Abstract
This study aimed to evaluate the association of longitudinal nutrient patterns with body composition in a cohort of 132 black South African middle-aged women over five years. Nutrient patterns were identified using principal component analysis at baseline and follow-up 5 years later. Associations between nutrient patterns and repeated body composition measures were evaluated using generalized estimating equations, before and after adjusting for baseline education and repeated measures of age, socio-economic status, physical activity and employment. The animal-driven nutrient pattern was associated with increases in repeated measures of visceral adipose tissue (VAT) (β coefficient, 5.79 [95% CI, 0.01-11.57] cm(2)), fat mass index (FMI) (0.47 [0.01-0.93] kg·m(-2)) and lean mass index (LMI) (0.50 [0.18-1.17] kg·m(-2)) (p < 0.05) after adjustment. Vitamin C, sugar, and potassium-driven nutrient pattern was associated with higher FMI (0.50 [0.12-0.88] kg·m(-2)) and LMI (0.58 [0.07-1.10] kg·m(-2)) before and after adjustment (p < 0.05). These findings suggest that dietary interventions to curb obesity in black middle-aged South African women should focus on attenuation of nutrient patterns centred on added sugar, animal fat and animal protein.