Within- and Between-Child Variation in Repeated Urinary Pesticide Metabolite Measurements over a 1-Year Period

BACKGROUND Children are exposed to pesticides from many sources and routes, including dietary and incidental ingestion, dermal absorption, and inhalation. Linking health outcomes to these exposures using urinary metabolites requires understanding temporal variability within subjects to avoid exposure misclassification. OBJECTIVES We characterized the within- and between-child variability of urinary organophosphorus and pyrethroid metabolites in 23 participants of the Children's Pesticide Exposure Study-Washington over 1 year and examined the ability of one to four spot urine samples to categorize mean exposures. METHODS Each child provided urine samples twice daily over 7- to 16-day sessions in four seasons in 2003 and 2004. Samples were analyzed for five pyrethroid and five organophosphorus (OP) metabolites. After adjusting for specific gravity, we used a customized maximum likelihood estimation linear mixed-effects model that accounted for values below the limit of detection to calculate intraclass correlation coefficients (ICC) and conducted surrogate category analyses. RESULTS Within-child variability was 2-11 times greater than between-child variability. When restricted to samples collected during a single season, ICCs were higher in the fall, winter, and spring than in summer for OPs, and higher in summer and winter for pyrethroids, indicating an increase in between-person variability relative to within-person variability during these seasons. Surrogate category analyses demonstrated that a single spot urine sample did not categorize metabolite concentrations well, and that four or more samples would be needed to categorize children into quartiles consistently. CONCLUSIONS Urinary biomarkers of these short half-life pesticides exhibited substantial within-person variability in children observed over four seasons. Researchers investigating pesticides and health outcomes in children may need repeated biomarker measurements to derive accurate estimates of exposure and relative risks.