buffer was chosen to ensure that parcels along the selected roads would be included but that most parcels located further from the road (e.g. behind those adjacent to the road) would not be selected. This method is based on the idea that land use encountered along roads is most important in characterizing a neighbourhood in the way it is experienced by residents walking through it, and land not accessible to the pedestrian, even if physically nearby, is not part of their 1 km walking neighbourhood. A 50 metre buffer of the road generally appears to be well-suited to constructing such a model and ensures that everything along the road is captured, that deeply extended parcels are not overrepresented and that inaccessible land, for example behind the houses along the road, is excluded. A 100 metre buffer was too large as it often included such inaccessible parcels and could bias the weight of parcels whose depth away from the road is disproportionate with their "shop front" profile experienced by walkers along the road. A 25 metre buffer was too narrow as it would sometimes miss properties set slightly back from the road along wide roadways or wide right-of-ways.
The third step involved intersecting the circular and line-based road network buffer for each respondent with high-resolution land use data to calculate the local land-use composition for each individual as a measure of the local built environment. Greater Vancouver Land Use Data, which assigns every parcel of land in our study area a detailed code indicating the specific use of the property (e.g. city hall, sports facility), was obtained. For this analysis a simplified layer of five land use categories was created from the more detailed land use codes . Recreational and park land includes parks, play grounds, fields, and trails/wooded areas. Residential land includes all private and rental dwellings such as high rises, low rises, garden/town homes, and single detached homes. Commercial land includes businesses with retail sales and services and professional offices. The category of mixed commercial and residential land (e.g. residential units above commercial properties) was divided between commercial and residential land uses. Institutional land includes public offices, hospitals, libraries, community centres, schools, city hall, and correction facilities. Industrial land includes factories, processing plant and industrial parks. An other land use category included water, open, undeveloped and agricultural land. The simplified land use layer was intersected with each respondent's circular and line-based road network buffer to create a layer including only land uses falling within the buffer area.
The total area (m2) of each of the five land use categories was calculated in ArcGIS for each respondent's circular and line-based road network buffer and then calculated as a proportion of the total area within each respondent's 1 km buffer. Land classified as 'other' was excluded from all further analyses. Proportions were chosen rather than an area measure of each land use category because the sizes of the line-based road network buffers were not equal among respondents. Even with the circular buffers, proportions were necessary as the total area of defined land use differs among respondents' buffers due to variations in the area of "empty" unclassified land occupied by streets and highways.
Logistic regression was used to examine the influence of the proportion of different land uses on the two outcome variables, walking for leisure and walking for errands. Parallel analyses were run using values for the proportion of land use categories as calculated by both the circular buffer and line-based road network buffer method, facilitating a comparison of the respective influence of each method of determining local land use composition for each respondent. A series of five models are presented beginning with a preliminary model (Model 1) that only includes individual characteristics. The following four models (Models 2–5) add each of the land use variables for both circular buffers (Model 2a-5a) and line-based road network buffers (Models 2b-5b). Model fit was assessed using the -2 Log Likelihood (-2LL) statistic, which indicates the amount of unexplained model variance. A lower -2LL statistic indicates a better model fit. The Likelihood Ratio Test (LRT) was used to determine if the addition of land use variables (i.e. Models 2ab-5ab) significantly improves the model fit compared to Model 1, which only includes individual variables. This LRT is calculated by subtracting the -2LL statistic for a model with land use characteristics from the preliminary model (e.g. Model 2a) and determining if this value exceeds the critical value of a chi-square distribution using a significance level of p = 0.05.
Descriptive statistics for individuals aged 20–60 living in the Greater Vancouver Regional District (n = 1311)
Variable (N = 1311)
Walk for leisure 15 minutes or less per day
Walk for errands less than one hour per week
Body Mass Index
Divorced or widowed
Less than $40,000
$40,000 – $80,000
More than $80,000
Comparison of circular and line-based road network buffers
Characteristics of 1 km circular and network land use buffers for study participants
Land use type
Recreational and park land
Other land uses
Logistic regression results
Logistic regression results for 'walking for leisure 15 minutes or less per day' are presented in Additional file 1. Model 1 includes only individual predictor variables with no land use characteristics. Being female significantly reduced the odds of 'walking for leisure 15 minutes or less per day'. Increasing BMI significantly increased the odds of 'walking for leisure 15 minutes or less per day' (OR 1.03; 95% CI 1.01,1.05). Income and marital status were not significant predictors of walking for leisure. Models 2 to 5 include individual and land use variables (added separately) for both the circular and line-based road network buffers. The -2LL is at the bottom of Additional file 1. Land use coefficients for Models 2, 3, 4 suggest recreational and park land, residential land and commercial land have no significant association with walking for leisure for both the circular and line-based road network buffers. The odds ratio for institutional land (OR 0.03, 95%CI 0.00, 0.33) was statistically significant for line-based road network buffers and indicates that increased institutional land reduces the odds of 'walking for leisure 15 minutes or less per day'. For institutional land (Model 5) the LRT indicates no significant difference between the individual model and circular model (Model 1 vs. Model 5a) but the line-based road network buffer significantly improves model fit (Model 1 vs. Model 5b).
