Abstract
Objective
To assess whether objectively-measured characteristics of the neighbourhood, route and school environments are associated with active commuting to school among children. We also explore whether distance acts as a moderator in this association.
Methods
A cross sectional study of 2012 children (899 boys and 1113 girls) aged 9-10 years attending 92 schools in the county of Norfolk, UK. During the summer of 2007 questionnaires were completed by children and parents. Attributes around the home and route to school were assessed using a Geographical Information System. School environments were assessed using a newly developed school audit and via questionnaires completed by head teachers. Data were analysed in 2008.
Results
Almost half of the children usually walked or cycled to school. Children who lived in a more deprived area and whose route to school was direct were less likely to walk or cycle to school, whilst those who had a higher density of roads in their neighbourhood were more likely to walk. Furthermore, children whose routes had a high density of streetlights were less likely to cycle to school. Distance did not moderate the observed associations.
Discussion
Objectively measured neighbourhood and route factors are associated with walking and cycling to school. However, distance did not moderate the associations found here. Creating environments which are safe, through improving urban design may influence children’s commuting behaviour. Intervention studies are needed to confirm the findings from this observational cross-sectional study.
Introduction
Physical activity provides a number of important benefits for children, including improved physical1 and mental health.2 Walking and cycling to school (or ‘active commuting’) is one way in which children can integrate physical activity into their everyday lives and has previously been identified as a possible target for increasing physical activity levels in children.3 Research suggests that children and adolescents who actively commute to school tend to be more active than those who do not4 and this may contribute significantly to children’s overall physical activity.5
According to the socio-ecological model of health behaviour,6 a variety of contextual influences are likely to be important in determining health behaviour. A recent review highlighted that in addition to individual factors such as age and gender, a broad range of environmental factors may influence children’s active commuting.7 Several authors, including Moudon and Lee,8 have suggested that three environmental components should be considered as possible influences on active commuting; the neighbourhood around the home, the route between home and school, and the environment of the school itself.
Studies that have examined one or more of these components have used either self-reported perceptions of the environment from participants or objective measures generated using geographical information systems (GIS) or street audits. Current research using objective methods suggests that the presence of pavements and mixed land use around the school is associated with a higher prevalence of walking or cycling to school,9, 10 whilst having to cross a busy road on route to school or having less route options have been negatively associated with children’s active commuting.11
There are four main limitations of work to date. Distance is an important determinant of travel behaviour7 and may moderate the association between environmental factors and active commuting,12 yet there is a lack of studies examining this in the context of travel to school. Secondly, many studies fail to separate walking and cycling. As the characteristics of environments encouraging walking among children may be very different to those supporting cycling, failure to consider the behaviours separately may mean that study outcomes and environmental exposures are not appropriately matched.13 Thirdly, the environmental measures commonly studied are often based on those developed to assess the supportiveness of the environment for walking among adults. Our understanding of how these factors are associated with cycling behaviour or how associations may differ in children is limited. Finally, the majority of studies to date have been conducted in urban areas in the USA and Australia, both countries with typical urban layouts that are not commonly found in regions such as Europe. Hence there is a need to examine how the environment might be associated with children’s active commuting in different environmental settings.
In order to address these limitations we quantified a range of associations between objectively measured environmental characteristics and walking and cycling to school in a large sample of British schoolchildren selected from urban and rural neighbourhoods. We also explored the hypothesis that the associations between environmental factors and children’s active commuting behaviours are moderated by the distance travelled to school.
Methods
Study design and setting
Children included in this analysis participated in the SPEEDY study (Sport, Physical activity and Eating behaviour: Environmental Determinants in Young people). The methods of recruitment, sampling and overall sample representativeness of the study have been described in detail elsewhere.14 Briefly, children were sampled through schools in the county of Norfolk, Eastern England. Ethical approval for the study was obtained from the University of East Anglia local research ethics committee and informed consent was provided by participating children and their parents. 2064 children aged 9-10 years from 92 schools across the county were recruited to the study. During summer 2007, a team of trained field workers visited each school to distribute questionnaires for children and their parents or guardians, and measure each child’s height and weight according to standard operating procedures. Body mass index (BMI) was calculated in kg/m2 and used to classify children into categories as described by Cole and colleagues.15
Children reported their usual travel mode to school (‘by car’, ‘by bus or train’, ‘on foot’ and ‘by bike’) via questionnaires. Responses were collapsed into three categories; ‘motorised travel to school’, ‘cycle to school’ and ‘walk to school’. Parents or guardians provided information on their access to or ownership of a car, mode of travel to work and educational qualifications by questionnaire. Based on the highest qualification reported, parents were assigned to one of three educational attainment categories; low (high school or less), medium (vocational qualifications above high school) and high (university education or above).
