Characteristics of the schools’ surrounding environment, distance from home and active commuting in adolescents from Curitiba, Brazil

REV BRAS EPIDEMIOL 2020; 23: E200065 ABSTRACT: Introduction: Active commuting to school could help increasing physical activity levels among adolescents. However, there is limited understanding on how the relationship between the environment in school surroundings, as well the distance to school, could affect this behavior. Aim: To analyze the characteristics of the environment and distance between house and school with objective measures and their association with active commuting between adolescents of Curitiba, Brazil. Methods: 493 adolescents were interviewed and 124 schools evaluated. The study variables included the schools’ surroundings accessibility characteristics obtained through systematic observation, and the distance between home to school was determined through Geographic Information Systems (GIS) data. Results: The presence of “safety signs” was inversely associated with active commuting (PR = 0.78; 95%CI 0.67–0.91; p = 0.003), as well distance 1,501–3,000 m (PR = 0.53; 95%CI 0.40 – 0.71; p < 0.001) and ≥ 3,501 m (PR 0.29; 95%CI 0.18 – 0.45; p < 0.001). Overall, schools’ surroundings showed walking friendly characteristics. Conclusion: Traffic safety and distance to school were associated with active commuting to school among the study participants. Policies aiming at integrating access to school and traffic safety could help to promoting active commuting among adolescents.


INTRODUCTION
Active commuting to school is an important means of promoting physical activity for adolescents 1 and health promotion 2 . This behavior is extremely relevant, since eight out of ten adolescents do not meet the recommended levels of physical activity 3,4 . Thus, actions such as walking and/or cycling to school can contribute to increasing the time spent in active behaviors 5 .
In some Brazilian cities, active commuting is carried out by less than half of adolescents. In Florianópolis, for example, only 41% actively commute to school 6 . On the other hand, in Pernambuco, 43% of adolescents were considered physically inactive when commuting to school 7 , which exposes a variation in this prevalence between the regions of Brazil [6][7][8][9] .
Some programs to encourage active commuting, such as Caminho da Escola 10 , Bicicleta na Escola 11 and the Bicicleta no seu Bairro 12 projects have been implemented to encourage walking and/or cycling among children and adolescents. However, the success of these programs still depends on the presence of attributes of the built environment, such as bike paths and bike lanes, sidewalks in good condition, and road signs, especially in the streets surrounding schools [13][14][15] . Furthermore, the distance to services and leisure spaces in the community, in addition to the distance traveled by teenagers to school [13][14][15] , are factors related to the chosen mode of commuting.
high-income countries with environmental and social characteristics that are different from Brazil 16 . For example, Brazilian cities have been affected by the sharp increase in private motorized transport, by the increase in public transport fares 17 , and by the precarious or lack of bicycling infrastructure 18 . In addition, the majority of available studies use only the perceptions of environmental attributes and distance to school, which do not permit the verification of factors of the built environment that can contribute positively or negatively with the outcome, making the findings inconsistent. Thus, the objective of this study was to analyze the association between characteristics of the environment around the school, distance from the residence, and the active commuting of adolescents from Curitiba, Brazil.

DESIGN, CHARACTERISTICS OF THE STUDY AND ETHICAL ASPECTS
The study had a cross-sectional design through a household survey 18 and an observational study conducted in the city of Curitiba, Paraná. The city has 1,851,215 inhabitants 20 and is mainly recognized for its urban planning and green areas 21 . The data of the present study are part of an international multicentric project called International Physical Activity and the Environment Network (IPEN), which is carried out in 19 countries, with similar data collection protocol and measures 22

SELECTION OF LOCATIONS
Altogether, there were 2,395 census sectors in Curitiba 23 , and it was considered the primary sampling unit. To maximize the variability of the data, the extremes of walkability and neighborhood income 24 were selected according to data from the Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística -IBGE) 23 . Walkability represents characteristics of the environment that can favor the practice of physical activity, defined by the sum of at least three attributes: residential density, street connectivity and mixed land use 25 . The variables were ordered in deciles and the combinations of the extremes called quadrants were defined: low income and low walkability; low income and high walkability; high income and low walkability; and high income and high walkability. In order to obtain a spatially representative sample, eight census sectors in each quadrant were intentionally selected, totaling the 32 necessary for the study.