Additional file 2 presents the model results for 'walking for errands less than one hour per week' and Model 1 presents the individual characteristics with no land use characteristics. Gender was not associated with walking for errands. Increasing BMI significantly increased the odds of walking less than one hour per week for errands and the magnitude of the odds ratios is similar to 'walking for leisure 15 minutes or less per day'. The odds of walking less than one hour per week for errands were significantly lower for individuals with low income (OR 0.65; 95% CI 0.49, 0.86) relative to middle income. Model 2 presents the results for recreational and park land for the circular and line-based road network buffers. For the line-based road network buffers increasing proportion of park land is associated with a decreased likelihood of walking for errands one hour or less per week. The LRT indicates that model fit was not improved by adding recreational land for the circular buffer but was significantly improved with the line-based road network buffer measure. Model 3 presents the results for residential land and shows that the proportion of residential land is not associated with walking for errands using the circular buffers but the line-based road network buffer shows increasing residential land significantly increases the likelihood that an individual will walk less than one hour per week for errands. Model 4 presents the results for commercial land and both buffers show that increasing commercial land significantly reduced the odds of walking less than one hour per week for leisure. The magnitude of the odds ratios were similar for both buffers (0.01 circular vs. 0.02 network). While the model fit, using the -2LL was significantly improved for both buffers compared to Model 1, the -2LL statistic was lower for the line-based road network buffer (1747.46 vs. 1755.77) suggesting that the line-based road network buffer may improve the model fit. Model 5 presents the results for institutional land use and results were only significant for the line-based road network buffers (OR 0.04; CI 95% 0.00, 0.35).
The principal goal of this paper was to present a methodology to create a line-based road network buffer that may better assess the land uses that influence walking. In this methodology a road network buffer that was 950 m long and 50 metres wide was constructed to assess aspects of land use that may influence walking. Previous methods to construct a network-based local neighbourhood have used journey endpoints (e.g. 1 km along the road network) as vertices from which to build a polygon delineating the neighbourhood. This method presents a vast improvement over the simple circular buffer method because it takes into account that the area of a "local neighbourhood" is necessarily restricted by the manner in which one is able to traverse the landscape – a pedestrian cannot travel "as the crow flies" but is instead forced to travel along existing roadways. However, the methodology currently presented even further improves the definition of a local neighbourhood by constraining the shape to more closely follow roads and thus exclude large inaccessible areas between diverging roads (see Figures 1 &2). This method may better represent the local neighbourhood of an individual as it would be experienced by that person as they walked through it. The advantage of the methodology presented is that it is less likely to be skewed by large features such as parks and industrial land as only land use within 50 metres of roads is selected. The methodology can be easily modified using GIS to meet the specific needs of other research projects. A shorter buffer length (e.g. 500 metres) may be more appropriate for studies with elderly populations or longer (e.g. 2 kms) for studies with younger populations. While we chose a buffer width of 50 metres, a wider or narrower may be more appropriate depending upon the built environment of the region under study.
The second purpose of this paper was to compare results of utilizing circular and line-based road network buffers to determine local neighbourhood land use composition, using walking for leisure and walking for errands as outcome variables. Greater association between land use and walking was found using the line-based road network buffers than the circular buffers suggesting that they may be better suited to examine relations between the built environment and walking. These results are important because they show that relations between the built environment and walking are sensitive to the choice of measure. For example, increasing residential land was associated with walking less for errands using line-based road network buffers but no association was found using circular buffers. The results of this study suggest that previous studies that have used circular buffers but found few associations with physical activity may see associations using line-based road network buffers. Based on these finding we suggest that when formulating policies to increase walking through modifying the built environment policy makers need to consider how the built environment was measured in different studies. Our results demonstrate that conclusions about which types of land use influence walking are sensitive to the type of buffer used.
