School environment and physical activity in adolescents from São Paulo city

We examined the association of physical activity (PA) facilities and access to school with total PA and domain-specific PA in adolescents. We enrolled 2610 adolescents (mean: 14.9 years) from Sao Paulo city. The number and presence of sports courts, swimming pools, locker rooms, running/athletics tracks, entrance accessible for student cyclists, bike racks, speed limit signal around the school, and pedestrian crossing were assessed in each school. All participants responded to a questionnaire about frequency and duration of physical education classes, leisure time, and active commuting. Total PA was obtained by adding up all PA domains. Presence of three or more (OR: 1.62; 95% CI: 1.15 to 2.30) sports courts, swimming pool available in usable conditions (OR: 1.45; 95% CI: 1.01 to 2.10), running/athletics tracks (OR: 2.35; 95% CI: 1.07 to 5.18), and bike racks (OR: 1.38; 95% CI: 1.07 to 1.78) were positively associated with total PA. Number of sports courts, swimming pool available in usable conditions, speed limit signals around the school, and pedestrian crossings were positively associated with physical education classes. The bike racks, speed limit signs around the school, and pedestrian crossings were positively associated with active commuting. School environment was associated with increased PA. Our findings should be considered in future epidemiologic studies and for educational and health policy makers.

Several studies have shown the benefits of physical activity for adolescents' health [1][2][3] . Physical activity during childhood and adolescence is associated with adiposity status, cardiometabolic biomarkers, mental, psychological, behavioral conduct/pro-social behavior, cognition/academic achievement, and physical fitness in the short, mid-and long-term 2 . Despite these benefits, fewer than 20% of adolescents worldwide and only 29% of Brazilian children aged 11-17 years participate in sufficient physical activity 4 . For adolescents aged 5-17 years, the 2020 World Health Organization Guidelines for physical activity recommends an average of 60 min/day of moderate-to-vigorous physical activity 1 .
Promoting physical activity in adolescents living in megacities is challenging, particularly in low-and middleincome countries which have faced remarkable urbanization in the last decades. By 2030, the world is expected to have 41 megacities, and more than 90% of the future urban population growth will be in low-and middleincome countries 5 . Sao Paulo (Brazil) is the largest megacity in Latin America and one of the largest megacities in the world. With approximately 20.5 million inhabitants 6 , Sao Paulo is characterized by regions with extreme poverty, high occurrence of violence, heavy traffic, air and noise pollution, insecurity spread throughout the city. Moreover, similar to other megacities in low-and middle-income countries, Sao Paulo has faced a fast-paced an unplanned urban growth 7,8 . These changes in the city's urban environment are likely to impact physical activity 9,10 , as demonstrated by our previous study which showed that only 12.7% of adolescents from Sao Paulo engaged in sufficient physical activity 9 .
Physical activity is determined not solely by individual behavior and choices, but also by sociocultural and environmental factors such as school environments 11,12 . Factors that may affect adolescents' physical activity during school hours include physical, social, or institutional structures. These factors are in line with ecological models of behavior change in which an environment can influence health behaviors 13,14 .
Scientific Reports | (2021) 11:18118 | https://doi.org/10.1038/s41598-021-97671-z www.nature.com/scientificreports/ In addition, monetary aspects may affect the ability of schools to provide opportunities to encourage physical activity among adolescents within the school environment. There is a growing body of research into the influence of the school environment, particularly barriers and enablers, in adolescents' physical activity 15,16 . A review 16 and a mixed-method review 15 demonstrated that schools are more able to enhance adolescents' physical activity when they emphasize resource provision for physical activity within the school day, create a ''culture'' of physical activity, train teachers to support a positive climate for physical activity promotion, and ensure extracurricular physical activity opportunities for all adolescents. However, few studies have shown the association between facilities for physical activity in the school and access to school with total and domain-specific physical activity in adolescents living in megacities from low-to middle-income countries 17 .
Given adolescents are likely to spend a substantially amount of daily hours in the school, alteration of physical activity in youth should consider the school environment 9,18,19 . Furthermore, non-availability of, or nonaccessibility to activities at school have been reported as main barriers to physical activity involvement among adolescents 20,21 . Therefore, physical activity facilities in schools and access to school are central reasons to be measured in the promotion of physical activity in adolescents 21 .
