Fig. 1.

Association of dietary patterns with urbanization, physical inactivity prevalence (%) and per capita gross domestic product (GDP). Dietary patterns were identified using k-means cluster analysis. A description of cluster analysis is provided in the methods section. GDP values were divided by a fixed factor (1000) to allow comparison with the other two groups. Urbanization, physical inactivity prevalence and per capita GDP were all statistically different (p < 0.001) between the three dietary patterns. Univariate analysis of variance (ANOVA) was used and variables were transformed if not normally distributed.