Prediction of children's blood lead levels on the basis of household-specific soil lead levels.

To help guide policy decisions about removing lead-contaminated soils, the authors estimated a regression model for predicting a child's blood lead level on the basis of his or her household-specific soil lead level. The data analyzed were blood lead levels (1-45 micrograms/dl) and household-specific soil lead levels (53-20,700 ppm) of 596 children aged 1-5 years who lived in the Helena Valley of Montana and the Silver Valley of Idaho during August 1983. A non-threshold, multiple linear regression model indicated that the estimated mean natural log transformed blood lead level increased by 0.231 micrograms/dl for each unit increase in natural log transformed soil lead level (ppm), after adjusting for the average number of daily outdoor play hours and whether someone in the household smoked. The model predicted that, at a soil lead level of 1,000 ppm, a child who does not play outside and who does not live in a household where someone smokes would be at low risk of lead toxicity (blood lead level between 4 and 24 micrograms/dl).