Studies on adolescents from the high-income countries found significant associations between physical activity facilities and total physical activity [22][23][24] . However, there are relatively few representative studies about these factors mega cities in low-to middle income countries, particularly in Latin America 25 . Furthermore, to analyze the associations between physical activity facilities and access to school may contribute to the physical activity opportunities of the Latin American adolescents. In this study, we examined the association of physical activity facilities and the access to school with total physical domain-specific physical activity in adolescents from Sao Paulo city, Brazil. . Written informed consent was obtained from a parent or legal guardian of all participants from the individual studies. All methods were carried out in accordance with relevant guidelines and regulations. Further details on SP-PROSO have been described elsewhere 9 .

Methods
A multi-stage sampling process was used to select participants of SP-PROSO. According to the school Census conducted in 2015, Sao Paulo had 175,854 ninth-grade students, who were enrolled across 2086 public (state and municipal) and private schools. Initially, 156 schools were randomly selected, of which 119 agreed to participate. Of the 61 private schools drawn, 26 refused to participate and 3 did not respond to our invitation and were excluded. Of the public schools, there was only one loss among the state schools and seven among the municipal schools. In each school, one class was selected and all students were eligible and invited to participle. Of 2816 eligible students, 2702 responded to a questionnaire. Our final analytical sample included 2610 participants with complete information on physical activity, which represented a final response rate of 93% (Fig. 1). Details on participant sampling, recruitment strategies, refused participants, and those who did not respond to our invitation have been published elsewhere 9,26 . Physical activity assessment. Participants reported their physical activity levels by completing a selfreported questionnaire. This questionnaire has been used in the National Survey of School Health (2009, and 2019) 27 . The indicators of physical activity from questionnaire showed satisfactory relative validity compared with three 24-h recalls for 300 min/week (sensitivity: 78%; specificity: 69%; accuracy index: 73%) and 150 min/ week (sensitivity: 88%; specificity: 49%; accuracy index: 78%) of adolescent´s physical activity 28 .
The participants were instructed to report their weekly frequency and duration of physical activity in the domains of physical education classes, leisure physical activity, and active commuting (walking or cycling) to or from school during the past week.
These questions were asked separately for each domain of physical activity. We calculated total physical activity by adding the minutes per day (min/day) of each domain of physical activity (physical education at school, leisure time physical activity, and active commuting). Total physical activity and leisure time of adolescents was categorized as < 60 min/day or as ≥ 60 min/day, according on the 2020 World Health Organization Recommendation for physical activity 1 . Physical education classes was categorized based on the number of classes offered per week in most Sao Paulo schools 29 as < 2 or ≥ 2 week. Since only 2.1% of the students met physical activity guidelines thresholds for commuting in our sample and there are no specific guidelines for active commuting, we opt to categorize active commuting on weekly frequency, as none or ≥ 1 week.
School-environment assessment. The school-environment was assessed using a questionnaire completed by a designated staff member, who observed each physical activity facility and access to school separately. Physical activity facilities and access to school considered were: number of sports courts (none to three or more; categorized as ≤ 1, 2 and ≥ 3), availability of swimming pool (does not have one, it is not in usable conditions, in usable conditions; categorized as none and not fit for use, and in usable conditions), locker room (does not have one, it is not in usable conditions, in usable conditions; categorized as none and not fit for use, and in usable conditions), running/athletics tracks (does not have one, has one), entrance accessible for student cyclists (no, www.nature.com/scientificreports/ yes), bike rack (no, yes), speed limit signal around the school (no, yes), and pedestrian crossing (no, yes). Type of school (municipal public, state public, and private) also was considered 9,21,30 .
We also generated a composite variable based on the "number of physical activity facilities" at the school by adding the number of physical activity facilities (yes = 1; no = 0) (sports courts, availability of swimming pool, locker room, running/athletics tracks, entrance accessible for student cyclists, bike racks, speed limit signal around the school, and pedestrian crossing. The variable ranged from zero (no facility available) to eight (five facilities evaluated were available and in usable conditions). The number of physical activity facilities was categorized as ≤ 1, 2-4, and ≥ 5. Sociodemographic variables. The socioeconomic level was assessed based on mother's education level (incomplete elementary, incomplete secondary, complete secondary or complete higher) and a series of enquiries that explored the possession of goods and the presence of a maid in the household. These questions have been used in the National School-Based Health Survey (Pesquisa Nacional de Saúde do Escolar-PeNSE). The PeNSE survey involved questions about the existence of goods in the home (the home where the adolescents actually lived). This questionnaire and method of evaluating socioeconomic level has been extensively used in Brazilian surveys 9,30 .
