Childhood exposure to lead can result in increased blood lead levels, which can have detrimental health effects. Models that can predict blood lead level based on exposure to floor and window sill dust lead at varying concentrations are useful tools for informing policy decisions. We compare an empirical model relating blood lead in children to the lead content of household dust, with a mechanistic model that simulates lead exposure in children through their daily activities. The empirical model is constructed using 1999–2004 NHANES data, while the mechanistic model is validated using 2011–2016 NHANES data. The models are evaluated at dust loadings ranging from 0.18 – 43 µg/ft2 and the resulting blood leads are compared. Specific demographic variables are chosen to simulate individuals are generated with “typical,” “high end,” and “low end” exposures in the empirical model; these are directly compared with the median, 95th, and 5th percentile results from the mechanistic model. For a child with “typical” lead exposure, the 95 % confidence interval for computed blood lead values contained the median blood lead value from the mechanistic model at all dust lead exposure levels considered. Similar results hold when comparing the 95th percentile to “high end” and 5th percentile to “low end”. The blood lead level computed by the mechanistic model increases more quickly with respect to dust loading than the empirical model, particularly at higher loadings. The empirical and mechanistic models generate similar blood lead distributions across the range of dust loadings considered, particularly at low loadings. The mechanistic model may be more appropriate for higher loadings due to limitations inherit in the structure of the empirical model.
Read full abstract