Neighbourhood delineation and environmental measures
Objective assessments of neighbourhood environmental factors were computed using a GIS (ESRI ArcGIS 9.2). Children’s home addresses were converted into a map location using the Ordnance Survey Address Layer 2 product, a dataset that identifies precise locations for all registered addresses in Great Britain.16 The neighbourhood of each child was defined using the street network (Ordnance Survey Integrated Transport Network) as the area within an approximate 10 minute walk (corresponding to 800m) of their home. The network included publicly accessible roads and pedestrianised streets, and public footpaths were added from maps supplied by local authorities.
Thirteen measures were chosen to reflect a variety of characteristics within the neighbourhood which might support walking and cycling or act as barriers (Table 1). These capture detailed characteristics of the area such as traffic safety, the provision of pavements for walking and street connectivity, as well as general indicators such as socioeconomic deprivation. These measures have either been hypothesised to be associated with adult’s walking or cycling17 or have been associated with active commuting in empirical work.7 Table 1 also provides an overview of the methods, data sources and classifications used. Classifications were based on those previously used in published studies, or where no prior method was available continuous data were collapsed into quartiles, and categorical data were simplified into two or three categories as appropriate.
Table 1.
Description and distribution of objectively measured environmental variables by distance to school
| Variables | Description | Data | Classification | Percentage prevalence (n) | P value | |
|---|---|---|---|---|---|---|
| source | Distance <1km | Distance >1km | ||||
| Neighbourhood (Area within 800m around child’s home) | ||||||
| Road outside child’s home | A major or minor road adjacent to the child’s home | A | A/B/Minor | 30.5 (232) | 31.0 (388) | n.s |
| Local & private | 69.5 (528) | 69.0 (864) | ||||
| Road density | Total road lengths divided by neighbourhood area | A | Lowest road density | 13.2 (100) | 32.0 (401) | 0.001 |
| Second quartile | 30.5 (232) | 21.8 (273) | ||||
| Third quartile | 28.4 (216) | 22.9 (287) | ||||
| Highest road density | 27.9 (212) | 23.2 (291) | ||||
| Proportion of primary (A) | Length of A roads divided by total road length | A | No A roads | 56.8 (432) | 58.1 (728) | n.s |
| roads | Some A roads | 43.2 (328) | 41.9 (524) | |||
| Building density | Total area of buildings divided by neighbourhood area | B & C | Lowest building density | 13.2 (100) | 32.2 (403) | 0.001 |
| Second quartile | 29.5 (224) | 22.3 (279) | ||||
| Third quartile | 29.2 (222) | 22.4 (281) | ||||
| Highest building density | 28.2 (214) | 23.0 (289) | ||||
| Streetlight density | Number of streetlights divided by total road length | D | Lowest streetlight density | 17.0 (129) | 31.9 (399) | 0.001 |
| Second quartile | 33.9 (258) | 17.6 (220) | ||||
| Third quartile | 22.8 (173) | 26.4 (330) | ||||
| Highest streetlight density | 26.3 (200) | 26.4 (303) | ||||
| Traffic accidents per km | Number of fatal or serious road traffic accidents between | E | None | 30.8 (234) | 41.2 (516) | 0.001 |
| 2002-2005 divided by total road length | Any | 69.2 (526) | 58.8 (736) | |||
| Pavement density | Area of pavements divided by total road length | B | Lowest pavement density | 13.2 (100) | 32.1 (402) | 0.001 |
| Second quartile | 25.9 (197) | 24.6 (308) | ||||
| Third quartile | 29.2 (222) | 19.6 (246) | ||||
| Highest pavement density | 31.7 (241) | 23.6 (296) | ||||
| Effective walkable area | Total neighbourhood area (the area that can be reached via | A | Lowest EFA | 23.2 (176) | 26.1 (327) | 0.001 |
| (EFA) | the street network within 800m from the home) divided by | Second quartile | 28.5 (216) | 22.8 (285) | ||
| the potential walkable area (the area generated using a | Third quartile | 28.3 (215) | 23.0 (288) | |||
| circular buffer with a radius of 800m from the home). | Highest EFA | 20.0 (152) | 28.1 (351) | |||
| Connected node ratio | Number of junctions divided by number of junctions and cul- | A | <0.7 Low connectivity | 13.2 (89) | 8.3 (169) | 0.001 |
| (CNR) | de-sacs | >0.7 High connectivity1 | 88.2 (671) | 91.7 (1083) | ||
| Junction density | Number of junctions divided by total neighbourhood area | A | Lowest junction density | 24.2 (184) | 26.0 (326) | 0.001 |
| Second quartile | 32.5 (247) | 20.2 (253) | ||||
| Third quartile | 30.9 (235) | 21.2 (265) | ||||
| Highest junction density | 12.4 (94) | 32.6 (408) | ||||
| Land use mix | Proportion of each land use2 squared and summed | C & F | Highest land use mix | 27.9 (212) | 23.2 (291) | 0.001 |
| Second quartile | 29.1 (221) | 22.5 (282) | ||||
