How does the built environment affect teenagers (aged 13–14) physical activity and fitness? A cross-sectional analysis of the ACTIVE Project

Built environments have been cited as important facilitators of activity and research using geographic information systems (GIS) has emerged as a novel approach in exploring environmental determinants. The Active Children Through Individual Vouchers Evaluation Project used GIS to conduct a cross-sectional analysis of how teenager’s (aged 13–14) environments impacted on their amount of activity and influences fitness. The ACTIVE Project recruited 270 participants aged 13–14 (year 9) from 7 secondary schools in south Wales, UK. Demographic data and objective measures of accelerometery and fitness were collected from each participant between September and December 2016. Objective data was mapped in a GIS alongside datasets relating to activity provision, active travel routes, public transport stops, main roads and natural resources. This study shows that fitness and physical activity are not correlated. Teenagers who had higher levels of activity also had higher levels of sedentary time/inactivity. Teenagers showed higher amounts of moderate-to-vigorous physical activity if their homes were closer to public transport. However, they were also more active if their schools were further away from public transport and natural resources. Teenagers were fitter if schools were closer to natural resources. Sedentary behaviour, fitness and activity do not cluster in the same teenagers. Policymakers/planning committees need to consider this when designing teenage friendly environments. Access to public transport, active travel, green space and activities that teenagers want, and need could make a significant difference to teenage health.


Introduction
Despite the well-documented physiological and psychological benefits of physical activity, many young people are not sufficiently active. Global guidelines recommend 60 minutes of moderate to vigorous physical activity (MVPA) per day in the form of play, games, sports, each participant between September and December 2016. This time of year was selected as it was the beginning of the school year and gave a baseline measure prior to significant engagement with school sport and the physical education curriculum which could have affected activity and cardiovascular fitness.

Data collection
Accelerometery including MVPA and inactivity/sedentary behaviour was measured via Actigraph GT3X+ accelerometers [18]. To improve wear compliance, the accelerometers were worn on the non-dominant wrist of the participant [19,20]. Apart from bathing or swimming, participants were asked to wear their accelerometer for 7 full days and set to record at a frequency of 30Hz [18]. Cut points for activity were taken from Chandler et al. [21]; sedentary behaviour was defined as periods with counts (the unit of measurement for activity used by Actigraph) below 305 counts per 5 seconds, light activity as 306-817 counts and MVPA defined as periods with counts >818. To be included in the analysis, participants needed to wear the accelerometers for at least 500 minutes on 3 or more days. Periods of >60 minutes with 0 count values was classified as non-wear time and excluded from analysis. This criteria is in line with previous studies [22,23]. Sedentary behaviour was also included in the analysis.
This study used the Cooper Run Test (CRT) to assess fitness [24]. The CRT is a 12-minute walk/run test where participants were asked to complete as many laps of a school sports hall as possible in the time [18]. The area to run was marked out by the researchers to ensure participants would complete the full lap. The facilities used to run the CRT differed in size depending on the school's provision and access to space. The number of laps was then converted into a total distance score (in metres). The CRT was carried out during physical education (PE) lessons during school time to avoid disruption [18]. Therefore, participants were split in to two groups so that one group could complete the CRT while the other group recorded scores.
Participants were asked to complete the modified Behavioural Regulation in Exercise Questionnaire (BREQ-2) to measure their motivation to exercise [25] prior to completing the CRT. The 19-item questionnaire provided a total motivation score which was used to define pupils as 'autonomous' or 'controlled' via five subscales; amotivation, external regulation, introjected regulation, identified regulation and intrinsic motivation [26]. The mean of the five subscales forms an idea of whether teenagers are motivated more autonomously or controlled [26]. In this study, the Relative Autonomy Index (RAI) was used to gain insight into the degree of autonomy the teenagers had. This was calculated by weighing each of the subscales and summing the weighted scores; the minimum score of RAI is -24 and the maximum is +20 [26]. In line with Self-Determination Theory (SDT) [27] being more autonomously motivated means that teenage participation in activity is attributed to enjoyment and personal values as opposed to controlled motivation and being made to feel guilty and external pressure to be active [28].
Deprivation scores for both the home and school was from the Welsh Index of Multiple Deprivation (WIMD) [29]. WIMD is based off numerous indicators such as income, access to services, safety, housing and education [29]. The continuous scale was used in analysis with 1 equating to the most deprived area and 1909 to the least deprived [29].
Objective data was mapped in a GIS alongside datasets relating to activity provision, active travel routes, public transport stops, main roads and natural resource. Lle [30], a geo-portal for Wales which is a partnership between Welsh Government and Natural Resources Wales, was used to access open source maps and create these datasets. Participant homes and schools and nearest Euclidean distance were measured from each school and home location to services and provision using QGIS 2.18. This created a database which was exported for statistical analysis.

