What is the contribution of each physical activity domain to total physical activity in adolescents?

– The present study aimed to verify the contribution of different physical activity domains to “total physical activity” in Brazilian adolescents. This is a cross-sectional study using secondary data from the third edition of the National School Health Survey - PeNSE, 2015. The sample consisted of 100,497 adolescents of both sexes enrolled in the 9 th grade of elementary schools. A linear regression model was used to verify how much each domain contributed to total physical activity, considering gender, type of municipality and region. The domain with the largest contribution to “total physical activity” regardless of sociodemographic and environmental variables was “extra-school physical activity” (R2 = 0.60), followed by “active commuting” (R2 = 0.34), and finally “Physical Education classes” (R2 = 0.23). The contribution of the different domains varied by gender, type of municipality and region, and it was concluded that “extra-school physical activity” make the greatest contribution to “total physical activity”, followed by “active commuting” and “Physical Education classes”. In addition, variation was observed in the contribution of domains by gender, type of municipality and region.


INTRODUCTION
The regular practice of physical activity provides health improvements and acts preventing diseases at all stages of life 1,2 . In adolescence, the adoption of an active lifestyle can estimate the involvement in physical activity in adulthood 3 .
Despite this evidence, it appears that between 11 and 17 years of age, regardless of country, less than 20% of adolescents meet the recommendations of 60 daily minutes of physical activity 4 , however, when analyzing data from Brazil, a prevalence of 30.7% is estimated 5

.
Aiming at increasing the level of physical activity in adolescents, action plans were developed considering the different physical activity domains (leisure, work / school, domestic and commuting) and the context in which they are and can be developed 4 . In this sense, the different domains in which adolescents will engage may make different contributions to total physical activity.
In adolescence, actions and investigations have shown that the main focus is Physical Education classes 6 , extra-school physical activity 7 and active commuting 8 , since there is greater engagement in these domains during this phase of life. However, it is observed that there is variation in the time spent among these domains 9 , indicating the need to verify the contribution of each of these in total physical activity in adolescents.
In this sense, understanding the contribution of the different domains that make up the total physical activity construct is essential for the development of interventions in which behavior change is one of the concerns, in a more responsive way, considering the needs of adolescents. In addition, there are few studies addressing these domains together during adolescence 10 .
Given the above, the present study aimed to verify the contribution of different physical activity domains to total physical activity in Brazilian adolescents.

Study Design
This is a cross-sectional study using secondary data from the third edition of the National School Health Survey -PeNSE, 2015. This edition was developed between April and September 2015 in order to identify risk factors and health protection of Brazilian adolescents regularly attending school in the day shift 11,12 . This study used data referring to the sample named "number one" of PeNSE, 2015. This sample consists of representative data of Brazilian adolescents enrolled in the 9 th grade of elementary school in 2015 11,12 . Fifty-three geographic strata were used, consisting of capitals and noncapital municipalities of each Federation Unit. In capitals, school and class sampling units were used; however, in non-capital municipalities, in addition to these, the IBGE agency was added 11,12

Participants
The survey was conducted with 3040 schools, 4159 classes and 102,301 students answered the survey questionnaire 11 . All students present in the selected classes were invited to participate in the research, but only those who agreed with the Free and Informed Consent Term, participated in the research 11 .

Instruments
Data were collected using an electronic questionnaire and the Personal Digital Assistant (PDA). To complete the questionnaire, students were instructed to consider the last seven days prior to the survey. Further information on PeNSE and its methodological aspects can be obtained from previous publication 11,12 .

Variables
For "active commuting", the average daily time accumulated by the student was used, with commuting to and from school on foot or bicycle. For physical activities performed in "Physical Education classes", the accumulated average duration in which the student practiced physical activity or sport during the period of Physical Education classes at school was used. For "extra-school physical activity", the average daily time accumulated by the student with some physical activity performed during the extra-school period was considered. "Total physical activity" was estimated based on the product between the number of days and the average time spent by students in physical activities in commuting, physical education classes and extra-school physical activity domains, considering the seven days prior to the survey 11 .
Sociodemographic (gender, age and race) and environmental (type of municipality, regions and study shift) variables were used. The characterization and cutoff points of variables used are presented in Box 1.

Statistical analysis
For data interpretation, descriptive analysis and confidence interval (95% CI) were used. Linear regression model was used to verify how much each domain contributed to "total physical activity", considering gender, type of municipality and region. Data are presented with their respective determination coefficient, beta estimator and confidence interval values, and throughout the analysis, 5% significance level (p ≤ 0.05) was considered. In all analyses, sample weights were used as weighting procedure considering instructions available for the PeNSE 2015 survey. Statistical treatment was performed using STATA version 15.0 software.