Model results find that low income respondents, relative to middle income respondents, walk more for errands. Similarly, a study of walking behaviours of high and low socioeconomic adults found low socioeconomic status adults walk more for errands than high socioeconomic adults . It is possible that lower income adults may walk more for errands due to reduced access to an automobile. Income was not significantly associated with walking for leisure. A study of women in Melbourne has also failed to find statistically significant differences in recreational walking by socio-economic status . Increasing BMI was associated with an increased likelihood of walking less for both leisure and errands. These finding are supported by a national survey of Canadians which found that adults who are physically active in their leisure time are less likely to be obese .
In this study recreational land was not associated with walking for leisure. Some previous studies have similarly found no relations and others have demonstrated a positive relationship between recreational land and walking [14, 16, 34, 35]. Future research should take into account the quality of park space and available facilities in the study areas. Increasing institutional land (e.g. libraries, hospitals, and community centres) reduced the odds of walking less for both leisure and errands (using line-based road network buffer) which was expected as institutional land serves both leisure and utilitarian functions. Increasing commercial land reduced the odds of walking less for errands and this was expected given similar findings in other studies [12, 18].
This study has several strengths. First, the geographic areas in this study were pre-selected to have a range of residential densities and income levels. Spatially sampling diverse geographic areas is important to be able to examine relations between the built environment and health outcomes . Second, in this study we used data on walking for leisure and walking for errands because these types of walking may be influenced by different land uses, therefore providing an opportunity to examine both network and circular buffers. Third, the availability of high-resolution land use data allowed us to measure aspects of the built environment in which respondents live.
There are several limitations to this study. The findings are based on cross-sectional data and therefore causality cannot be determined. Another limitation is that six digit postal codes were used as a proxy for respondent's home address and respondents who live in neighbourhood clusters with higher densities may have postal codes that are more accurate than those in lower density neighbourhood clusters. The survey was conducted in the month February, which is typically rainy and wet and the responses may be more conservative than if the survey was conducted in the summer. The survey item assessing walking for errands specifically asked about time spent walking from home to do errands; however the item assessing walking for leisure did not specify walking from home for leisure. In this study we only compared walking for leisure and walking for errands and other types of walking or physical activities were not examined but may also be related to aspects of the built environment. Comprehensive data on total daily physical activity was not available to enable to the construction of cut-off points indicating if respondents met Canada's Physical Activity Guidelines . As such, cut-offs were selected to compare circular and network buffers and may have less utility from a public health perspective. Because we lacked information on where respondents walked we cannot confirm that the line-based road network buffer is a better approximation than the circular buffer of the local neighbourhood that influences walking. Line-based road network buffers were constructed on the assumption that walking occurs on sidewalks or along roads. Data was not available on the location of foot paths in parks or recreational areas where walking may also take place. The current research could be complemented by future studies that examine walking routes and how aspects of the built environment influence these routes. In this study aesthetics or the quality of the built environment was not assessed but may also influence walking behaviours. The data used was from a survey of eight neighbourhood clusters in suburban municipalities with contrasting income levels and residential density. As such, the respondents are not necessarily representative of the Greater Vancouver Regional District or Canadians.
In this study BMI was calculated using self-reported heights and weights which tend to underestimate BMI compared to direct measurements . As the purpose of this study was to examine the influence of land use characteristics on walking using circular and line-based road network buffers we have not accounted for the full range of environmental factors such as social capital, safety, aesthetics, traffic flow, and quality of sidewalks that may influence walking . In this study multilevel analysis, which is appropriate for the hierarchical data structure of individuals nested in eight neighbourhood clusters was not conducted due to an insufficient number of neighbourhood units . The logistic regression models used were unable to account for the non-independence of observations nested within neighbourhoods .
There is increasing interest in examining the influence of the built environment on walking, physical activity and obesity. The results of this study are important because they demonstrate that the selection of spatial units to measure the built environment influences the results of analysis. As high-resolution spatial data is increasingly being used to examine relations between the built environment and physical activity the results of this study demonstrate that researchers need to carefully consider the choice of spatial measure. For many models, greater associations between land use and walking were found using the line-based road network buffer suggesting that these buffers may be more sensitive than circular buffers to detect associations with walking. Existing studies finding no relations between the built environment and physical activity using circular buffers may find associations using line-based road network buffers. We suggest that when making conclusions about the influence of the built environment on physical activity researchers need to carefully consider the methodology used to measure the built environment. The results of this research highlight that future research is needed to develop appropriate measures of the built environment in order to better understand relations with physical activity.