Statistical analysis
Multivariate linear regression models were estimated in STATA (Version 15). Three models were created to answer how the environment influences a) MVPA and, b) fitness. Prior to conducting the analysis, the data was cleaned to remove any readings that were not required for analysis (i.e. METS/kcals) and any outliers, in particular accelerometry measures which did not meet wear time. Assumptions underlying regression models were confirmed and model fit. Checks of random variation of residuals was undertaken for all models. S1 Table presents the correlation data between the independent variables.
Regression models used to assess the associations of variables in the GIS dataset on MVPA, as well as fitness and motivation (level of significance = p < 0.05). Sex was included in the models as physical activity and fitness levels in particular differ by sex. Relationships between the environment and physical activity was shown via structural equation modelling (SEM) in STATA and also fitness and motivation as secondary aims.

Results
Demographic data (Table 1) shows that there were more boys than girls in the study population (62%). In accordance with previous findings, boys were significantly fitter than girls. Yet, they were not statistically more active (using accelerometer data). Distances to active travel, public transport, main roads, natural recourses and activity providers were similar for the participant's homes and schools on average showing that these built environments have similar provisions. However, boys lived further away from school or from a main road compared to girls.
Due to not meeting the inclusion criteria of accelerometry wear time (>500 minutes per day), 23 participants were excluded from the analysis (n = 247).  19) shows that teenagers showed higher MVPA levels if their homes were closer to public transport. Conversely, they were also more active if their schools were further away from public transport and natural resources. Interestingly, teenagers who had higher levels of activity also had higher levels of physical inactivity, which shows a contrasting relationship between MVPA and inactivity. In this study, over 60% of teenagers met government's recommendations of 60 minutes of MVPA per day on average across the week.

Moderate-to-vigorous physical activity (MVPA)
S2 and S3 Tables show that in terms of MVPA there are some small differences between sex. Girls were more active if they were from more deprived homes and their homes were closer to public transport (S3 Table). Boys were more active if their schools were further away from public transport and natural resources. Boys were also more active if they had higher time spent inactive (S2 Table). Table 3 (R^2 = 0.27) shows boys had higher levels of fitness. Teenagers were fitter if schools were closer to natural resources, which is in contrast to findings regarding activity levels. Teenagers were fitter if they had higher motivation.

Cardiovascular fitness
S4 and S5 Tables show very little influenced boys' fitness but for girls, attending a school further away from a main road and having higher motivation had a significant relationship with being fitter (S5 Table).

Path analysis
Combining all variables using a path analysis model (Fig 1) showed the there was no relationship between levels of MVPA and fitness. The school environment appears integral to fitness; an increasing distance to natural resource and public transport shows a negative effect on fitness. Whereas being further away from active travel and the nearest activity provision shows higher fitness levels. Being more active was influenced by distance to public transport. Teenagers who showed higher levels of MVPA also showed higher inactivity, but inactivity did not affect fitness.

PLOS ONE
How does the environment affect teenagers activity and fitness?

Discussion
The impacts of the built environment on a teenager's physical activity/fitness are not straightforward according to this study. Fitness was associated with going to school near natural resources (e.g. green space). While associations between natural resources and PA have been