| Third quartile | 29.9 (227) | 22.0 (276) | ||||
| Lowest land use mix | 13.2 (100) | 32.2 (403) | ||||
| Socioeconomic deprivation | Population weighted scores for neighbourhood G Least deprived | G | Least deprived | 25.9 (197) | 24.4 (305) | 0.004 |
| Second quartile | 21.4 (163) | 27.2 (341) | ||||
| Third quartile | 24.1 (183) | 25.6 (320) | ||||
| Most deprived | 28.6 (217) | 22.8 (286) | ||||
| Urban-rural status | Urban-rural classification of child’s home address | H | Urban | 43.4 (330) | 37.1 (795) | 0.001 |
| Town and Fringe | 34.7 (264) | 24.6 (308) | ||||
| Village | 21.8 (166) | 38.3 (479) | ||||
|
| ||||||
| Route (Area within a 100m buffer of the shortest route to school) | ||||||
| Streetlight density | Number of streetlights within 100m of route divided by | D | Lowest streetlight density | 46.2 (351) | 26.2 (328) | 0.001 |
| route length | Second quartile | 6.8 (52) | 22.0 (276) | |||
| Third quartile | 18.6 (141) | 28.8 (361) | ||||
| Highest streetlight density | 28.4 (216) | 22.9 (287) | ||||
| Traffic accidents per km | Number of fatal or serious road traffic accidents between | E | None | 72.8 (553) | 30.5 (380) | 0.001 |
| 2002-2005 within 100m of route divided by route length | Any | 27.2 (207) | 69.5 (866) | |||
| Main road en route | Presence of primary (A) road as part of route | A | No | 86.8 (660) | 49.1 (615) | 0.001 |
| Yes | 13.2 (100) | 50.9 (737) | ||||
| Main or Secondary road en | Presence of primary (A) or secondary (B) road as part of | A | No | 77.1 (586) | 33.2 (416) | 0.001 |
| route | route | Yes | 22.9 (174) | 66.8 (836) | ||
| Route length ratio | Route length divided by the straight line distance between | A | ≤1.6 Low directness | 26.6 (202) | 26.4 (330) | n.s |
| the home and school | <1.6 High3 | 73.4 (558) | 73.6 (922) | |||
| Percentage of route to | Proportion of route which passes through urban area | H | <100% urban | 11.7 (89) | 59.3 (742) | 0.001 |
| school within an urban area | 100% urban | 88.3 (671) | 40.7 (510) | |||
| Land use mix | Proportion of each land use2 within 100m of route squared | C & F Highest land use mix | Highest land use mix | 27.9 (212) | 23.2 (291) | 0.001 |
| and summed | Second quartile | 29.1 (221) | 22.5 (282) | |||
| Third quartile | 29.9 (227) | 22.0 (276) | ||||
| Lowest land use mix | 13.2 (100) | 32.2 (403) | ||||
|
| ||||||
| School | ||||||
| Travel plan | Presence of school has a travel plan (a formal document, | I | No | 15.0 (114) | 16.5 (206) | n.s |
| which identifies ways to encourage walking, cycling or use | Yes | 85.0 (646) | 83.5 (1046) | |||
| of public transport to school) | ||||||
| Walking bus | Presence of walking bus (where a group of children walk to | I | No | 95.8 (728) | 95.8 (1200) | n.s |
| school along a route accompanied by adults, picking up | Yes | 4.2 (32) | 4.2 (52) | |||
| ‘Walk to School’ initiative | The school has a walk to school initiative (period during | I | No | 31.3 (238) | 27.9 (347) | n.s |
| which children are encouraged to walk to school) | Yes | 68.7 (522) | 72.3 (905) | |||
| Pedestrian training | The school offers pedestrian training | I | No | 59.2 (450) | 55.6 (696) | n.s |
| Yes | 39.8 (302) | 44.4 (547) | ||||
| Entrance for | The school has separate entrance(s) for pedestrians and | I | No | 26.6 (202) | 25.9 (324) | n.s |
| pedestrians/cyclists | cyclists | Yes | 72.1 (548) | 72.4 (906) | ||
| Lollypop person | The school has a lollypop person (road crossing guard/school | I | No | 54.2 (412) | 63.3 (793) | 0.001 |
| crossing supervisor/school road patrol) | Yes | 43.4 (330) | 35.5 (445) | |||
| Cycle racks | The school has cycle racks for use by children | J | No | 10.8 (82) | 12.3 (154) | n.s |
| Yes | 89.2 (678) | 87.7 (1098) | ||||
| Land use mix around the | Single or mixed land use surrounding school | J | Single land use | 70.4 (535) | 70.1 (878) | n.s |
| school | Mixed land use | 29.6 (225) | 29.9 (374) | |||
| Pavements | Pavements (sidewalks) visible from the school entrance | J | None/On one side | 86.0 (653) | 65.4 (819) | 0.001 |
| On both sides | 14.1 (107) | 34.6 (433) | ||||
| On road/shared cycle paths | Cycle paths visible from the school entrance | J | No | 87.6 (666) | 92.1 (1153) | 0.004 |
| Yes | 12.3 (94) | 7.9 (99) | ||||
| Traffic calming | Traffic calming measures visible from the school entrance | J | No | 58.9 (448) | 69.9 (872) | 0.001 |
| Yes | 41.1 (312) | 30.4 (380) | ||||
| Pedestrian crossing | Pedestrian crossing visible from the school entrance | J | No | 89.6 (681) | 93.7 (1173) | 0.001 |
| Yes | 10.4 (79) | 6.3 (79) | ||||
A OS Integrated Transport Network, B OS Mastermap, C Address Layer 2, D Local Authority, E Norfolk & Suffolk Constabulary, F Land Cover 2000, G Index of Multiple Deprivation, H Urban-rural classification, I Teacher Questionnaire, J School grounds audit.
