Faculty of Civil and Environmental Engineering, Technion, Haifa, Israel; Technion Center of Excellence in Exposure Science and Environmental Health (TCEEH), Technion, Haifa, Israel.
Technion Center of Excellence in Exposure Science and Environmental Health (TCEEH), Technion, Haifa, Israel.
Sci Total Environ. 2017 Feb 15;580:1401-1409. doi: 10.1016/j.scitotenv.2016.12.105. Epub 2016 Dec 27.
Accurate estimation of exposure to air pollution is necessary for assessing the impact of air pollution on the public health. Most environmental epidemiology studies assign the home address exposure to the study subjects. Here, we quantify the exposure estimation error at the population scale due to assigning it solely at the residence place. A cohort of most schoolchildren in Israel (950,000), age 6-18, and a representative cohort of Israeli adults (380,000), age 24-65, were used. For each subject the home and the work or school addresses were geocoded. Together, these two microenvironments account for the locations at which people are present during most of the weekdays. For each subject, we estimated ambient nitrogen oxide concentrations at the home and work or school addresses using two air quality models: a stationary land use regression model and a dynamic dispersion-like model. On average, accounting for the subjects' work or school address as well as for the daily pollutant variation reduced the estimation error of exposure to ambient NO/NO by 5-10ppb, since daytime concentrations at work/school and at home can differ significantly. These results were consistent regardless which air quality model as used and even for subjects that work or study close to their home. Yet, due to their usually short commute, assigning schoolchildren exposure solely at their residential place seems to be a reasonable estimation. In contrast, since adults commute for longer distances, assigning exposure of adults only at the residential place has a lower correlation with the daily weighted exposure, resulting in larger exposure estimation errors. We show that exposure misclassification can result from not accounting for the subjects' time-location trajectories through the spatiotemporally varying pollutant concentrations field.
准确估计空气污染暴露对于评估空气污染对公众健康的影响是必要的。大多数环境流行病学研究将家庭住址暴露分配给研究对象。在这里，我们量化了由于仅在居住地分配而导致的在人群规模上的暴露估计误差。使用了以色列约 95 万名 6-18 岁的大多数学童队列和 38 万名 24-65 岁的以色列成人代表性队列。对于每个研究对象，家庭和工作或学校地址都进行了地理编码。这两个微环境共同构成了人们在大多数工作日期间所在的位置。对于每个研究对象，我们使用两种空气质量模型来估计家庭和工作或学校地址的环境氮氧化物浓度：一个是固定的土地利用回归模型，另一个是动态类似扩散的模型。平均而言，考虑到研究对象的工作或学校地址以及每日污染物变化，可以将暴露于环境 NO/NO 的估计误差降低 5-10ppb，因为工作/学校和家庭的日间浓度可能有很大差异。这些结果在使用哪种空气质量模型时都是一致的，甚至对于那些在家庭附近工作或学习的研究对象也是如此。然而，由于他们通常的通勤时间较短，仅将学童的暴露分配在其居住地进行估计似乎是合理的。相比之下，由于成年人的通勤距离较长，仅将成年人的暴露分配在居住地与每日加权暴露的相关性较低，从而导致更大的暴露估计误差。我们表明，由于没有考虑到研究对象在时空变化的污染物浓度场中的时间-位置轨迹，可能会导致暴露分类错误。
