Spatial prediction of blood lead levels in children in Toledo, OH using fuzzy sets and the site-specific IEUBK model

• Blood lead was predicted and mapped using the USEPA IEUBK model and fuzzy sets.
• Spatial variables influencing blood lead levels were evaluated and weighted.
• Elevated blood lead levels are strongly linked with older housing.
• Predicted blood lead levels decrease moving away from urban to rural areas.

Lead poisoning in young children remains a concern in many urban areas, even 30 years after the banning of Pb-based paint and leaded gasoline. In 2010, 16.6% of 6550 children tested in Toledo, OH had blood lead levels (BLLs) above the Centers for Disease Control (CDC) lead poisoning reference level of 5 μg/dL. The aim of this study was to spatially quantify the risk of lead poisoning to children in Toledo, OH and to reduce lead poisoning risk through educational outreach and citizen science.The educational outreach component of the study was implemented through the sampling strategy. Students in Toledo area schools were instructed on proper USEPA soil sampling guidelines and were asked to collect soil samples from their residential yards. A subsample of 81 soils was analyzed for total lead and bioavailable lead. Site-specific total lead and bioavailability data were used in the USEPA Integrated Exposure Uptake and Biokinetic (IEUBK) model to predict BLLs. Predicted BLLs were then used to extract expert knowledge for development of an index model using analytical hierarchy process (AHP) and weighted linear combination (WLC). Expert knowledge was used to standardize predictor datasets and to produce continuous BLL risk for each predictor variable. Fuzzy sets were implemented in the model to account for uncertainties in the sampling method.It was found that 8.6% of sampled sites had total soil lead concentrations above the USEPA action level of 400 mg/kg, but 28.4% of soil samples yielded predicted elevated BLLs, suggesting the action level is set too high. The spatial variables influencing risk of lead poisoning from most important to least important were age of housing, road density, percent impervious surfaces, home value, household income and soil type. The highest risk for lead poisoning was in the city center and decreased moving towards the rural areas surrounding the city. The index model paired with the unique outreach driven sampling approach proved successful both at providing quality soil samples and educating the community about the ongoing risks of lead in soil.