Aim:

We intend to quantify the impact of diet on children's BMI z-scores. Instead of classifying food consumption based on nutritional components or dietary pattern, we propose examining individual foods or food categories following the method by Mozzafarian et al (2011, NEJM). By tracking dietary consumption from age 2 to age 13, we aim at assessing whether higher dietary intake of certain food items or categories is associated with higher BMI z-scores, thus increasing propensity of childhood obesity.

Hypotheses:

1. Using a multivariate regression model on first differences of BMI z-scores, more frequent consumption of potato chips, potatoes, sugar-sweetened beverages, unprocessed red meat and processed meats is positively associated with BMI z-score increase; whereas more frequent consumption of vegetables, fruits, whole grains, nuts, and yogurt is negatively associated with BMI z-score increase.

2. When using a fixed-effect (FE) model, hypothesis1 still holds.

3.Using a quantile regression, the impact of diet on BMI z-score is stronger in the upper BMI z-score quantile.

4.Using a dynamic panel data model, an induction period can be identified. The effect of diet on weight is more significant after the induction period.

Exposure variables:

Food consumption for each food category

Outcome variable:

BMI z-score (calculated using weight, height, gender and age)

Confounding variables:

Socioeconomic measures such as ethnicity and family income, lifestyle behaviours including diet control, TV viewing, physical activity, tobacco use, and sleep duration.

References:

Mozaffarian D, Hao T, Rimm EB, Willett WC, Hu FB. Changes in diet and lifestyle and long-term weight gain in women and men. New England Journal of Medicine 2011;364:2392-2404.