Population Studies and Surveillance, Cancer Care Ontario, Toronto, Ontario, Canada M5G 2L7.
J Agric Food Chem. 2010 Sep 22;58(18):10253-61. doi: 10.1021/jf101209g.
Epidemiologic studies of pesticides have been subject to important biases arising from exposure misclassification. Although turf applicators are exposed to a variety of pesticides, these exposures have not been well characterized. This paper describes a repeated measures study of 135 TruGreen applicators over three spraying seasons via the collection of 1028 urine samples. These applicators were employed in six cities across the United States. Twenty-four-hour estimates (μg) were calculated for the parent compounds 2,4-D, MCPA, mecoprop, dicamba, and imidacloprid and for the insecticide metabolites MPA and 6-CNA. Descriptive statistics were used to characterize the urinary levels of these pesticides, whereas mixed models were applied to describe the variance apportionment with respect to city, season, individual, and day of sampling. The contributions to the overall variance explained by each of these factors varied considerably by the type of pesticide. The implications for characterizing exposures in these workers within the context of a cohort study are discussed.
农药的流行病学研究受到了暴露分类错误带来的重要偏差的影响。尽管草坪施用者会接触到多种农药，但这些暴露情况并未得到很好的描述。本文通过收集 1028 份尿液样本，对 135 名特绿草坪施用者在三个喷洒季节进行了重复测量研究。这些施用者分布在美国六个城市。计算了母体化合物 2,4-D、MCPA、麦草畏、二甲四氯、草甘膦和噻虫啉以及杀虫剂代谢物 MPA 和 6-CNA 的 24 小时估计值（μg）。描述性统计用于描述这些农药的尿液水平，而混合模型则用于描述与城市、季节、个体和采样日有关的方差分配。这些因素对总体方差的解释贡献因农药类型而异。本文讨论了在队列研究背景下描述这些工人暴露情况的意义。
