Which environmental sample type and statistical approach provides most information about metal exposure in young...

Identification of sources and pathways is critical in minimizing exposure of young children to toxic materials. We monitored 108 children <5 years old 6-monthly for up to 5 years in a major urban setting. Samples (ñ7000) included interior handwipes (W1) and after exterior playing (W2), interior house dust (PD1) and day care dust (PD2) using petri dishes, exterior dust-fall accumulation, exterior dust sweepings, garden soil, blood and urine. Here we describe multi-element results to determine which sampling method and analysis of the data provide the most reliable indicators of metal exposure to young children. Samples were analysed by ICPMS for Ca, Cd, Cr, Cu, Fe, Mg, Mn, Ni, Pb, Ti, V and Zn. Pearson Correlations showed the highest number of significant correlations are for: W1 and W2, dust sweepings and soil. Mixed model analyses (MMA) for the blood levels as the dependent variable and environmental predictor variables showed the most consistent results were for W1, PD1 and sweepings. MMA to investigate the association between each metal (e.g. Ca) and the other 11 metals showed the largest numbers of significant relationships are for W1 and sweepings. Cluster analyses showed that the 'best' clusters in W1 and W2 are for Fe-Zn-Mg and Mn-Pb-Ni. For PD1 and PD2 the 'best' clusters were Fe-Zn-Mg, Cr-Ni-Ti, and Cu-Mn-Pb. Clusters for dust sweepings and soil are generally similar. Principal component analysis (PCA) loadings for W1 and PD1 accounted for >50% of the variance. Metals comprising loading 1 component for both sample types included Ca, Fe, Mg, and Mn. Overall cluster analyses provided more information than PCA loadings. In summary: The most suitable methods and analyses are MMA of W1 and sweepings, and cluster analyses of W1and PD1. Resuspension from outdoor surfaces and soils and deposition in the residences is a likely pathway for most metals.