n.s not significant. P values indicate the differences in neighbourhood, route, and school categorical measures between those children who lived more or less than 1km from school.
Connectivity; Classification previously used by Schlossberg et al., (2005).
Seventeen different land uses were classified: farmland, woodland, grassland, uncultivated land, other urban, beach, marshland, sea, small settlement, private gardens, parks, residential, commercial, multiple use buildings, other buildings, unclassified buildings and roads. This score is also known as the Herfindahl-Hirschman Index developed by Rodriguez and Song (2005)
Route length ratio; Classification previously used by Dill (2004)
Route identification and environmental measures
To enable the delineation of a route between the home and school location for each child, the locations of all access points, such as gates and driveways, at each school were noted by researchers by on-foot survey. Assuming that children would use their nearest access point, the shortest route between each child’s home and school was identified in the GIS using the modified street network and route lengths were calculated. Each route was then buffered by a distance of 100m, as this was felt to be an appropriate distance of the environment proximal to the road, and seven measures of the environment of the route falling within this zone were computed.
School environment characteristics
A team of trained fieldworkers conducted audits of the school grounds and assessed three components of the school environment; facility provision for walking and cycling in the area surrounding the school and in the school grounds, as well as school policy towards walking and cycling. Further details of the tool developed for this purpose are available from the authors. All head teachers also completed a questionnaire which included seven items that allowed school policy towards active commuting to be determined.
Data Analysis
Analyses were undertaken in 2008 using SPSS version 16 (SPSS Inc). Cross tabulations were generated to compare the number of children reporting the use of different travel modes across personal and demographic groups. Differences in neighbourhood, route, and school categorical measures between those children who lived more or less than 1km from school were tested using chi-squared tests. To identify predictors of children’s travel mode, multilevel statistical models were fitted using the MLWin version 2.10 package.18 A two-level structure of children (level 1) nested within schools (level 2) was applied in order to account for clustering of children’s characteristics, including behaviours, within schools. Multinomial outcomes were specified in the models with a three category outcome of ‘motorised travel’, ‘walking’, or ‘cycling’, with motorised travel as the reference category. Analysis was stratified by two route length categories; less than or equal to 1km (n= 760 children), and greater than 1km (n= 1252 children). These cut-offs were chosen to ensure adequate statistical power to detect associations in each strata.
Two sets of models were created. The first, partially adjusted, examined the effects of the factors of interest listed in Table 1 separately, whilst adjusting for the hypothesised confounding effect of age, gender, child BMI, household car access, within category variation in route length, and maternal travel mode to work. Maternal, rather than paternal, travel mode to work was chosen as weaker associations were found with the latter measure. The second set of ‘best fit’ models fully adjusted for all predictors included in the model. Variables were retained in models based on the goodness of fit (a statistical significance of α ≤ 0.05), and if the direction of effect was expected and consistent between partially and fully adjusted models. If variables showed strong correlations with each other, the one which was most strongly associated with active commuting behaviour was selected for multivariate analysis. Strong associations (r>0.75) were seen between neighbourhood building density and streetlight density, as well as road density, junction density and urban rural status (all between r=0.72 and r=0.84). Of the route measures, associations were seen between road traffic accidents per km and whether the route included a primary or secondary road (r=0.54) as well as whether the route included a primary road and whether the route included a primary or secondary road r=0.75). For the school environment, only measures of school pedestrian training and school walking initiatives were strongly associated (r=0.68). In order to investigate any moderating effects of distance on associations, an interaction term (distance × predictor) was added to the models.