PLOS ONE
How does the environment affect teenagers activity and fitness?
positive [4,12,13], this finding shows there may also be associations with fitness too. This is an important link to make as it highlights the cardiovascular benefits to being nearer natural resource for young people and should be considered when planning environments.
In terms of natural resources around schools, the finding that increased proximity improves fitness could mean that some schools' PE/break time provision is better suited to improving fitness due to better space/resources as highlighted in previous research [31]. For example, these school may have increased access to green space for opportunities to participate in wide range of sports across the PE curriculum. They could also facilitate more outdoor, active classrooms. Thus, improving fitness in a variety of ways.
Moreover, these are spaces that teenagers could use their bikes/scooters safely away from roads or play with their friends in green spaces after-school. These spaces could facilitate forms of more structured sport (e.g. football/rugby) or encompass less conventional activities such as den building [7]. Schools with better access to natural resource may be in less urban/ built-up areas thus, further away from high traffic, main roads. Research has attributed roads to impacting activity [9] as safety concerns over traffic are cited in previous studies as reasons why young people are inactive in their communities [32]. These characteristics may encourage inherently more physically active families to live nearby or send their children to schools near natural resource as they are perceived to be conducive to an active lifestyle [32]. This could also underpin the finding that natural resource proximity influences fitness.
When building schools or planning existing school development, this study shows that the school grounds are integral to providing opportunities for better fitness and activity. Currently, school grounds are under-utilised for child-led play and activity when the teaching day ends [33]. Whilst the importance of community access to schools, particularly in more deprived communities is recognised [34], the focus for teenagers is often on adult led and structured activities. Previous studies show this is not what teenagers want and need from provision [5].
There are other findings to consider. Firstly, despite declines in fitness, activity levels actually improve when public transport and natural resource is further away from schools, especially for boys. It could be that the increased walking distance to access these provisions increases activity; suggesting walking can contribute to moderate-to-vigorous activity. This finding highlights that supportive environments and local activity, so teenager's would not need to travel, should be valued to improve their fitness [16,17] but activity levels can be sustained despite distance. Conversely, this study also suggests that activity time improves if teenagers lived closer to public transport and went to schools closer to active travel routes. Therefore, giving them the opportunity to travel independently to activity provision/school without the help of parents/guardians. Previous analysis has shown that being within walking distance of provision is beneficial for teenage fitness [4,14] and therefore, the importance of considering the needs of teenagers when planning environment's should not be overlooked. It would seem that teenagers value enjoyment from activity and therefore willing to travel further to do things they like [28] rather than the convenience of accessing whatever activity is on your doorstep or it is simply a case that there is nothing for them to do in their local communities. Access to provision teenagers like may be more relevant for girls, who are more active if they live closer to public transport suggesting they may be travelling to provision. This is in line with previous research [11], which has acknowledged how important transport infrastructure is, particularly when overcoming accessibility barriers [5,9,10].
This study provides evidence that improving activity and improving fitness are not intrinsically linked. Physical inactivity increases as activity does, especially for boys. Thus, MVPA cannot predict sedentary behaviour/time spent inactive or that being more inactive/sedentary cannot be a single determinant of poor activity, health and fitness [35]. Activities that are likely to influence activity, such as structured, competitive sports may have high periods of inactivity outside this formal training period. These could be the types of activities that boys are more likely to participate in outside of school. Research has shown that girls see high-intensity, competition as a barrier to being active [8]. This finding would suggest that we should promote different types of activity (e.g. light, moderate and vigorous) which have shown benefits to cardiovascular health and fitness, rather than restricting to focus on addressing sedentary behaviour [35]. Environments should improve access to and uptake of a variety of activities that promote beneficial physical activity, rather than simply aiming to reduce time spent inactive.
Interestingly, this study shows that more deprived teenagers are more fit. Moreover, additional analysis shows that girls are more active if they are from more deprived homes. It might be that pupils in deprived schools spend more time doing active travel compared to those who are less deprived. Despite being less likely to engage in activity in the form of structured, competitive sports clubs, it may be that teens from more deprived areas engage more in active travel due to the cost of running a car or getting public transport [36]. Girls, in particular, may enjoy active travel over other types of activity. Motivation was a significant factor in increased fitness in girls, therefore enjoyment is a key principal to focus on when prompting fitness and activity to this group. With this in mind, schools should focus on promoting and maintaining active travel and active travel infrastructure. As shown by previous research, the importance of promoting different types activity in young people cannot be overstated [35,37,38].
Generally speaking, over half of the teenagers involved in this study met the recommendation of 60 minutes of moderate-to-vigorous activity across the week which is high compared to previous data [2]. This could be due to the smaller sample size and consent rates to participate being higher in those more interested in being active. The prominence of the school setting in all outcomes highlights that to improve MVPA and fitness, interventions should centre on the school as a hub for teenagers as this study has shown it as a crucial setting. Where most environment-based studies and interventions focus on the home [4,9,13,[39][40][41], these findings suggest that focus is due on environments around the school to improve fitness and teenagers motivation to be active.

Limitations
This study was only able to measure and ask the opinions of teenagers who consented, and this group may have been more interested in being active and more motivated to be active. Therefore, this study may investigate outcomes of predominantly more active teenagers. Furthermore, this study reports the findings from south Wales, which may not be generalizable to the whole population. Future work should include a larger sample size.
Future study should explore how teenagers travel around their environments; be it by foot, bicycle, car or public transport. This would help provide further context to the time spent in sedentary/MVPA. For example, some teenagers may be more inactive if they are travelling by car to provision rather than travelling by foot or cycling. Moreover, the study's inclusion criteria for accelerometry analysis required participants to wear the device on 3 or more days regardless of weekday or weekend. Future studies should look to include at least one weekend day to note any difference between activity levels throughout the week.
GIS accessibility measures were developed using Euclidean distances which provides an indicative measure of access. In addition, access to public transport is based on bus stop locations as opposed to more sophisticated origin-destination measures. Further work could include more sophisticated network measures of access which take into account urban morphology and whether a destination (e.g. leisure centre) is served by a public transport route from an origin (e.g. home or school).

Conclusions
This study highlights the importance of the school setting in improving MVPA and particularly, fitness for teenagers. Access to public transport, active travel infrastructure, access to green space and bringing activities that teenagers want and need closer to schools could make a significant difference to teenage health. Therefore, policy-makers/planning committees need to consider these provisions when designing teenage friendly environments namely, creating opportunities for teenagers to walk/cycle independently to activity provision around their homes and schools. Additionally, school communities could be utilised to support teenagers to make better use of these spaces when formal learning ends.
Environments that improve PA and fitness for teenagers should focus on bringing activity provision that teenager's want and need into local communities. This provision should include different types of activity (e.g. light, moderate and vigorous) rather than addressing sedentary behaviour as there is evidence there is conflict between improving MVPA and increasing periods of inactivity/sedentary time.
Supporting information S1