Hierarchical Analysis of Physical Activity Determinants in Brazilian Adolescents: A Cross-Sectional Study

Background: Physical inactivity during adolescence remains a major public health concern, yet its multifactorial determinants are not fully understood in low- and middle-income settings. Objective: To identify and quantify the determinants of physical activity (PA) among Brazilian adolescents using a hierarchical logistic regression model that reflects the theoretical ordering of distal to proximal factors. Methods: A cross-sectional study was conducted with 856 adolescents (13-19 years). Data were obtained from the validated Brazilian Youth Risk Behavior Survey and the Brazilian Economic Classification Criteria (ABEP) socioeconomic questionnaire. PA was dichotomized according to World Health Organization recommendations. Hierarchical logistic regression examined five theoretical blocks: sociodemographic, anthropometric, substance use, weight/diet, and sedentary behavior. Results: Overall, 5 out of 17 predictors were significant in the full model. extended Body mass index (eBMI) was negatively associated with physical activity (OR = 0.331, 95% CI: 0.169-0.647, p = 0.001), while body mass was positively associated (OR = 2.078, 95% CI: 1.045-4.135, p = 0.037). Working status (OR = 1.235, 95% CI: 1.035-1.475, p = 0.019) and weight loss attempts (OR = 1.327, 95% CI: 1.042-1.690, p = 0.022) increased the odds of being active, whereas current smoking reduced it (OR = 0.715, 95% CI: 0.517-0.990, p = 0.043). Model discrimination improved progressively (AUC = 0.577 to 0.692). Conclusions: Physical activity among Brazilian adolescents was primarily influenced by behavioral and perceptual rather than demographic factors. These findings highlight the need for integrated interventions promoting muscle development, body positivity, and smoking prevention.