Mixed-effect models for evaluating multiple measures of atrazine exposure among custom applicators

The exposure of custom (or commercial) applicators to the herbicide atrazine was measured in environmental (hand wash and dermal patch) and biological (urine and saliva) samples. Surrogate exposure data, such as amount of atrazine sprayed, were also collected. A systematic sampling design was used that included both spray and nonspray days. Fifteen applicators were sampled 5 to 7 days each during a 6-week spring spray season for a total of 89 sampled days. Mixed-effect regression modeling was used to examine the relationship among the surrogate, environmental, and biological atrazine exposure measures. Surrogate measures of atrazine application (either kg of atrazine sprayed or spray atrazine [yes/no]) were significantly associated with increased levels of atrazine or atrazine equivalents (eq) in hand wash, thigh patch, 4-6 p.m. saliva, and 24-hour urine samples. Two days of spraying information (day of sampling and day before sampling) were needed to optimally estimate atrazine biomarkers in the biological samples, whereas only 1 day of spraying information (day of sampling) was needed to estimate atrazine levels in the environmental samples. Thigh and hand atrazine exposures were significantly associated with increased atrazine and atrazine eq. levels in the 4-6 p.m. saliva and 24-hour urine samples, respectively. Levels of 4-6 p.m. salivary atrazine were also significantly associated with increased levels of 24-hour urinary atrazine eq. Atrazine levels in the 4-6 p.m. saliva samples tracked most closely with evening and next morning urinary atrazine eq. Number of days into the study at the time of sample collection predicted urinary and salivary atrazine levels independent of other fixed effects. These results indicate that either surrogate, environmental, or biological exposure measures can be used in appropriately specified models to estimate urinary and salivary atrazine biomarker levels.