As a sensitivity analysis, we explored the potential
for confounding by the remaining factors available in our dataset including
socio-economic status (education level (primary, technical, secondary,
university)), and physical activity (moderately inactive, moderately
active, active).