SELECTION OF RESIDENCES AND PARTICIPANTS
To select the residences, blocks were considered secondary sampling units. The enrollment process was carried out on all of the blocks, and the first was located at the southwest end of the sector. The residences were approached from the upper left side of the block (all houses present, one by one in a clockwise direction). In case of refusal or if a teenager did not reside there, the next house on the left was visited. In each household, an adolescent and a guardian were selected. The criteria for the selection of adolescents were: female -the youngest and male -the oldest, thus allowing for equal selection between the sexes. If the adolescent refused, another member of the same residence could be invited to the study voluntarily. The minimum sample for the project was 300 adolescents 22 .
Adolescents aged between 12 and 17 years old, living in the city of Curitiba, Paraná, were considered eligible, specifically in the selected census sectors, for at least one year. Furthermore, the teenager had tto study at a school located in the city of Curitiba and not have any physical and/or cognitive limitations that would make physical activity impossible.

HOUSEHOLD SURVEY
The household survey was conducted in two stages. First, an invitation was made for the adolescent and their guardian to participate, and the interview was scheduled. Second, the interviews (guardians and adolescents) were carried out on the days and times previously scheduled. For the guardians, a questionnaire was applied with 215 questions involving environmental characteristics, physical activity and demographic information. For the adolescents, the questionnaire had 285 questions about environment, physical activity, psychosocial aspects, sedentary behavior, occupation, school and sociodemographic information. All interviews were conducted by members of the Research Group on Physical Activity and Quality of Life (Grupo de Pesquisa em Atividade Física e Qualidade de Vida -GPAQ), who received 12-hour theoretical and practical training on the selection criteria, how to approach to residences, how to fill out the forms, how to apply the questionnaire, how to ask the questions, and the appropriate answers. A trial was carried out to simulate the data collection process and to solidify the procedures.

IDENTIFICATION OF SCHOOLS AND THE ASSESSMENT TOOL
The evaluation of the school environment occurred simultaneously with the household survey. The collection took place between the months of August 2013 and May 2014 in public and private schools in the city of Curitiba, Paraná. From the information obtained in the questionnaires in the household survey, the schools in which the adolescents studied were identified, namely: the name of the school, the managing body, address and telephone REV BRAS EPIDEMIOL 2020; 23: E200065 number. At the time, the city had 1,034 schools (212 municipal, 167 state and 655 private) 26 . For access to the institutions, the Municipal Education Secretariat (Secretaria Municipal de Educação -SME) and the Paraná State Education Secretariat (Secretaria Estadual de Educação do Paraná -SEED) were contacted, requesting permission for the research. In addition to the authorization of the secretariats, the school principals had to sign the Informed Consent Form (ICF). All schools selected in the study were located in the city of Curitiba, and offered elementary and/or high school education and included Physical Education in the curriculum. Schools that did not complete all stages of the assessment were excluded.
The school environment and its surroundings were assessed through systematic observation with the School Audit Tool instrument, developed by Jones et al. 27 in the United Kingdom and adapted to the Brazilian context. Characteristics related to the practice of physical activity were observed and divided into four sections: • Access to school; • Area surrounding the school; • Aesthetics and; • School grounds.
The variables of this study are contained in section B of this instrument, and were composed of 14 items. The evaluations were carried out by four independent researchers, who underwent an eight-hour theoretical-practical training.

Characteristics of the environment around the school
The environment around the school was characterized by the structures present in the streets around the space where the school is physically located 27 . These structures were independently assessed in the section "area around the school" (bike paths, bike lanes, sidewalk on both sides, sidewalk on one side, crosswalk, traffic lights or speed reducers, school signs, traffic signs and route signs for cyclists): "no" for not present and "yes" for present. For analysis purposes, the sidewalk variable was added and categorized as "no" for not present and "yes" for present. Thus, it is possible to understand how each variable is related to the outcome. Also, based on the sum of the individual items around the school, a general score was computed, divided into tertiles, and later classified as "low", "medium" and "high".

DISTANCE FROM HOME
The geocoding of the participants occurred based on the self-reported information of the street, such as street name, house number and postal code (CEP). To compose the distance variable, the geographic locations of the residence and the school were considered. Thus, calculating the shortest distance between them through the network of city streets was performed using the command "Network analyst> Route" in ArcGIS 10.0. ESRI ® software 28 . As there is no pre-established cutoff point to quantify the distances, the categorization of the variable was performed based on the findings in the literature. As such, three categories were assigned in this study: ≤ 1,500 m; 1,501-3,500 m; and ≥ 3,501 m 15 .

