The impact of water consumption, point-of-use filtration and exposure categorization on exposure misclassification of ingested drinking water contaminants

The use of population-level indices to estimate individual exposures is an important limitation of previous epidemiologic studies of disinfection by-products (DBPs). We examined exposure misclassification resulting from the use of system average DBP concentrations to estimate individual-level exposures. Data were simulated (n=1000 iterations) for 100 subjects across 10 water systems based on the following assumptions: DBP concentrations ranged from 0-99 microg/L with limited intra-system variability; water intake ranged from 0.5-2.5 L/day; 20% of subjects used bottled water exclusively; 20% of subjects used filtered tap water exclusively; DBP concentrations were reduced by 50% or 90% following filtration. DBP exposure percentiles were used to classify subjects into different exposure levels (e.g., low, intermediate, high and very high) for four classification approaches. Compared to estimates of DBP ingestion that considered daily consumption, source type (i.e., unfiltered tap, filtered tap, and bottled water), and filter efficiency (with 90% DBP removal), 48-62% of subjects were misclassified across one category based on system average concentrations. Average misclassification across at least two exposure categories (e.g., from high to low) ranged from 4-14%. The median classification strategy resulted in the least misclassification, and volume of water intake was the most influential modifier of ingestion exposures. These data illustrate the importance of individual water use information in minimizing exposure misclassification in epidemiologic studies of drinking water contaminants.