This research was made possible through the support of a grant from the Canadian Institutes for Health Research (CIHR) (# 149353) and a grant from the Canadian Institute for Health Information. We would like to thank Anna-Maria Meyer for contributing extensively to the construction of both sets of buffers and providing additional technical assistance.
- Tjepkema M: Adult obesity. Health Rep. 2006, 17: 9-25.PubMedGoogle Scholar
- Hill JO, Peters JC: Environmental contributions to the obesity epidemic. Science. 1998, 280: 1371-1374.PubMedView ArticleGoogle Scholar
- Handy SL, Boarnet MG, Ewing R, Killingsworth RE: How the built environment affects physical activity: views from urban planning. Am J Prev Med. 2002, 23: 64-73.PubMedView ArticleGoogle Scholar
- Gordon-Larsen P, Nelson MC, Page P, Popkin BM: Inequality in the built environment underlies key health disparities in physical activity and obesity. Pediatrics. 2006, 117: 417-24.PubMedView ArticleGoogle Scholar
- Frank LD, Andresen MA, Schmid TL: Obesity relationships with community design, physical activity, and time spent in cars. Am J Prev Med. 2004, 27: 87-96.PubMedView ArticleGoogle Scholar
- CFLRI: Local opportunities for physical activity and sport: Trends from 1999–2004. 2005, Ottawa, ON: Canadian Fitness and Lifestyle Research InstituteGoogle Scholar
- PHAC: Canada's Physical Activity Guide to Healthy Living. 1998, Ottawa: Public Health Agency of Canada (PHAC)Google Scholar
- Frank LD, Schmid TL, Sallis JF, Chapman J, Saelens BE: Linking objectively measured physical activity with objectively measured urban form: findings from SMARTRAQ. Am J Prev Med. 2005, 28: 117-25.PubMedView ArticleGoogle Scholar
- Wendel-Vos W, Droomers M, Kremers S, Brug J, van Lenthe F: Potential environmental determinants of physical activity in adults: a systematic review. Obes Rev. 2007, 8 (5): 425-440.PubMedView ArticleGoogle Scholar
- Lake A, Townshend T: Obesogenic environments: exploring the built and food environments. J R Soc Health. 2006, 126: 262-7.View ArticleGoogle Scholar
- Lopez RP, Hynes HP: Obesity, physical activity, and the urban environment: public health research needs. Environ Health. 2006, 5: 25-PubMedPubMed CentralView ArticleGoogle Scholar
- Boer R, Zheng Y, Overton A, Ridgeway GK, Cohen DA: Neighborhood Design and Walking Trips in Ten U.S. Metropolitan Areas. Am J Prev Med. 2007, 32: 298-304.PubMedPubMed CentralView ArticleGoogle Scholar
- Michael Y, Beard T, Choi D, Farquhar S, Carlson N: Measuring the influence of built neighborhood environments on walking in older adults. J Aging Phys Act. 2006, 14: 302-12.PubMedGoogle Scholar
- Li F, Fisher KJ, Brownson RC, Bosworth M: Multilevel modelling of built environment characteristics related to neighbourhood walking activity in older adults. J Epidemiol Community Health. 2005, 59: 558-64.PubMedPubMed CentralView ArticleGoogle Scholar
- Ball K, Timperio A, Salmon J, Giles-Corti B, Roberts R, Crawford D: Personal, social and environmental determinants of educational inequalities in walking: a multilevel study. J Epidemiol Community Health. 2007, 61: 108-14.PubMedPubMed CentralView ArticleGoogle Scholar
- Giles-Corti B, Donovan RJ: Socioeconomic Status Differences in Recreational Physical Activity Levels and Real and Perceived Access to a Supportive Physical Environment. Prev Med. 2002, 35: 601-11.PubMedView ArticleGoogle Scholar
- Forsyth A, Oakes J, Schmitz K, Hearst M: Does Residential Density Increase Walking and Other Physical Activity?. Urban Stud. 2007, 44: 679-97.View ArticleGoogle Scholar
- Cervero R: Mixed land-uses and commuting: Evidence from the American Housing Survey. Transportation Research Part A: Policy and Practice. 1996, 30: 361-77.Google Scholar
- Berke EM, Koepsell TD, Moudon AV, Hoskins RE, Larson EB: Association of the built environment with physical activity and obesity in older persons. Am J Public Health. 2007, 97: 486-92.PubMedPubMed CentralView ArticleGoogle Scholar
