Testing ecological and universal models of body shape and child health using a global sample of infants and young children

Aim: To test whether a risk of child illness is best predicted by deviations from a population-specific growth distribution or a universal growth distribution. Subjects and methods: Child weight for height and child illness data from 433 776 children (1–59 months) from 47 different low and lower income countries are used in regression models to estimate for each country the child basal weight for height. This study assesses the extent to which individuals within populations deviate from their basal slenderness. It uses correlation and regression techniques to estimate the relationship between child illness (diarrhoea, fever or cough) and basal weight for height, and residual weight for height. Results: In bivariate tests, basal weight for height z-score did not predict the country level prevalence of child illness (r² = –0.01, n = 47, p = 0.53), but excess weight for height did (r² = 0.14, p < 0.01). At the individual level, household wealth is negatively associated with the odds that a child is reported as ill (beta = –0.04, p < 0.001, n = 433 776) and basal weight for height was not (beta = 0.20, p = 0.27). Deviations from country-specific basal weight for height were negatively associated with the likelihood of illness (beta = –0.13, p < 0.01), indicating a 13% reduction in illness risk for every 0.1 standard deviation increase in residual weight-for-height Conclusion: These results are consistent with the idea that populations may differ in their body slenderness, and that deviations from this body form may predict the risk of childhood illness.