This study aimed to identify the predictors and health consequences of screen-time changes from 11 to 15 years of age in a birth cohort from Southern Brazil. To our best knowledge, this is the first study in Latin America to address this subject and one of the few in the international literature. We investigated both the predictors and outcomes of screen-time change in adolescents.
Some limitations of our study are the self-reported information of screen time and the short follow-up period. Another possible limitation was to have analyzed change in time spent watching TV, playing video game, and using the computer as a composite measure. Nonetheless, stratifying the analyses for each component revealed that most associations were observed in the same direction, although the strongest associations have seen a change in computer time. An issue to be considered when interpreting our findings is that adolescent boys and girls are growing at rapid rates from age 11 to 15 years. This might influence physical activity practice, and as a consequence, screen time as well.
A systematic review showed that screen time tracks from childhood and adolescence, thus suggesting that interventions should start as early as possible 
. Our study focused on some key behaviors such as watching TV, using computer, and playing video games. We found an increase of 60 minutes (95% confidence interval [CI]: 53.0–66.0) in the overall screen time from age 11 to 15 years. Computer use increased (82 minutes; 95% CI: 78.0–85.0), TV viewing declined (−12.3 min; 95% CI:−16.8 to −7.8), and no change occurred in time spent playing video games (.4 minutes; 95% CI: −3.0 to 2.4).
Previous studies found that watching TV/videos/DVDs and using a computer for fun were the most popular sedentary behavior among students, corresponding with more than one-half of all sedentary time spent by Australian adolescents 
. Consistent with previous studies [1,4,20,21]
, our study found that boys reported more hours of screen time than girls. In American adolescents, there was no difference in mean time of watching TV/video between sexes. However, boys spent more time playing computer games than girls, whereas girls spent more time sitting and listening to music and talking on the telephone than boys 
. In Spanish children, males also spent more time playing games—consoles—and engaged in more time spent in all screen-viewing behaviors than females 
Our findings showed greater screen time among adolescents most biologically matured compared with those in previous stages. However, when adjusted for all variables, biological maturation lost its association, suggesting that other factors might influence this relationship. Researchers have shown that more screen time occurs in British adolescents in greater biological development stages 
. Indeed, with the advancement of puberty stages, adolescents may feel more attracted to new and complex technologies.
Screen time in our study was greater in adolescents of white skin color, higher maternal education, and improved assets index. Apart from skin color, these variables remained significant in the final model. Another study also found lower screen time among black adolescents compared with white teens 
. However, a British study revealed that black students accumulated more screen time than their white counterparts 
. Recently, a review of studies revealed an inverse association among sedentary behaviors and ethnicity (Caucasian), socioeconomic status, and parental education in adolescents [8,26]
. Conversely, in a systematic review involving only prospective studies, authors found insufficient evidence for the socioeconomic determinants of sedentary behavior 
This scenario shows that these associations are still inconclusive and inconsistent, suggesting that predictors of screen time might be influenced by particular characteristics of the studied population, advances of and access to technology, and the local culture of technology consumption in each country. Nevertheless, more evidence involving prospective studies is extremely relevant to understanding predictors of screen time and the reasons why some studies have found inverse association [8,26]
, no association [9,19]
, or direct associations (e.g., our study). A survey using data from countries participating in the Health Behaviour in School-Aged Children study found that the engagement in screen activities may have different meaning in different cultures and may be influenced by behavioral traditions 
, which partially may explain these differences in findings, reinforcing our suppositions.
Our findings indicated that overall screen time was lower in adolescents with poor school performance in all tested models. Several studies with babies and children demonstrated that high exposure to TV could result in poorer cognitive development, short-term memory, attention, language skills, and academic achievement [7,27]
. However, as pointed out by Roberts et al 
, media is becoming more integrated into the lives of young people, and its impact on academic performance could be attenuating or changing direction in the past decades. It is possible that this phenomenon is associated with computer use—the behavior with the most important increase in our study—for academic purposes. Today, many screen electronic components might be used as tools to help students with homework and diverse school content. Adolescents with greater academic interest may seek access to computers, particularly to have greater availability to information, didactic materials, and others facilitators of the learning process.
