van Wendel de Joode Berna, van Hemmen Joop J, Meijster Tim, Major Vicky, London Leslie, Kromhout Hans
Department of Chemical Exposure Assessment, TNO Chemistry, 3700 AJ Zeist, The Netherlands.
J Expo Anal Environ Epidemiol. 2005 Jan;15(1):111-20. doi: 10.1038/sj.jea.7500369.
Valid and reliable semi-quantitative dermal exposure assessment methods for epidemiological research and for occupational hygiene practice, applicable for different chemical agents, are practically nonexistent. The aim of this study was to assess the reliability of a recently developed semi-quantitative dermal exposure assessment method (DREAM) by (i) studying inter-observer agreement, (ii) assessing the effect of individual observers on dermal exposure estimates for different tasks, and (iii) comparing inter-observer agreement for ranking of body parts according to their exposure level. Four studies were performed in which a total of 29 observers (mainly occupational hygienists) were asked to fill in DREAM while performing side-by-side observations for different tasks, comprising dermal exposures to liquids, solids, and vapors. Intra-class correlation coefficients ranged from 0.68 to 0.87 for total dermal exposure estimates, indicating good to excellent inter-observer agreement. The effects of individual observers on task estimates were estimated using a linear mixed effect model with logged DREAM estimates as explanatory variable; "task", "company/department", and the interaction of "task" and "company/department" as fixed effects; and "observer" as a random effect. Geometric mean (GM) dermal exposure estimates for different tasks were estimated by taking the exponent of the predicted betas for the tasks. By taking the exponent of the predicted observer's intercept (exp(omega i)), a multiplier (M(O)) was estimated for each observer. The effects of individual observers on task estimates were relatively small, as the maximum predicted mean observers' multiplier was only a factor 2, while predicted GMs of dermal exposure estimates for tasks ranged from 0 to 1226, and none of the predicted individual observers' multipliers differed significantly from 1 (t-test alpha = 0.05). Inter-observer agreement for ranking of dermal exposure of nine body parts was moderate to good, as median values of Spearman correlation coefficients for pairs of observers ranged from 0.29 to 0.93. DREAM provides reproducible results for a broad range of tasks with dermal exposures to liquids, solids, as well as vapors. DREAM appears to offer a useful advance for estimations of dermal exposure both for epidemiological research and for occupational hygiene practice.
适用于不同化学制剂、用于流行病学研究和职业卫生实践的有效且可靠的半定量皮肤暴露评估方法实际上并不存在。本研究的目的是通过以下方式评估一种最近开发的半定量皮肤暴露评估方法（DREAM）的可靠性：（i）研究观察者间的一致性；（ii）评估个体观察者对不同任务的皮肤暴露估计值的影响；（iii）比较观察者间根据身体部位暴露水平进行排名的一致性。进行了四项研究，共邀请了29名观察者（主要是职业卫生学家）在对不同任务（包括皮肤接触液体、固体和蒸气）进行并排观察时填写DREAM。总皮肤暴露估计值的组内相关系数范围为0.68至0.87，表明观察者间的一致性良好至优秀。使用线性混合效应模型估计个体观察者对任务估计值的影响，该模型以记录的DREAM估计值作为解释变量；“任务”、“公司/部门”以及“任务”与“公司/部门”的交互作用作为固定效应；“观察者”作为随机效应。通过取任务预测β值的指数来估计不同任务的几何平均（GM）皮肤暴露估计值。通过取预测观察者截距的指数（exp(ωi)），为每个观察者估计一个乘数（M(O)）。个体观察者对任务估计值的影响相对较小，因为预测的观察者平均乘数最大值仅为2倍，而任务的皮肤暴露估计值的预测GM范围为0至1226，且预测的个体观察者乘数均与1无显著差异（t检验α = 0.05）。九个体表部位皮肤暴露排名的观察者间一致性为中等至良好，因为观察者对的Spearman相关系数中位数范围为0.29至0.93。DREAM为广泛的涉及皮肤接触液体、固体以及蒸气的任务提供了可重复的结果。DREAM似乎为流行病学研究和职业卫生实践中的皮肤暴露估计提供了有益的进展。
