Resolving intertracer inconsistencies in soil ingestion estimation.

In this article we explore sources and magnitude of positive and negative error in soil ingestion estimates for children on a subject-week and trace element basis. Errors varied among trace elements. Yttrium and zirconium displayed predominantly negative error; titanium and vanadium usually displayed positive error. These factors lead to underestimation of soil ingestion estimates by yttrium and zirconium and a large overestimation by vanadium. The most reliable tracers for soil ingestion estimates were aluminum, silicon, and yttrium. However, the most reliable trace element for a specific subject-day (or week) would be the element with the least error during that time period. The present analysis replaces our previous recommendations that zirconium and titanium are the most reliable trace elements in estimating soil ingestion by children. This report identifies limitations in applying the biostatistical model based on data for adults to data for children. The adult-based model used data less susceptible to negative bias and more susceptible to source error (positive bias) for titanium and vanadium than the data for children. These factors contributed significantly to inconsistencies in model predictions of soil ingestion rates for children. Correction for error at the subject-day level provides a foundation for generation of subject-specific daily soil ingestion distributions and for linking behavior to soil ingestion.