In the present study, screen time was greater in adolescents with excellent relationships with parents in crude analyses and when adjusted to screen time at baseline. However, in the final model, an association between relationship with parents and increase in screen time was not found. Perhaps, the enjoyable family environment may increase the time adolescents spend at home; thus, a good relationship with parents might exert stronger influence on adolescent screen time. A Chinese study identified that children who often watched TV with their parents were 2.3 times more likely to spend more screen time than those who seldom watched TV with their parents 
. In addition, findings with Spanish children suggest that parental screen-viewing rules and family co-viewing practices appeared to be predictors of screen viewing 
. The Health Behaviour in School-Aged Children data reported a positive association between screen-based media sedentary behaviors and quality of peer relationships, in all regions studied. In contrast, a negative relationship with quality of family relationships was found 
. In our study, this association did not persist after adjustment for confounders; one explanation is that the involvement of parents in the types of screen activities practiced by adolescents was not asked. Possibly, the engagement of parents in these activities may directly influence overall screen time.
In our study, teens who perceived spending more time indoors compared with peers did not have more screen time than those who perceived spending more time outdoors. Similarly, the feeling of safety about the neighborhood where they live did not affect screen time. Some possible explanations for these findings are that in Brazil, it is common to find commercial enterprises with easy access to screen technologies in neighborhoods regardless of security, such as Internet cafes, video gaming facilities, and other options. The fact that time spent indoors or outdoors was not a strong predictor of screen time could be because screen activities might be found in both environments; young people have access to video games and laptop computers and may transport them to different locations, access the Internet anytime, or even attend houses specializing in video games or other electronics. In Chinese youth, the odds of high screen time were strongly associated with having access to Internet at Internet cafes or at home 
. Few studies have assessed the relationship of these predictors with screen time, and more evidence is necessary.
Our study found no evidence of prediction of change in screen time related to physical activity engagement at age 11 years. Literature consistently points out that screen time has no or just moderate associations with leisure-time physical activity, indicating that they are two different and independent behaviors [11,30]
Alternatively, we found a positive association between change in screen time and leisure-time physical activity at age 15 years, owing mainly to video game playing time. Other researchers have found a relationship between video game playing specifically and leisure-time physical activity [2,31,32]
. However, the explanation is not clear. A hypothesis is that video games featuring sports may encourage playing sports in real life. As pointed out by Marshall et al 
, it is possible that the relationship between screen time and physical activity is multifaceted and dependent of the behavior analyzed.
In the case of BMI, we found the same pattern as seen with physical activity engagement. We did not find evidence to suggest that change in screen time predicts BMI at baseline in the final model. However, a positive association between the change in screen time from age 11 to 15 years and BMI was found. Our results are consistent with other recent longitudinal studies 
and literature reviews [6,7]
, although there are still many controversies [5,34]
. Some hypotheses try to explain the relationship between screen time and body fat (1) obesity itself increases screen time; (2) displacement of physical activities with screen time, resulting in a reduction in total energy expenditure; (3) reduced resting energy expenditure during screen time; and (4) higher consumption of unhealthy food (such as sweets, cakes, and fast foods), especially during TV viewing 
In terms of blood pressure, our study did not find evidence of association, except between change in TV viewing and diastolic blood pressure (data not shown). In a recent review, Tremblay et al 
found that four of nine cross-sectional studies observed a positive association between screen time and blood pressure. Only two longitudinal studies are available in the literature about this subject: one did not find any association 
, and the other did not find association with overall screen time, TV viewing, and computer use, but found a relationship with video game playing 
In summary, screen time increased from early to mid adolescence. This rise was higher among boys and the wealthiest adolescents. Increases in screen time from age 11 to 15 years were related to body composition at age 15 years, with a negative implication on adiposity. Further studies are needed to expand the body of evidence on predictors and health consequences of screen-time change among adolescents.