The socioeconomic level was designed through principal component analysis (PCA) 30 . PCA was run adding to the model: maternal educational level, type of school, landline, cell phone, computer with internet access in the room, car, and number of bathrooms within the household. Details about the components of the PCA and the equation were published earlier 9,30 . Socioeconomic level was characterized as terciles of the total wealth scores, in which the first tercile (low) is the poorest group and the third tercile (high) is the wealthiest group.
Participants also self-reported their skin color/ethnicity [white, black, mixed, others (yellow, indigenous)] and mother's employment status (none, part-time, and full-time).
Statistical analysis. Descriptive statistics included frequency and proportions of adolescent characteristics by total physical activity level (< 60 min/day vs ≥ 60 min/day) and school characteristics. We also calculated the proportion and its respective 95% confidence interval (95% CI) of total and domain-specific physical activity (physical education classes, leisure time physical activity, and active commuting) according to school characteristics.
A multilevel logistic regression model with the number of enrolled students in the school as the first level and type of schools as the second level, were used to estimate the odds ratios (OR) and 95% confidence intervals (95% CI) for the associations of school environment with total physical activity and domain-specific physical activity, adjusted for sex, age (years), skin color/ethnicity, and mother's employment status. We performed unadjusted and adjusted models including physical activity facilities and access to school separately (number of sports courts, swimming pools, locker rooms, running/athletics tracks, accessible entrance for student cyclists, bike racks, speed limit signal around the school, pedestrian crossing, and number of physical activity facilities) for total physical activity and each domain of physical activity.
The proportion of adolescents who did ≥ 60 min/day of total physical activity was 13%, with a higher proportion in boys (18.1%) than girls (7.4%); those in the third tercile of wealth score (17.6%) than in the first tercile of wealth score (9.2%); those whose mothers completed higher education (15.2%) than those whose mothers did not complete middle school (10.9%); and those whose mothers had a full time job (14.0%) than those whose mothers were unemployed (6.7%) ( Table 1).
Most of the schools were municipal public (40.0%) and had at least two sports courts (53.0%); 92% of schools did not have swimming pools in usable condition, 74% did not have locker rooms in usable condition (73.9%) and 98% did not have running/athletics tracks (98.3%). Over half schools had accessible entrances for student cyclists, pedestrian crossings, and two or more physical activity facilities. Over 50% of schools did not have bike racks and speed limit signs around the school (  (Table 4).
We did not observe associations between any physical activity facilities and extracurricular sports activities in schools with leisure-time physical activity (here defined as ≥ 60 min/day) ( Table 5).

Discussion
This cross-sectional study examined the associations of school environment with total physical activity and domain-specific physical activity in adolescents from the largest city in Latin America, Sao Paulo. We found that higher number of sports courts (≥ 3), swimming pool available in usable conditions, running/athletics tracks, and bike racks were positively associated with total physical activity (≥ 60 min/day). We found specific associations between school characteristics and domains of physical activity in adolescents. Our study provides highlight the importance of school environment to promote physical activity for adolescents.
We found a positive association between the presence of two or more sports courts, swimming pools available in usable conditions, speed limit signs around the school, and pedestrian crossings and participation in physical education classes. Previous studies have shown that adolescents spent more time in physical education classes in schools that had sports court availability 21,32 . A systematic review concluded that students attending schools with the higher number of physical environment features may have increased physical activity (compared to lower number of facilities) 33 . Therefore, improving the amount, variety and condition of the facilities in the school environment may be relevant strategies to increase physical activity at school and enhance the participation in physical education classes. The present results corroborate previous studies that have attributed a significant association of the school context variability on physical activity. It is likely that, in the case of the www.nature.com/scientificreports/ present study, this fact was due to the sampling randomness in each school, as well as the possible similarity regarding the infrastructure.
We not found significant associations between any physical activity facilities and extracurricular sports activities in schools with leisure time physical activity (≥ 60 min/day). On the other hand, Rezende et al. 21 used data collected for PeNSE (Pesquisa Nacional de Saúde do Escolar-National Survey of School Health) showed positive associations between number of sports court (≥ 2), the presence of running/athletics track, and swimming pool available in usable condition with leisure-time physical activity in adolescents from Brazil. The difference in these findings may be due to the different cut-offs. The authors concluded that school context has important potential to increase the engagement of adolescents in physical activity. Therefore, school environment has a fundamental role in allowing adolescents, who spend part of their day in school, to have a more active lifestyle 21 . Furthermore, there is still no physical activity guidelines for each domain of physical activity 1 . Of note, leisure time physical activity contribute to 41.3% of total physical activity in adolescents from Sao Paulo 9 . Therefore, physical activity interventions, especially recreational activities, are important to increase the total physical activity in adolescent from Sao Paulo.