RESULTS
A total of 102,301 adolescents were interviewed, but 1.804 were excluded due to the lack of important information such as age and / or gender, in addition to data from those who did not attend the 9th grade of elementary school, resulting in a sample with 100,497 participants. The sample consisted of adolescents of both sexes, mostly composed of females aged "≤ 14 years", "non-white", living in capitals and studying in a "part time" system ( Table 1).
The domain that made the greatest contribution to "total physical activity" was "extra-school physical activity", followed by "active commuting" and "physical education classes", respectively ( Table 2). Table 3 shows the contribution of "extra-school physical activity" to the total physical activity of adolescents. For female adolescents living in capitals, "total physical activity" was best explained by "extra-school physical activity" in the Mid-western region, with 56% (R² = 0.56), followed by the Southern region, with 54% (R² = 0.54). However, for those who do not live in capitals, this domain presented the largest contribution in the Southern region, with 60% (R² = 0.60), followed by the Northern region, with 53% (R² = 0.53).
Regarding males, regardless of type of municipality, the largest contribution of "extra-school physical activity" to total physical activity occurred in the Southern region, with 67% (R² = 0.67) and 68% (R² = 0.68) for "capital" and "non-capital", respectively.  Table 4 presents the contribution of "active commuting" to "total physical activity" in adolescents. For female adolescents living in capitals, Note: R2 = determination coefficient; B = estimator B; 95% CI = confidence interval the largest contribution of "active commuting" to "total physical activity" was found in the Northern region, with 42% (R² = 0.42), followed by the Northeast region, with 41% (R² = 0.41). However, regarding those who did not live in capitals, this contribution was higher in the Northeastern region, with 49% (R² = 0.49), followed by the Mid-western region, with 48% (R² = 0.48).
Considering male adolescents living in capitals, the contribution of "active commuting" to "total physical activity" was higher in the Southeast region, with 34% (R² = 0.34); however, for those who did not live in capitals, the largest contribution was in the Southern region, with 35% (R² = 0.35).  Table 5 presents the contribution of "Physical Education classes" to "total physical activity" in adolescents. It was found that for female adolescents, regardless of type of municipality, the largest contribution of "Physical Education classes" to "total physical activity" occurred in the Southern region, with 23% (R² = 0.23) and 24% (R² = 0.24) for "capital" and "non-capital", respectively. Considering males living in capitals, the region in which "Physical Education classes" best explained "total physical activity" was the Mid-western region, with 25% (R² = 0.25); however, for those who do not live in capitals, this domain presented greater explanation in the Southeast region, with 26% (R² = 0.26).

DISCUSSION
As the main result, it was found that the degree of contribution of physical activity domains to "total physical activity", regardless of gender, type of municipality and region, were, respectively, "extra-school physical activity", "active commuting" and "Physical Education classes".
There is agreement between results of the present study and data obtained from PeNSE 2012, in which "extra-school physical activity" made the largest contribution to "total physical activity", followed by "active commuting" and "Physical Education classes" 10 .
Regarding "extra-school physical activity", it was found that for female adolescents living in capitals, this domain had the greatest contribution to "total physical activity" in the Midwestern region, however, for those who did not live in capitals, the greatest contribution occurred in the Southern region. For males, regardless of type of municipality, "extra-school physical activity" made the largest contribution to "total physical activity" in the Southern region.
One possible explanation for the greater engagement in "extra-school physical activity" in adolescents is their autonomy in choosing the activity to engage in 13 and the social support offered, especially by friends, for involvement in extra-school physical activities 14 .
The result of the present study corroborates with previous study 15 in which there was greater contribution of extra-school sports practice to "total physical activity". Similarly, study 9 found that during weekdays, adolescents had higher energy expenditure in sports practiced in the extra-school period. Study 16 found that the practice of "extra-school physical activity" may be a strong predictor of subsequent practice over the years.
For female and male participants living in capitals, "active commuting" made the largest contribution to "total physical activity", respectively, in the Northern and Southeastern regions. However, when considering those who do not live in capitals, this domain made the largest contribution to "total physical activity" in the Northeastern and Southern regions for females and males, respectively. These findings are compatible with previous study 8 , who identified the mode of "active commuting" to school as a correlate of levels of physical activity for both sexes. In contrast, Smith, Aggio, Hamer 17 did not identify association between type of commuting and levels of physical activity. One aspect that should be considered is that, depending on the region of the country, "active commuting" is a compulsory activity, in which sociodemographic and environmental factors can be considered determinant for its practice 8,18 .
Regarding "Physical Education classes" domain, it was found that adolescents of both sexes and living in capitals presented higher values in the Mid-western region compared to the other regions, indicating greater contribution of this domain to the "total physical activity" accumulation. However, for those who do not live in capitals, the "Physical Education classes" domain made the largest contribution to "total physical activity" for females in the Mid-western region and for males in the Southeastern region. These findings can be explained by the variation in the number of Physical Education classes offered in the different regions, considering that regions with the highest economic development have the highest number of Physical Education classes and greater involvement in physical activity during this period 19 .
In this sense, study 20 observed that the greater the offer of Physical Education classes, the greater the involvement of students in moderate to vigorous physical activities throughout the day. Similarly, previous studies have found that, on the days when Physical Education classes are offered, increase in levels of physical activities is observed, and the largest increase in these levels occurred mainly in inactive students 19,21 .
The limitation presented by this study is that the amount of practice in each physical activity domain was self-reported, thus, values may be overestimated or underestimated and may not express the real values of their practice. However, studies have been conducted using subjective methods and the methodological rigor adopted increased the reliability of information collected. As strength, the study has representative sample size and information obtained can be extrapolated, ensuring its internal and external validity.

CONCLUSION
It could be concluded that "extra-school physical activity" make the greatest contribution to "total physical activity", followed by "active commuting" and "Physical Education classes". In addition, variation was observed in the contribution of domains by gender, type of municipality and region.