Comportamentos de movimento de 24 horas, inequalidades e disparidades na universidade: evidências preditivas e suas implicações para a formação de estudantes em regiões economicamente desfavorecidas

Abstract

Objective: To test the predictive capacity of inequalities, social disparities, and academic context to identify healthier 24-hour movement behavior patterns in university students from economically disadvantaged regions.Methodology: We analyzed data from 407 baseline students of the longitudinal project “24-hour movement behavior and metabolic syndrome study” (24h- MESYN), conducted at two universities, a private one in Imperatriz, Maranhão, and a public one in Miracema, Tocantins, both located in economically disadvantaged regions with a Gini index ≤ 0.56. We collected information through an online questionnaire, including inequalities (age, biological sex), social disparities (race/ethnicity, maternal education, family and individual income), and 24-hour movement behaviors (physical activity, sedentary behavior, and sleep), converted into minutes per day. We constructed clusters based on nine behaviors using the k- medians method, identifying four clusters (Calinski-Harabasz index = 495.93). We applied multinomial logistic regression to test associations between exposures and clusters, using Cluster 2 (healthiest profile) as the baseline outcome. We also assessed the predictive accuracy of these variables in identifying Cluster 2 (target outcome). Results: Considering Cluster 2 as a base, we observed significant associations with Cluster 1 for city of residence in other locations of Maranhão, Tocantins or Pará (OR, 11.22; 95%CI, 1.20–105.29), non-white race/ethnicity (OR, 5.68; 95%CI, 1.60–20.18), individual monthly income of 1 to 2 minimum wages (OR, 0.19; 95%CI, 0.04–0.82) and morning/full-time shift (OR, 4.82; 95%CI, 1.15–20.12). Compared with Cluster 2, Cluster 3 was also associated with non-White race/ethnicity (OR, 3.34; 95% CI, 1.06– 10.53) and morning/full-time shift (OR, 4.49; 95% CI, 1.12–17.93). The variables retained in the multinomial model showed acceptable accuracy (AUROC = 0.785) in identifying Cluster 2. Inequalities, social disparities, and academic context showed high sensitivity at low cutoff points (<0.10), and a balance between sensitivity (75.0%) and specificity (75.0%) in the ranges of 0.15 to 0.20. Conclusion: Inequalities, social disparities, and academic context are associated with 24- hour movement behaviors and have predictive power to identify students with healthier patterns of these behaviors. The variables city of residence, race/ethnicity, individual personal income, and course shift offer relevant operational accuracy, with a cutoff point between 0.15 and 0.20 providing the best balance between sensitivity and specificity for screening purposes.