Results
Sample Characteristics
From the sample of 2064 children who participated in the SPEEDY study, 52 participants (2.5%) were excluded; 41 failed to provide an address which could be located and 11 gave no information on travel mode to school. No significant differences were noted between participants excluded from analysis and the main sample.
The sample contained more girls than boys (55.3% versus 44.7%), mean age (± SD) was 10.25 (± 0.3) years. Forty percent of children reported usually walking to school, 9% usually cycled and the remainder used a motorised form of travel. 77% of children were normal weight and 40% lived in an urban area. Most parents had access to or ownership of a car (95%), usually travelled to work using motorised travel (60.6 % of mothers) and were in the low (39.0%) or middle educational attainment categories (41.0%).
Neighbourhood, route and school based environmental factors
The two rightmost pairs of columns in Table 1 present the prevalence of neighbourhood, route, and school categorical measures according to the distance to school. Overall, 77% of children lived in neighbourhoods which were deemed to have high connectivity and over half of children had a route to school which was completely within an urban area, although the prevalence of many of the environmental measures varied by distance to school. The majority of schools reported having policies which would promote active commuting, with most children attending schools with a school travel plan (84%).
Correlates of active commuting behaviour
Table 2 shows the direction and statistical significance of associations from the partially adjusted models for both walking and cycling behaviours, stratified by distance. In the overall sample many of the variables, including those from neighbourhood, route and school environments, were significantly associated with walking or cycling behaviours. The results are generally inconsistent. Nevertheless, there is some evidence that school policies to promote walking and cycling are associated with more active commuting behaviour. In addition, the presence of cycling infrastructure was associated with more cycling. The table shows no evidence that these effects are modified by distance.
Table 2.
Direction of association between neighbourhood, route and school variables and active travel, stratified by distance to school
| Overall sample | Distance to school <1km | Distance to school >1km | ||||
|---|---|---|---|---|---|---|
| Environmental attributes | Cycle to school n=186 Direction |
Walk to school n=805 Direction |
Cycle to school n=89 Direction |
Walk to school n=533 Direction |
Cycle to school n=97 Direction |
Walk to school n=272 Direction |
| Neighbourhood characteristics | ||||||
| Road outside child’s home (A/B/minor= referent) | − n.s | + n.s | + n.s | + n.s | − n.s | + n.s |
| Road density (lowest density=referent) | − n.s | + * | − n.s | + n.s | − * | +* |
| Proportion of primary (A) roads (lowest proportion= referent) | n.s | − ** | − * | − n.s | − n.s | − n.s |
| Building density (lowest density=referent) | − n.s | + n.s | − * | + n.s | − ** | + ** |
| Streetlight density (lowest density=referent) | − ** | + n.s | − ** | − n.s | − n.s | + ** |
| Traffic accidents per km of roads (none =referent) | − n.s | − n.s | + n.s | + n.s | − n.s | + n.s |
| Pavement density (lowest density =referent) | − n.s | + n.s | − n.s | + n.s | + n.s | − n.s |
| Effective walkable area (lowest connectivity=referent) | + n.s | + * | − n.s | + n.s | + n.s | + ** |
| Connected node ratio (low connectivity=referent) | − n.s | − ** | − n.s | − n.s | − ** | − ** |
| Junction density (lowest density= referent) | + n.s | − * | + n.s | + n.s | + n.s | − ** |
| Land use mix (highest mix= referent) | + * | − n.s | + n.s | − n.s | + n.s | − n.s |
| Socioeconomic deprivation (least deprived= referent) | − ** | − ** | − ** | − n.s | − n.s | − * |
| Urban-rural status (urban= referent) | − n.s | − ** | + * | − n.s | − ** | − ** |
|
| ||||||
| Route characteristics | ||||||
| Streetlight density(lowest density=referent) | − * | + n.s | − n.s | − n.s | − n.s | + ** |
| Traffic accidents per km of route (none =referent) | − n.s | −n.s | + n.s | − n.s | − n.s | − n.s |
| Main road en route (no =referent) | − n.s | − ** | − n.s | + n.s | − n.s | + n.s |
| Main or Secondary road en route (no =referent) | − n.s | − n.s | − n.s | − n.s | − n.s | + n.s |
| Route length ratio (low directness=referent) | − n.s | − ** | + n.s | +** | − n.s | − ** |
| Percentage of route within an urban area (<100%= referent) | + n.s | + ** | − n.s | − n.s | + n.s | + n.s |
| Land use mix (highest mix= referent) | + * | − n.s | + n.s | − n.s | + n.s | − n.s |
|
| ||||||
| School characteristics | ||||||
| Travel plan (no=referent) | + ** | + ** | + n.s | + ** | − n.s | − * |
| Walking bus (no= referent) | + n.s | − n.s | + n.s | + n.s | + n.s | − n.s |
| ‘Walk to school’ initiative (no=referent) | + n.s | + ** | + ** | + ** | + n.s | − n.s |