DEPENDENT VARIABLE
The practice of active commuting to school was assessed by the question "In a normal school week, how many days and how much time during the day do you use the following means of commuting (to and from school)?", considering active commuting: walking, riding a bicycle or using skates to go to and/or from school. For analysis purposes, this variable was dichotomized into: "Do not", for those who did not actively commuting during the week, and "Do", for those who did it ≥ 1 time per week. This classification is commonly used in the literature 13,29 .

COVARIABLES
The adolescent's sex ("male"; "female") was observed, and the age was classified into three age groups ("12-13 years", "14-15 years" and "16-17 years"). Socioeconomic status was assessed with the questionnaire from the Brazilian Association of Research Companies (Associação Brasileira de Empresas de Pesquisa -ABEP) 30 , which was later classified into seven levels. For the analysis, the participants were classified into three categories: "high" (class A), "medium" (class B) and "low" (class C + D + E). Vehicle ownership was assessed according to the number of vehicles present in the residence: "no", when there were no vehicles, and "yes", for ≥ 1 vehicle. The type of school administration was classified as "public" or "private".

DATA ANALYSIS
The frequency distribution and χ 2 test for heterogeneity, linear trend and Fisher's exact test were used to describe the characteristics of the sample. Poisson regression was used to test the crude association between the characteristics of the environment around the school, the distance from the residence and active commuting. In the adjusted analysis, the potential covariates identified in the crude analysis, with p <0.05, were inserted in the final model. The analyzes were performed using Statistical Package for the Social Science (SPSS) 20.0 and STATA 12.0 statistical software, with the significance level maintained at 5%.

RESULTS
Of the 930 adolescents eligible to participate in the project, 493 actually made up the final sample, with a refusal rate of 53.0%, distributed similarly between sex and neighborhood income. One hundred and sixty-three schools were identified, of which ten (6.1%) were ineligible and 29 (17.8%) refused to participate in the study. There was less refusal among public schools than among private schools (public 5.5% versus private 12.9%). The distribution of participants was balanced in terms of sex (girls = 51.2%), but the sample was predominantly composed of adolescents aged 12-13 years (43.1%), of the average socioeconomic level (65.2% ), from public schools (70.6%) and who owned at least one car at home (81.0%). At least three in four schools had at least one bus stop (84.8%), sidewalks (99.0%), pedestrian crossings (75.2%), speed reducers (80.4%), road signs (90.4%) and safety signs (96.3%). Approximately one in ten schools had bicycle paths (12.0%), bicycle lanes (1.5%) and signposting routes for cyclists (11.3%). Most participants lived less than 1,500 m from the school (55.9%) ( Table 1).
Among adolescents who live up to 1,500m from school, the proportion of those who actively commute to school is higher among those enrolled in a public school (public 59.7% versus private 13.6%). Walking is more common among adolescents from public and private schools at distances <1,500 m (59.1 and 13.6%, respectively). In relation to the use of bicycles, for trips to public schools, they were more used in distances <1,500 m (1.5%), while for trips to private schools, the distances they were used was > 3,501 m (2.5%) (Figure 1).

DISCUSSION
This is the first study conducted in Brazil that explored the association between the characteristics of the environment around the school, assessed through systematic observation 27 ,  31 , with the active commuting of adolescents. The combination of methods made it possible to identify quantitative and qualitative attributes of the environment built around schools and the possible routes between homes and schools, which contributes to the understanding of how people choose their mode of transport, and are relevant and innovative factors of the study. Active commuting to school was reported by 62% of adolescents in the sample, with a higher proportion among those enrolled in public schools (public 59.7% versus  private 13.6%), especially when homes are located at a distance up to 1,500 m from the institution. This data can be justified by the enrollment procedure adopted by the State Department of Education (Secretaria Estadual de Educação -SEED), which prioritizes the enrollment of students in public institutions close to their homes, while private institutions do not follow this criterion 32 . However, the farthest distance observed in this study was 22,400 m, suggesting that enrollment in the education system does not necessarily follow the distance to the school as suggested by SEED. As for the type of active commuting, while walking is a popular way to get to school, the use of bicycles in Brazil is less common among teenagers, mainly because there is no culture linked to the use, little or no infrastructure in the neighborhood and around schools, in addition to topographic variations 33 .
Individual characteristics may be one of the factors that justify the choice of active commuting, since the outcome, in the present study, is less frequent among girls and those whose family owns at least one car. In the study by Rech et al. 34 , 56.8% of girls were active commuters to school. The same pattern was observed in the study by Silva et al. 29 , in which 48.6% of the adolescents actively commuted to school in some way. This relationship could be explained by psychological and socio-cultural factors of greater overprotection by parents or guardians in relation to girls, especially when the educational level of parents or guardians is medium-low 34,35 . Furthermore, a family's purchasing power can  interfere in the mode of transport due to the convenience and the perception of safety. Ownership of cars can influence this transport choice, giving preference to car commuting instead of traveling on foot [35][36][37] .