Our study also showed an association between school environment (bike racks, speed limit signs around the school, and pedestrian crossings) and active commuting (≥ 1 week). Active commuting is an essential source of physical activity and has been associated with higher physical activity levels in adolescents 9 . Safety, social support for commuting and the built environment have been reported to be important determinants of active commuting in adolescents from high-income countries 14 . Furthermore, density of exercise facilities and urbanization (i.e., urban versus rural residences) are positively associated with physical activity 13 . Distance from home to school is the most common barrier to active commuting to school, especially because it is the primary factor in the Table 3. Association between school environment and total physical activity in adolescents. Values in bold indicate significant associations; 95% CI 95% confidence intervals, OR odds ratio. *Number of adolescents enrolled in the school as first level and type of school as second level. **Number of adolescents enrolled in the school as first level, type of school as second level, adjusted by sex, age (years), skin color/ethnicity, and mother's employment status. www.nature.com/scientificreports/ parents' decision-making process for or against allowing their child to walk or cycle to school 34 . In addition, it may not be feasible to walk or cycle to school if the distance is too far in medium-sized towns possibly because of higher traffic density 35 . Sao Paulo is among the 10 most urbanized, highly disorganized cities, with heavy traffic, air and noise pollution, rising crime rates and high-income inequality 8 . Future research should address the physical and social environmental determinants of active commuting such as safety, presence of sidewalks and bicycle lanes, and availability of infrastructure to verify this hypothesis in adolescents from low-income to middle income countries. Our results are also consistent with previous studies showing a positive association between the amount of school facilities and adolescent physical activity 36,37 . A significant number of physical activity facilities must be implemented in schools, as this might help to offer a diversity of opportunities for physical activity 38 . Possibly, more recreational activities and spaces with equipment and/or 'obstacles (natural or built), which stimulate the increase of the total physical activity levels 39 . The proposal for a diverse infrastructure and open school would enhance the capability of the school to promote societal knowledge related to body culture and understanding of concepts such as leisure goals, expression of feelings, affections and emotions, and opportunities to promote, restore and maintain health, one of the aims of the Programa Saúde na Escola (Health in Schools Program), in Brazil 40 . Table 4. Association between school characteristics and participation in physical education classes in adolescents. Values in bold indicate significant associations; 95% CI 95% confidence intervals, OR odds ratio. *Number of adolescents enrolled in the school as first level and type of school as second level. **Number of adolescents enrolled in the school as first level, type of school as second level, adjusted by sex, age (years), skin color/ethnicity, and mother's employment status. www.nature.com/scientificreports/ Our study included a large representative sample of adolescents from the city of Sao Paulo to examine the association between school environment and physical activity. However, our findings should be interpreted in the light of some limitations. First, the cross-sectional associations must be interpreted cautiously, as they may not be appropriate for causal inference. Second, SP-PROSO study did not include adolescents of all ages and the results may not applicable to all ages. Third, physical activity information was collected through a self-reported questionnaire and therefore measurement error probably occurred. However, this physical activity questionnaire has shown a high relative 28 . The variable of mothers' education level has considerable missing data. Adolescents with missing maternal education data showed similar physical activity pattern 9 .

Conclusions
We found that schools with a higher number of physical activity facilities had higher odds of adolescents reaching the physical activity guidelines. Physical activity facilities were differentially associated with domain-specific physical activity, which can be useful to inform the development of future physical activity programs. The school environment may have an important role in allowing adolescents, who spend part of their day in school, to have a more active lifestyle. Prospective studies of school characteristics and physical activity are needed, as well as evidence from intervention studies in order to advance our understanding of these relationships and better inform educational and public health policy, programs and decision-makers. Table 5. Association between school characteristics and leisure time physical activity in adolescents. 95% CI 95% confidence intervals, OR odds ratio. *Number of adolescents enrolled in the school as first level and type of school as second level. **Number of adolescents enrolled in the school as first level, type of school as second level, adjusted by sex, age (years), skin color/ethnicity, and mother's employment status.  Table 6. Association between school characteristics and active commuting to/from school in adolescents. Values in bold indicate significant associations; 95% CI 95% confidence intervals, OR odds ratio. *Number of adolescents enrolled in the school as first level and type of school as second level. **Number of adolescents enrolled in the school as first level, type of school as second level, adjusted by sex, age (years), skin color/ ethnicity, and mother's employment status.