Determinants of changes in physical activity levels in late adolescence; prospective analysis in urban communities

Purpose: Physical activity levels (PA-levels) significantly decline during adolescence, and sport participation during childhood and adolescence is frequently emphasized as protec-tive factors of PA-decline. However, there is a lack of studies which specifically examined sport-related factors and its influence on changes in PA (PA-changes) in adolescence. This study aimed to prospectively observe sport factors as: (i) correlates of PA-levels and (ii) pre-dictors of PA-changes in the period between 16 and 18 years of age among urban adoles-cents from Bosnia and Herzegovina. Methods: The sample of participants comprised 324 adolescents (44% females) who were prospectively observed over two testing waves: (i) baseline, when participants were 16 years old; and (ii) follow-up, 20 months later (18 years of age). The variables were collected by previously validated questionnaires including questions on predictors (sociodemographic variables and various sport factors [current/former/ever participation in individual and team sports, experience in sports, competitive result achieved]), and criteria (PA level obtained at study baseline and follow up, measured by Physical Activity Questionnaire for Adolescents [PAQ-A], and difference between PA-levels at baseline and follow-up). The t-test was used to compare PA-levels. The associations between variables were evidenced by: (i) Spear-man’s rank order correlations (between predictors and PA-levels), and (ii) logistic regression analysis (between predictors, and PA-changes observed as binomial criterion [PA-incline vs. PA-decline] – excluding those participants who reported active sport participation at study baseline). Results: The PA-level significantly declined over the study course (t-test: 6.60, p < 0.01). Sport-related predictors were significantly associated with PA at baseline (Spearman’s R: 0.33–0.45, p < 0.01), and PAat follow-up (Spearman’s R: 0.32-0.45, p < 0.01). Meanwhile, there was no significant correlation between studied predictors and differences in PA-levels between baseline and follow-up. Also, logistic regression did not reveal any significant influ-ence of predictors obtained at study baseline and PA-changes observed as binomial criterion (PA-incline vs PA-decline). Conclusion: While studied sport-related predictors significantly influence the PA-levels in the age of 16 and 18, with the higher level of PA among those adolescents who are actively in-volved in sports, sport-participation do not predict changes in PA-levels over the observed period of life. Knowing the influence of PAon overall health status, future studies should pro-vide additional details on possible predictors of PA-changes in adolescence.