REV BRAS EPIDEMIOL 2020; 23: E200065
The presence of safety signs was inversely associated with active commuting (PR = 0.78; 95%CI 0.66 -0.91). This result is in contradiction with the literature, in which the presence of signing tends to provide a greater perception of safety and increase the chances of active commuting 16 . The findings can be explained, in part, by the fact that, in developing countries such as Brazil, in the busiest places -with regard to traffic or flow of motor vehicles -, there is a greater amount of signage, which aim at greater movement control and the reduction of possible accidents 16,38 . This can inhibit active commuting, considering the large flow of cars and the negative perception of safety, especially by parents or guardians [39][40][41] .
In fact, the findings of the present study demonstrated an inverse association between greater distances, such as 1,501-3,500 m (PR = 0.53; 95%CI 0.40 -0.70) and ≥ 3,501 m (PR = 0.29; 95%CI 0.18 -0.45), and active commuting to school. The results suggest a consistency with the literature that analyzed the distance traveled from the residence to the school, both for perceived measures 41,42 and for objective measures 8 . A study carried out in Belgium by D'Haese et al. 13 pointed out that distances of up to 1,500 m between home and school are suitable for walking, and distances of up to 3,000 m are suitable for cycling. A study carried out in Ireland by Nelson et al. 14 demonstrated that distances above 4,000 m are characterized as a barrier to active commuting, reinforcing that the proximity of the home can stimulate this practice 1,15,43,44 , provided that factors such as perceived safety of adolescents and parents or guardians, in addition to the presence of attributes (walkability, density and accessibility), are linked 16,38 . In Brazil, Silva et al. 8 identified that the main barrier to active commuting was the distance to school. Thus, the results seem to strengthen the importance of policies that integrate access to school and traffic safety. It is also important to highlight that these commuting alternatives have direct implications for the education and health of the population, as well as for the economy and sustainability of cities 45, 46 .
Some limitations must be considered for the proper interpretation of the results of this study. Active commuting was self-reported, which prevents a more accurate estimation of the behavior 47 . The evaluation of the schools' surroundings was restricted to the streets around the block where the institution was located, which does not allow to extrapolate the characteristics to other streets in the community environment. The sample of schools is not representative of the city, and their selection was associated with the institutions in which the adolescents were enrolled. Although the shortest distance through the network of streets, between the residence and the school of the adolescents was measured using GIS, this does not reflect the real route taken by them, like, for example, data provided by Global Positioning System (GPS) 48 . Therefore, in the present study, it was not possible to assess the "quality" of the total area that covers the path that the teenager frequently used or was exposed to, for example, areas of sausage buffer of 25 or 75 m of street segments or sidewalk 49 . Finally, the cross-sectional design limits the causal interpretation between the variables. REV BRAS EPIDEMIOL 2020; 23: E200065

CONCLUSION
The presence of safety signs and the distance between home and school showed an inverse association with active commuting among adolescents from public and private schools in Curitiba, Paraná. The findings indicate that traffic safety and proximity to home can contribute to active commuting to school. Policies that integrate access to schools close to home and traffic safety can contribute to encouraging active commuting to school among teenagers, and also involve the perceptions of teenagers and their parents or guardians. It is necessary, then, to provide improvements in the conditions of the neighborhood and surrounding schools for the development of effective interventions 50 .

ACKNOWLEDGMENTS
The Coordination for the Improvement of Higher Education Personnel (CAPES), the National Institutes of Health (NIH) and the members of the Research Group on Physical Activity and Quality of Life (Grupo de Pesquisa em Atividade Física e Qualidade de Vida -GPAQ).