| Pedestrian training (no=referent) | − n.s | + n.s | + n.s | − n.s | − n.s | + * |
| Entrances for pedestrians/cyclists (No= referent) | − * | − n.s | + n.s | + n.s | − n.s | + * |
| Lollypop person (no=referent) | + n.s | − n.s | + n.s | −n.s | + n.s | −n.s |
| Land use mix around the school (Single land use =referent) | − n.s | − n.s | + n.s | + n.s | − n.s | + ** |
| Pavements (none = referent) | − n.s | − * | − n.s | + n.s | + n.s | − n.s |
| On road /shared cyclepaths (no=referent) | + ** | − n.s | + n.s | + n.s | + n.s | − n.s |
| Traffic calming (no=referent) | + n.s | − n.s | − n.s | + n.s | − n.s | + * |
| Pedestrian crossing (no=referent) | − n.s | + n.s | − n.s | + n.s | + n.s | + ** |
Direction indicates direction of association (+ = positive association, − = negative association) when variables tested for trend, n.s.= not statistically significant
=p<0.05
=p<0.01
All associations are adjusted for child age, gender, BMI, parental car access, maternal travel mode to work and journey length.
Fully adjusted models are shown in Table 3. These results indicate that children who lived in a more deprived area were less likely to walk or cycle to school. Those who had a higher density of roads in their neighbourhood were more likely to walk, whilst children whose route was direct were less likely to walk. Furthermore, children whose routes had a high density of streetlights were less likely to cycle to school. When the moderating effects of distance on the associations between environmental factors and active commuting were tested, none of the interaction terms were statistically significant (α ≤ 0.05).
Table 3.
Fully adjusted associations (best-fit model) between neighbourhood, route and school variables and active travel
| Characteristic | Cycle to school OR (95%CI) |
Walk to school OR (95%CI) |
|---|---|---|
| Neighbourhood characteristics | ||
| Connected node ratio (low connectivity = referent) Road density (lowest density=referent) |
0.96 (0.50-1.84) | 0.49 (0.31-0.76)** |
| Second quartile | 1.39 (0.86- 2.27) | 1.94 (1.33-2.82)** |
| Third quartile | 1.53 (0.90-2.54) | 2.74 (1.85-4.07)** |
| Highest density | 1.31 (0.72-2.36) | 3.22 (2.09-4.94)** |
| Socioeconomic deprivation (least deprived=referent) | ||
| Second quartile | 0.59 (0.35-0.97)* | 0.85 (0.59-1.22) |
| Third quartile | 0.72 (0.43-1.20) | 0.55 (0.37-0.81)** |
| Most deprived | 0.47 (0.26-0.85)* | 0.45 (0.29-0.70)** |
| Route characteristics | ||
| Distance to school >1km (<1km =referent) | 0.27 (0.18-0.39)** | 0.10 (0.06-0.11)** |
| Route length ratio (low directness= referent) | 0.62 (0.43-0.90)* | 0.47 (0.36-0.61)** |
| Main road en route (no= referent) | 0.50 (0.32-0.78)** | 0.65 (0.48-0.89)** |
=p<0.05
=p<0.01
All associations are adjusted for child age, gender, BMI, parental car access, maternal travel mode to work and journey length
Discussion
This is one of the first studies to examine the association between objectively measured characteristics of neighbourhoods, routes, and schools and children’s reported active commuting in the UK, and to consider the moderating effects of distance required to travel to school. We found evidence that children who lived in highly connected, more deprived neighbourhoods, with routes to school which were short, direct and included a busy road were less likely to walk or cycle to school. These associations were not moderated by distance required to travel to school.
Even though levels of active commuting are higher than those reported in many similar studies conducted in the US19, the findings are generally similar to those reported here; that busy roads11, 20 and direct routes11 are barriers to active commuting in children. Although direct routes may be associated with more walking in adults, Timperio and colleagues11 have suggested that the contrary finding in children may be because high route directness is often associated with greater traffic flow, and busy roads may be particularly avoided by children, although we found the association was independent of the presence of a main road en-route in our sample.
In contrast to much of the literature which reports a positive association between deprivation and active commuting (for example, Harten and Olds21), we found that children from more deprived neighbourhoods were less likely to walk or cycle to school, even after controlling for distance travelled and car ownership. Whilst this is somewhat counter-intuitive, the same finding was reported by Timperio and colleagues11 although the effect of deprivation disappeared in that work after controlling for maternal education. The reason for our finding is not apparent, although it is possible that unmeasured factors influencing child travel behaviour are socially patterned in a way that is associated with area deprivation. For example, we have shown that parental perceptions of neighbourhood safety are predictors of children’s travel mode22 and it could be that more positive perceptions in more affluent areas are acting to encourage active travel amongst children.
