Looking for Lead in all the wrong places?
An examination of the Environmental Protection Agency's
Integrated Exposure Uptake Biokinetic Model
using Rhode Island case data.

Cassandra Stubbs

Lead poisoning effects one out of every six children in the country. Recently, attention has been devoted to the effects of lead on increasingly lower levels of exposure, heightening the sense of importance of the issue. Additionally, the multiple pathways of exposure to lead have become a major focus of interest in an effort to explain the major mechanisms that children are poisoned. In 1994 the Environmental Protection Agency (EPA) released the Integrated Exposure Uptake Biokinetic Model for Lead in Children (IEUBK Model), a computer model intended to predict blood lead levels for children 6 months to 7 seven years using exposure and intake data. The model is important because it allows the scientific and regulatory communities the capability to determine levels of lead in the soil, dust, water, or air that would likely lead to blood lead levels. Although the Model was released in 1994, the Model has not been validated with field studies.

For my senior thesis, I applied Rhode Island case data to the Model to compare children's actual blood lead levels with levels predicted by the Model. All children included in the project had received home environmental lead inspections by the Rhode Island Department of Health because they had blood lead levels equal to or above 25 mg/dl. As part of the inspection process, water, dust, and soil samples are gathered and analyzed for lead concentrations. I matched the environmental lead concentration data with the children's blood lead level data and regional lead in air concentrations for input into the model. In addition to applying the case data to the model, I analyzed exposure data, blood lead data, and demographic data using regression analysis.

Comparisons of children's actual blood lead levels with the blood lead levels predicted by the IEUBK Model found that the model underpredicted children's blood lead levels for 82% of the cases. For over half the children, the model predicted a blood lead level 20 mg/dl lower than their actual blood lead level (the definition for blood lead poisoning is 10 mg/dl). The analysis of the IEUBK Model suggests the need for further investigation of the model's ability to predict blood lead level's accurately for case data. Regression analysis of the data included in the model as well as data about the children's race and gender found that no variables correlated strongly with higher blood lead levels, although a weak correlation was found between gender and blood lead levels. The primary explanation for not finding stronger associations is the lack of a control group (children with low blood lead levels) and the individual variability among lead poisoned children.