A group of researchers is interested in investigating the effect of maternal exposure to manganese at high doses on child neuro-development in a sample of mother-infant pairs in Bangladesh. Bangladesh is experiencing an epidemic of metals and heavy metals poisoning. The soil of the country is contaminated with arsenic and manganese. Occupational and environmental exposure to lead is also a concern. The levels of arsenic, manganese, and lead found in Bangladesh can be extremely high and the support of these environmental toxicants poorly overlaps with what observed in Europe and in the USA.

In a sample of 400 mother infant pairs, the investigators measured child cognitive development (Y) at 24 months using the Bayley scales for infant neurodevelopment. Furthermore, they obtained a measure of concentrations of manganese (X1, mu/dl), the primary exposure of interest, as well as of arsenic (X2, mu/dl) and lead (X3, mu/dl) in cord blood. This measure reflects maternal exposure to these metals during pregnancy. An extensive questionnaire was filled at baseline by the mothers to collect information on potential confounders of the manganese-cognitive development relationship. These include child age at testing (Z1, months), mother education (Z2, years), mother age (Z3, centered at the mean), mother IQ (Z4, z-scored), home quality score (Z5), daily nutritional intake of egg (Z6), meat (Z7), fish (Z8), and maternal smoking status (Z9). The data is generated so that observations are be independent of each other and identically distributed (conditional on covariates). Under a certain model selection NUCA will hold. Note that not all covariates will be true confounders.