We hypothesised that distance from school would act as a moderator in the relationship between environmental factors and active commuting. Not surprisingly children living closer to school were more likely to walk or cycle, but there was no evidence that distance moderated any other associations. We found that the environmental correlates for both walking and cycling behaviours were similar, although the structure of the pedestrian network was found to be more strongly associated with walking. Findings from the US10 which suggest that urban form is an important driver of children’s active commuting, are supported here. However, in our study children were more likely to walk to school if they lived in neighbourhoods with low network connectivity but high density. Taken together these results suggest that environments with a high number of route choices but with less connected, and hence quieter, streets are particularly supportive of walking for children.
Although we examined the effects of the school environment, no school related measures proved to be statistically significant predictors of behaviour in our fully adjusted models. While this may be related to a lack of heterogeneity for these variables, our findings suggest that interventions within neighbourhood environments may be more successful in encouraging children’s active commuting. Indeed, our view that addressing components of road safety and urban design will be central to interventions to increase children’s active commuting is supported by recent UK National Institute for Clinical Excellence recommendations.23 Nevertheless, given the pivotal role that parents play in determining children’s travel behaviour,12 it is unlikely that environmental modifications alone will be sufficient unless they also gain parental support.
Our study has a number of strengths and limitations. Data were collected in a large, population-based study incorporating sampling to generate significant variation in environments. We also used objective methods to measure the characteristics of neighbourhoods and routes which might influence the decision to walk or cycle. However, our data are cross-sectional and hence we cannot infer causality to the observed relationships. Furthermore, we purposefully sampled children from a narrow age range to obtain a homogenous study population, however this limits the generalisability of our findings to children of other ages. Our sample consisted of predominantly white children so we were not able to examine how our findings might be influenced by ethnicity. Whilst we used a modified version of the street network to delineate neighbourhoods and routes, we did not have information on informal cut-throughs. Our modelled routes were also based on the assumption that children would choose the shortest distance, and may not exactly match with the actual routes used.11 As we assessed usual travel mode to school, we were unable to assess the importance of weather on children’s active commuting, although the climate in Norfolk is relatively benign and weather may be less important in this setting than other authors have observed. We acknowledge that parents perceptions of the environment are important influences on children’s active commuting patterns, and have previously reported on these associations22. The aim of the current paper was to study this influence of objectively measured characteristics only.
Further work, using large sample sizes in heterogeneous environments would confirm the findings reported here. As the environment in which children travel to school can often change markedly, for example during the transition from primary to secondary school, we believe that there is particular potential for studies that utilise longitudinal designs to examine the impact of changes on active commuting behaviours.
Conclusion
A combination of objectively measured neighbourhood and route characteristics were associated with children’s active commuting behaviour. We found no evidence that these associations varied according to the distance to school, or that differences in the school environment were important. Creating neighbourhoods that provide safe and quiet routes to school may increase participation in active commuting
Footnotes
The final published version of this article can be found at http://www.ajpmonline.org/article/S0749-3797(09)00848-4/abstract
doi: 10.1016/j.amepre.2009.10.040
References
- 1.Andersen L, Harro M, Sardinha L, Froberg K, Ekelund U, Brage S, et al. Physical activity and clustered cardiovascular risk in children: a cross-sectional study (The European Youth Heart Study) Lancet. 2006;368:299–304. doi: 10.1016/S0140-6736(06)69075-2. [DOI] [PubMed] [Google Scholar]
- 2.Biddle S, Gorely T, Stensel D. Health-enhancing physical activity and sedentary behaviour in children and adolescents. Journal of Sports Sciences. 2004;22(8):679–701. doi: 10.1080/02640410410001712412. [DOI] [PubMed] [Google Scholar]
