Cardiometabolic Risk Factors and Physical Activity Patterns Maximizing Fitness and Minimizing Fatness Variation in Malaysian Adolescents: A Novel Application of Reduced Rank Regression

Patterns of physical activity (PA) that optimize both fitness and fatness may better predict cardiometabolic health. Reduced rank regression (RRR) was applied to identify combinations of the type (e.g., football vs. skipping), location and timing of activity, explaining variation in cardiorespiratory fitness (CRF) and Body Mass Index (BMI). Multivariable regressions estimated longitudinal associations of PA pattern scores with cardiometabolic health in n = 579 adolescents aged 13-17 years from the Malaysian Health and Adolescent Longitudinal Research Team study. PA pattern scores in boys were associated with higher fitness (r = 0.3) and lower fatness (r = -0.3); however, in girls, pattern scores were only associated with higher fitness (r = 0.4) (fatness, r = -0.1). Pattern scores changed by β = -0.01 (95% confidence interval (CI) -0.04, 0.03) and β = -0.08 (95% CI -0.1, -0.06) per year from 13 to 17 years in boys and girls respectively. Higher CRF and lower BMI were associated with better cardiometabolic health at 17 years, but PA pattern scores were not in either cross-sectional or longitudinal models. RRR identified sex-specific PA patterns associated with fitness and fatness but the total variation they explained was small. PA pattern scores changed little through adolescence, which may explain the limited evidence on health associations. Objective PA measurement may improve RRR for identifying optimal PA patterns for cardiometabolic health.