This study aimed to identify the prevalence and sociodemographic correlates of domain-specific PA and sitting time among South American adolescents. The prevalence of active adolescents ranged from 7.5% (Brazil) to 19.0% (Suriname). The percentage of students who reported three or more physical education classes per week was from 2.2% (Peru) to 37.5% (Argentina and Colombia), while active commuting to school ranged from 43.1% (Guyana) to 67.8% (Argentina). In addition, we noted that the prevalence of sitting for three or more hours per day was from 24.6% (Bolivia) to 55.6% (Argentina). Our results revealed that boys were more prone to be physically active, engaged in physical education classes, active in commuting to school, and less likely to present higher sitting time compared to girls. Older adolescents (16y +) were less active and more likely to spend three or more hours in sitting time than younger adolescents (12-13y). Adolescents with food insecurity presented a higher frequency of physical education classes and active commuting to school per week, as well as had less higher sitting time than their food security pairs.
Our results extend previous findings showed by Aguilar-Farias et al.  to other countries (Brazil and Paraguay). We also included nationally representative data from Colombia and Ecuador, as well as updating the data from Argentina (from 2012 to 2018) and Suriname (from 2009 to 2016) . Our findings also extended the recent gender inequalities in PA presented by Brazo-Sayavera et al.  for other exposures, such as age group and food security status.
We noted heterogeneity in the percentage of adolescents who reported three or more physical education classes. For instance, approximately one third of students from Argentina, Bolivia, Chile, Colombia, and Suriname reported at least three physical education classes per week, while less than 10% of the Brazilians and Peruvians reported three or more physical education classes per week. UNESCO  reported that in 2013, approximately 21% of Latin America countries were not implementing physical education classes according to obligations or expectations, which may be linked to several factors, such as the autonomy of schools concerning the weekly frequency of physical education classes, and/or the unavailability of adequate infrastructure in schools . In addition, although it is relevant to recognize that the number of Latin American countries with prescribed national physical education curricula is increasing, the weekly frequency of these classes is still low [8, 25]. These results demonstrate the need for national efforts to promote physical education classes across South American countries.
The prevalence of active commuting to school was from 43.1% (Guyana) to 67.8% (Argentina). Indeed, variations in active commuting to school between countries are expected, and may be related to differences in total population, population density, and gross domestic product . Moreover, countries with a low prevalence of active commuting to school should consider implementing policies (i.e., investments in resources, public health regulations) and changes in the built environment (i.e., bike lanes and walking routes to school) to effectively promote active commuting to school.
We also observed variation in sitting time between countries. While less than 30% of adolescents in Peru and Bolivia reported sitting for three or more hours a day, in Argentina, Brazil, Colombia, and Chile, this was reported by more than 40% of the students. These findings show that strategies aimed at reducing sitting time may have different priorities across countries.
We found that boys were more likely to be physically active, and to participate in physical education classes than girls. These findings corroborate previous studies [27, 28] and highlight a gender inequality related to PA across South American countries. Gender inequality in PA is a worldwide challenge, and previous studies have discussed possible explanations, such as parental influence, social support [15, 29], and stereotypes . In addition, our results also revealed that the magnitude of the associations between gender and PA domains differed across the countries observed. For instance, boys were more physically active than girls in almost all countries, while gender disparities related to physical education classes (Argentina, Brazil, and Paraguay) and active commuting to school (Argentina, Brazil, Colombia, Ecuador, and Guyana) were less frequent, which may be related to the high disparities in leisure-time PA observed previously . Beyond the reduction in gender inequality related to leisure-time PA, the abovementioned countries should seek to reduce gender inequality related to physical education classes, given the beneficial effects of participating in these classes on health  and the importance of physical education classes for adolescents to reach the physical activity recommendation [32, 33].
The meta-analysis revealed that older adolescents were less prone to be active than their youngest pairs. Likewise, important results were observed concerning specific countries, where the oldest adolescents from Brazil, Ecuador, Guyana, and Paraguay were less likely to have three or more physical education classes per week than their youngest pairs. Decreasing PA during adolescence has been reported in the literature and different explanations have been suggested, such as lack of time to practice PA and decreasing family support over time . However, our study could raise some insights that may ameliorate this scenario. In this sense, promoting more weekly physical education classes and stimulating the participation of older adolescents, especially from above mentioned countries, in these classes could reduce barriers to PA and increase the odds of reaching the international guidelines.