- 3.Tudor-Locke C, Ainsworth B, Popkin B. Active Commuting to School An Overlooked Source of Childrens’ Physical Activity? Sports Med. 2001;31(5):309–313. doi: 10.2165/00007256-200131050-00001. [DOI] [PubMed] [Google Scholar]
- 4.Lee MC, Orenstein MR, Richardson MJ. Systematic Review of Active Commuting to School and Children’s Physical Activity and Weight. Journal of Physical Activity & Health. 2008;5(6):930–949. doi: 10.1123/jpah.5.6.930. [DOI] [PubMed] [Google Scholar]
- 5.van Sluijs EMF, Fearne V, Mattocks C, Riddoch C, Ness A, Griffin S. The contribution of active travel to children’s physical activity levels: Cross-sectional results from the ALSPAC study. Prev Med. 2009;48:519–524. doi: 10.1016/j.ypmed.2009.03.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Sallis J, Owen N. Ecological models. In: Glanz K, Lewis FM, Rimer BK, editors. Health Behaviour and Health Education. Theory, Research and Practice. second edition Jossey-Bass; San Francisco: 1997. pp. 403–424. [Google Scholar]
- 7.Davison KK, Werder JL, Lawson CT. [[accessed 18.08.08]];Children’s active commuting to school: Current knowledge and future directions. Prev Chronic Dis [serial online] 2005. 2008 Oct;5(3) [PMC free article] [PubMed] [Google Scholar]
- 8.Moudon A, Lee C. Walking and Bicycling: An Evaluation of Environmental Audit Instruments. American Journal of Health Promotion. 2003;18(1):21–37. doi: 10.4278/0890-1171-18.1.21. [DOI] [PubMed] [Google Scholar]
- 9.Kerr J, Frank L, Sallis JF, Chapman J. Urban form correlates of pedestrian travel in youth: Differences by gender, race-ethnicity and household attributes. Transportation Research Part D: Transport and Environment. 2007;12(3):177–182. [Google Scholar]
- 10.McMillan TE. The relative influence of urban form on a child’s travel mode to school. Transportation Research Part A: Policy and Practice. 2007;41(1):69–79. [Google Scholar]
- 11.Timperio A, Ball K, Salmon J, Roberts R, Giles-corti B, Simmons D, et al. Personal, Family, Social, and Environmental Correlates of Active Commuting to School. American Journal of Preventive Medicine. 2006;30(1):45–51. doi: 10.1016/j.amepre.2005.08.047. [DOI] [PubMed] [Google Scholar]
- 12.Panter JR, Jones AP, van Sluijs EMF. Environmental determinants of active travel in youth: A review and framework for future research. International Journal of Behavioral Nutrition and Physical Activity. 2008;5(1):34. doi: 10.1186/1479-5868-5-34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Giles-Corti B, Timperio A, Bull F, Pikora T. Understanding physical activity environmental correlates: increased specificity for ecological models. Exerc Sport Sci Rev. 2005;33:175–181. doi: 10.1097/00003677-200510000-00005. [DOI] [PubMed] [Google Scholar]
- 14.van Sluijs EM, Skidmore PML, Mwanza K, Jones AP, Callaghan AM, Ekelund U, et al. Physical activity and dietary behaviour in a population-based sample of British 10-year old children: the SPEEDY study (Sport, Physical activity and Eating behaviour: Environmental Determinants in Young people) BMC Public Health. 2008;8(8) doi: 10.1186/1471-2458-8-388. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide:international survey. British Medical Journal. 2000;320(7244):1240–1243. doi: 10.1136/bmj.320.7244.1240. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ordnance Survey [[accessed 18.08.08]];OS MasterMap Address Layer. 2006 Available at: http://www.ordnancesurvey.co.uk/oswebsite/products/osmastermap/layers/address/index.html.
- 17.Pikora T, Giles-Corti B, Bull F, Jamrozik K, Donovan R. Developing a framework for assessment of the environmental determinants of walking and cycling. Social Science & Medicine. 2003;56(8):1693–1703. doi: 10.1016/s0277-9536(02)00163-6. [DOI] [PubMed] [Google Scholar]
- 18.Rasbash J, Steele F, Browne W, Prosser B. A user’s guide to MLwiN version 2.0. Institute of Education; London: 2004. [Google Scholar]
- 19.Fulton JE, Shisler JL, Yore MM, Caspersen CJ. Active transportation to school: Findings from a national survey. Research Quarterly for Exercise and sport. 2005;76(3):352–357. doi: 10.1080/02701367.2005.10599306. [DOI] [PubMed] [Google Scholar]
- 20.Schlossberg M, Greene J, Phillips P, Johnson B, Parker B. School Trips: Effects of Urban Form and Distance on Travel Mode. Journal of American Planning Association. 2006;72(3):337–346. [Google Scholar]
- 21.Harten N, Olds T. Patterns of active transport in 11-12 year old Australian children. Australian and New Zealand Journal of Public Health. 2004;28(2):167–72. doi: 10.1111/j.1467-842x.2004.tb00931.x. [DOI] [PubMed] [Google Scholar]
- 22.Panter JR, Jones AP, van Sluijs EMF, Griffin SJ. Attitudes, social support and environmental perceptions as predictors of active commuting behaviour in school children. J Epidemiol Community Health. 2009 doi: 10.1136/jech.2009.086918. jech.2009.086918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.National Institute for Health and Clinical Excellence . Promoting physical activity for children and young people. London: 2009. [Google Scholar]