Rondeau Virginie, Berhane Kiros, Thomas Duncan C
Institut National de la Santé et de la Recherche Médicale, Equipe Mixte INSERM E03 38 (Biostatistique), Université Bordeaux2, 146 rue Léo Saignat, 33076 Bordeaux Cedex, France.
Stat Med. 2005 Apr 15;24(7):1103-15. doi: 10.1002/sim.1980.
A three-level model is proposed to simultaneously examine the effects of daily exposure to air pollution and individual risk factors on health outcomes without aggregating over subjects or time. We used a logistic transition model with random effects to take into account heterogeneity and overdispersion of the observations. A distributed lag structure for pollution has been included, assuming that the event on day t for a subject depends on the levels of air pollution for several preceding days. We illustrate this proposed model via detailed analysis of the effect of air pollution on school absenteeism based on data from the Southern California Children's Health Study.
提出了一种三级模型，用于同时检验每日接触空气污染和个体风险因素对健康结果的影响，而无需对个体或时间进行汇总。我们使用了具有随机效应的逻辑转换模型来考虑观测值的异质性和过度分散性。已纳入污染的分布滞后结构，假设个体在第t天的事件取决于前几天的空气污染水平。我们基于南加州儿童健康研究的数据，通过对空气污染对学校缺勤率影响的详细分析来说明这个提出的模型。