Older adolescents (16y +) were also more sedentary compared to younger adolescents (12-13y). Our results corroborate findings from high and low-middle- income countries [24, 35], which might be related to increased use of the internet and greater sitting time to perform tasks . Although older adolescents had a higher probability for elevated sitting time (≥ three hours per day), countries such as Argentina, Brazil, Chile, and Colombia presented an elevated prevalence of these behaviors even among the youngest adolescents. Possible explanations for this might be related to high rates of urbanization accompanied by higher access to the internet in these countries , which may facilitate access to electronic devices. However, more studies are needed to confirm this hypothesis. Therefore, countries should consider actions to guide the population, such as advising parents and students about the importance of limiting sedentary time .
Our results revealed that adolescents with food insecurity reported a higher frequency of weekly physical education classes than their peers in a food security situation. This unexpected association may occur because adolescents with food insecurity could attend schools more often, since schools represent a safe environment against some social problems, such as food insecurity . Especially in low- and middle-income countries, physical education classes can be one of the few free-of-charge opportunities for access to sport and different types of structured PA, which might help to reduce existing social inequalities related to PA . Likewise, adolescents with food insecurity had more odds for active commuting to school, which might the preferable way to reach the school given that walking or cycling represents low costs than other passive transportation modes. On other hand, the meta-analysis revealed that the probability of reaching the PA guidelines might to be lower among adolescents with food insecurity [OR = 0.91 (CI95% 0.83; 1.00), I2 = 35.1%]. Although our data do not allow us to analyze leisure PA, previous studies enable us to suggest that the discrepancy in leisure PA could explain discrepancies in PA related to food security status. For instance, a previous cross-national study comparing 52 countries also found that adolescents in the highest wealth quintile were more active in leisure than their peers in the lowest quintile, which can explain the inequality in PA . Therefore, policies to stimulate PA outside school should also be focused on adolescents with economic disadvantage.
Pooled analysis revealed that adolescents who reported food insecurity were less prone to sit more. These findings corroborate a study carried out with 66 low and middle-income countries . Plausible explanations may be related to the greater access to electronic devices by individuals with higher socioeconomic conditions, as well as the need for food-insecure adolescents to be involved in other activities, such as occupational ones . On the other hand, food insecurity was associated with higher sitting time only in Chile. Chile presents the highest income among the countries analyzed, and interestingly, a review carried out by Mielke et al.  revealed that contrary to low-middle income countries, food-insecure adolescents are more prone to present higher sedentary behavior in high-income countries, which highlights the need for country-specific interventions.
Strengths and limitations
Our results need to be interpreted with caution. Studies on the reliability and validity of the questions used in the present study were not available for the countries analyzed. Thus, it is not possible to estimate whether the differences observed between the countries could be partly explained due to biases in the self-reported estimates. In general lines, a previous review showed that self-reported measures tend to overestimate PA, especially among girls . Therefore, it is possible that the gender differences related to PA could be even greater than our estimates. Also, younger adolescents frequently report less accurate information on PA than their older peers , however, the direction of this bias is unknown. Similarly, how the report of PA according to food insecurity is unknown. However, the directions of our findings are in line with studies carried out with objective measures , reinforcing that the inequalities that we found in PA and sitting time are not overridden by bias in the self-report.
The strengths of this study are the use of a large sample from 11 South American countries and the use of comparable instruments to assess domains of PA and sitting time as well as to identify the sociodemographic correlates. On the other hand, some limitations need to be pointed out. 1) The surveys were carried out in different years, which may introduce bias in the analysis. 2) The PeNSE, ENSE, and ENSANUT studies adopted different sampling procedures to the GSHS survey, which could lead to potential differences. However, we minimized some of these differences by seeking to harmonize the age groups. 3) There were slight differences in the questionnaires across surveys. 4) PA and sitting time were assessed by self-reported questionnaires, which can lead to memory, cultural and social desirability biases. Also, information on reliability or validity was not available for the countries analyzed. However, given that technological and financial resources are not always feasible to assess objective PA and sedentary behavior data, self-reported questionnaires have been widely used by the scientific community to guide actions in PA surveillance [6, 17]. 5) GSHS, PeNSE, and ENSE are surveys restricted to adolescents who are in school, which limits the extrapolation of results to adolescents not enrolled in schools. Lack of representativeness of adolescents who do not attend school is a challenge in surveillance of PA and sedentary behavior in several countries, and additional efforts to include this population in future studies are needed . 6) It was not possible to estimate the type of sitting time (e.g., TV-viewing, computer use, listening to music), and future studies could further explore this information, given that health outcomes are dependent on the types of sedentary behaviors performed.