- Macintyre S, Ellaway A, Cummins S: Place effects on health: how can we conceptualise, operationalise and measure them?. Soc Sci Med. 2002, 55: 125-39.PubMedView ArticleGoogle Scholar
- Northridge M, Sclar E, Biswas P: Sorting out the connections between the built environment and health: A conceptual framework for navigating pathways and planning healthy cities. J Urban Health. 2003, 80: 556-568.PubMedPubMed CentralView ArticleGoogle Scholar
- Owen N, Humpel N, Leslie E, Bauman A, Sallis JF: Understanding environmental influences on walking: Review and research agenda. Am J Prev Med. 2004, 27: 67-76.PubMedView ArticleGoogle Scholar
- Leslie E, Coffee N, Frank L, Owen N, Bauman A, Hugo G: Walkability of local communities: Using geographic information systems to objectively assess relevant environmental attributes. Health Place. 2007, 13: 111-22.PubMedView ArticleGoogle Scholar
- Ross N, Tremblay S, Khan S, Crouse D, Tremblay MS, Berthelot J: Body Mass Index in Urban Canada: Neighbourhood and Metropolitan Area Effects. Am J Public Health. 2007, 97 (3): 500-508.PubMedPubMed CentralView ArticleGoogle Scholar
- Rutt CD, Coleman KJ: Examining the relationships among built environment, physical activity, and body mass index in El Paso, TX. Prev Med. 2005, 40: 831-41.PubMedView ArticleGoogle Scholar
- Nelson MC, Gordon-Larsen P, Song Y, Popkin BM: Built and Social Environments: Associations with Adolescent Overweight and Activity. Am J Prev Med. 2006, 31: 109-17.PubMedView ArticleGoogle Scholar
- Kirtland KA, Porter DE, Addy CL, Neet MJ, Williams JE, Sharpe PA, Neff LJ, Kimsey CD: Environmental measures of physical activity supports: Perception versus reality. Am J Prev Med. 2003, 24: 323-31.PubMedView ArticleGoogle Scholar
- Giles-Corti B, Donovan RJ: The relative influence of individual, social and physical environment determinants of physical activity. Soc Sci Med. 2002, 54: 1793-812.PubMedView ArticleGoogle Scholar
- Schuurman N, Fiedler RS, Grzybowski SCW, Grund D: Defining rational hospital catchments for non-urban areas based on travel-time. Int J Health Geogr. 2006, 5: 43-PubMedPubMed CentralView ArticleGoogle Scholar
- Bow CJ, Waters NM, Faris PD, Seidel JE, Galbraith PD, Knudtson ML, Ghali WA: Accuracy of city postal code coordinates as a proxy for location of residence. Int J Health Geogr. 2004, 3: 1-5.View ArticleGoogle Scholar
- Schuurman N, Leszczynski A, Fiedler R, Grund D, Bell N: Building an integrated cadastral fabric for higher resolution socioeconomic spatial data analysis. Progress in spatial data handling. Edited by: Riedl A, Kainz W, Elmes G. 2006, Berlin: Springer, 897-920.View ArticleGoogle Scholar
- Ford E, Merritt R, Heath G, Powell K, Washburn R, Kriska A, et al: Physical Activity Behaviors in Lower and Higher Socioeconomic Status Populations. Am J Epidemiol. 1991, 133: 1246-56.PubMedGoogle Scholar
- Shields M: Overweight and obesity among children and youth. Health Rep. 2006, 17: 27-42.PubMedGoogle Scholar
- 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: 1127-32.PubMedView ArticleGoogle Scholar
- Neuvonen M, Sievänen T, Tönnes S, Koskela T: Access to green areas and the frequency of visits – A case study in Helsinki. Urban Forestry & Urban Greening. 2007Google Scholar
- Lee C, Moudon AV, Courbois JP: Built Environment and Behavior: Spatial Sampling Using Parcel Data. Ann Epidemiol. 2006, 16: 387-94.PubMedView ArticleGoogle Scholar
- Gorber SC, Tremblay M, Moher D, Gorber B: A comparison of direct vs. self-report measures for assessing height, weight and body mass index: a systematic review. Obes Rev. 2007, 8: 307-26.PubMedView ArticleGoogle Scholar
- Leyden KM: Social capital and the built environment: the importance of walkable neighborhoods. Am J Public Health. 2003, 93: 1546-51.PubMedPubMed CentralView ArticleGoogle Scholar
- Roux , Diez Ana: Investigating Neighborhood and Area Effects on Health. Am J Public Health. 2001, 91: 1783-9.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.