Microsimulation model of child and adolescent overweight: making use of what we already know

New Zealand has high rates of child overweight and obesity when compared with other countries. Despite an abundance of research documenting the problem, it is unclear what the most effective policy changes or interventions are, and how policy changes might unfold over time within complex systems. We use estimates derived from meta-analyses to create a dynamic microsimulation model of child overweight (including obesity). Using census records we created a synthetic birth cohort of 10,000 children. Information on parental education, ethnicity and father’s socio-economic position at birth were taken from census records. We used the New Zealand Health Survey to estimate population base rates for the prevalence of overweight and obesity. Information on other modifiers (such as maternal smoking, breastfeeding, preterm birth, regular breakfast consumption and so forth) were taken from three birth cohorts: Christchurch Health and Development Study, The Dunedin Multidisciplinary Health and Development Study and the Pacific Islands Families Study. Published intervention studies were used to derive plausible estimates for changes to modifiers. Reducing the proportion of mothers classified as overweight and obesity (−3.31(95% CI −3.55; −3.07) percentage points), reducing the proportion of children watching two or more hours of TV (−3.78(95% CI −4.01; −3.54)), increasing the proportion of children eating breakfast regularly (−1.71(95% CI −1.96; −1.46)), and reducing the proportion of children born with high birth weights (−1.36(95% CI −1.61; −1.11)), lead to sizable decreases in the estimated prevalence of child overweight (including obesity). Reducing the proportion of mothers giving birth by caesarean (−0.23(95% CI −0.49; −0.23)) and increasing parental education (−0.07(95% CI −0.31; 0.18)) did not impact upon child overweight rates. We created a working simulation model of New Zealand children that can be accessed by policy makers and researchers to determine known relationships between predictors and child overweight, as well as potential gains from targeting specific pathways.