This cross-sectional study enrolled adolescents of both genders and indicated that, even after adjustment for important confounders associated with PA, regular participation in sports was associated with active leisure time behaviors, but not with sedentary behaviors.
The presented study has several strengths which provide adequate external validity for the results found, such as its large sample size and the control of the design effect in the analyses. Additionally, the high agreement level for collected data is a good indicator of internal validity. However, several limitations also should be recognized. Although the results found were similar to other publications which measured the total PAL [4, 13, 17], the measurement of a fraction of the PA (participation in sports during leisure time) instead of the total amount should be considered as a limitation which could affect the observed high rate of inactivity. Another considered limitation of this study is the sample size calculation, which assumed a prevalence of 41.8% when the observed overall rate of the outcome was approximately 15%. This discrepancy could have implications on the power of the study and, therefore, in the detection of the differences. However, a new sample size was estimated within the same parameters (error of 2.5%, power of 80% and significance of 5%) and including the discovered rate of regular participation in sports, 14.8% (n = 775). A sub-sample was randomly selected and the statistical analyses were again carried out (data not shown). Even with slight differences in the PR values, there was similarity between all sub-sample analyses and the overall sample results in both crude and adjusted models; this fact minimizes the likelihood of bias in the presented results.
Although adopting different cut-offs, the regular engagement in sports observed in these adolescents was low (14.8%) in comparison with those analyzed by Baumert et al.  who identified a rate of American adolescents engaged in sports of 59%. These findings are in agreement with the high prevalence of sedentary lifestyle observed among Brazilian adolescents , and indicated that in this population effective strategies for physical activity promotion are needed.
Furthermore, there was an associated between male gender and higher engagement in sports, which concurs with other studies performed in both more and less developed settings [4, 7]. Recently, Gonçalves et al.  reported that Brazilian male adolescents had more social and family support to engage in physical activities than female adolescents. Furthermore, in Brazil the prevalence of perceived personal barriers to engagement in active leisure-time behaviors is more frequent in the female gender .
The relationship observed between the regular participation in sports and all active behaviors confirms the first hypothesis of this study, since the positive and significant effect of the practice of sports on two active behaviors examined remained statistically significant after adjustment for important confounders associated with PA, such as gender and family socio-economic status. These findings are consistent with those of Baumert et al.  in which adolescents not engaged in sports less often reported getting 30 minutes of nonstop exercise three times a week than did those engaged in sports.
Likewise, in a previous publication, Larson et al. , who analyzed the association between the participation in sports and unhealthy behaviors, indicated that American adolescents who smoke cigarettes might be less likely to participate in ≥1 team sports. Among Brazilian female adolescents, the participation in sports was a protective factor associated with a lower likelihood of pregnancy and early onset of sexual intercourse . These data indicate that the regular participation in sports during adolescence is associated with other healthy lifestyle attitudes; thereby, its relevance for physical health is confirmed, which justifies its continued presence in developing countries' public health programs.
Marshall et al.  in a meta-analysis indicated that, although too small, there is a statistically significant relationship between TV viewing and body fatness. Further, in a recent review, Bryant et al.  presented four possible mechanisms that might explain this relationship (impact of TV viewing on weight gain), and one of these mechanisms was that television displaces time that would otherwise be used for physical activity.
In this present study, the results did not indicate that the regular participation in sports decreased TV viewing during leisure time, and for this reason, the secondary hypothesis of the present study was rejected.
Although several important differences in the methods used in this study, such as the cross-sectional design and PA questionnaire use, must be considered for comparison of our results with previous data, this dissociation agrees with other studies from more developed nations. Ekelund et al.  among European children and adolescents in cross-sectional analysis, observed that the time spent on TV viewing and free time PA measured by accelerometer were not correlated (r = 0.01). Burke et al. , among Australian adolescents, also observed no relation between the time spent in PA during weekdays and TV viewing. Furthermore, Taveras et al.  in a longitudinal study (4 years of follow-up), indicated that changes in TV viewing were not associated with changes in leisure time activities of higher intensities.
Thus, this lack of association seems to occur in both developed and developing countries. Future national surveys, principally in the context of developing nations, should analyze physical activity and inactivity as independent entities and not as the same variable anymore. Besides this, the implications of these results for public health policies are that the newer PA-promotion strategies should not only increase PAL, but also create separate strategies for decreasing the time spent on sedentary behaviors. Another implication is that in clinical practice, health professionals must take sedentary behaviors into account as important risk factors which are not beneficially influenced by PA and, hence, must create tools for combating these behaviors among pediatrics populations.