Exposure misclassification in studies of agricultural pesticides - Insights from biomonitoring
Exposure misclassification in studies of agricultural pesticides - Insights from biomonitoring
复制标题
DOI：
10.1097/01.ede.0000190603.52867.22
复制
发表时间：
2006-01-01
期刊：
影响因子：
5.4
通讯作者：
Gustin, C
中科院分区：
文献类型：
--
作者：
Acquavella, JF;Alexander, BH;Gustin, C
背景：流行病学家经常通过询问参与者特定农药的使用和相关的工作实践来评估终身农药暴露。最近，Dosemeci和他的同事提出了一种算法，可以根据调查问卷信息估计一生中的平均暴露强度。我们对使用草甘膦(n=48)、2，4-D(n=34)或毒死蜱(n=34)的农民的尿液农药浓度进行了评估。方法：根据训练有素的现场观察员和农民对应用条件的评价，分别计算算法分数。统计分析包括非参数相关性、分类一致性评估和暴露分布的分类评估。结果：根据现场观察员的评估，算法得分与草甘膦和2，4-D的尿药浓度之间存在中等相关性(r=0.47；95%可信区间[CI]=0.21~0.66)和2，4-D(0.45；0.13~0.68)。当算法得分基于参与者的自我报告时，相关性较低(对于草甘膦，r=0.23[CI=-0.07到0.48]；对于2，4-D，r=0.25[-0.10到0.54])。对于毒死蜱，根据观察者和参与者的输入，液体(0.42；0.01至0.70)和颗粒剂(-0.44；-0.83至0.29)存在相反的相关性。3种农药在分类分析中的符合率在20%到44%之间，暴露分布在不同类别之间存在明显的重叠。结论：我们的结果证明了收集类型的农药制剂的重要性，并表明通用暴露评估可能导致许多农药的暴露错误分类。
Background: Epidemiologists often assess lifetime pesticide exposure by questioning participants about use of specific pesticides and associated work practices. Recently, Dosemeci and colleagues proposed an algorithm to estimate lifetime average exposure intensity from questionnaire information. We evaluated this algorithm against measured urinary pesticide concentrations for farmers who applied glyphosate (n = 48), 2,4-D (n = 34), or chlorpyrifos (n = 34).Methods: Algorithm scores were calculated separately based on trained field observers' and farmers' evaluations of application conditions. Statistical analyses included nonparametric correlations, assessment of categorical agreement, and categorical evaluation of exposure distributions.Results: Based on field observers' assessments, there were moderate correlations between algorithm scores and urine concentrations for glyphosate (r = 0.47; 95% confidence interval [CI] = 0.21 to 0.66) and 2,4-D (0.45; 0.13 to 0.68). Correlations were lower when algorithm scores were based on participants' self-reports (for glyphosate, r = 0.23 [CI = -0.07 to 0.48]; for 2,4-D, r = 0.25 [-0.10 to 0.54]). For chlorpyrifos, there were contrasting correlations for liquid (0.42; 0.01 to 0.70) and granular formulations (-0.44; -0.83 to 0.29) based on both observers' and participants' inputs. Percent agreement in categorical analyses for the 3 pesticides ranged from 20% to 44%, and there was appreciable overlap in the exposure distributions across categories.Conclusions: Our results demonstrate the importance of collecting type of pesticide formulation and suggest a generic exposure assessment is likely to result in appreciable exposure misclassification for many pesticides.