Spatial and Temporal Modeling of PM2.5 and Infant Morbidity

描述（由申请人提供）：环境PM2.5暴露与婴儿细支气管炎和中耳炎风险之间的潜在关联是一个重要的公共卫生问题。越来越多的毒理学和流行病学证据支持了这一担忧，即环境PM2.5暴露会加剧疾病进程。婴儿毛细支气管炎和中耳炎可能对受影响的婴儿产生严重的长期后果，并且与医疗保健相关的费用是巨大的。然而，现有的研究很少，而且有重要的局限性。本研究将解决许多这些弱点，并对理解环境PM2.5暴露与婴儿细支气管炎和中耳炎之间的关系做出重大贡献。第一个目标是通过用卫星遥感数据补充地面监测数据，与早期研究相比，改进PM2.5暴露评估。不是将婴儿的邮政编码质心与最近的监测仪的数据联系起来，而是使用先进的统计模型来预测婴儿地理编码地址处的PM2.5浓度。与现有研究相比，这将减少暴露错误分类。第二个目标是将PM2.5暴露与来自马萨诸塞州（MA）与医院数据和MA出生缺陷登记处相关的大型纵向出生队列的数据整合起来。有超过40,000例住院和门诊的婴儿细支气管炎和125,000例住院和门诊的中耳炎，分析将有足够的能力来发现小的影响，并检查易感亚群，包括早产儿或已有呼吸或心脏疾病和/或先天性异常的婴儿。链接的PELL数据集允许对短期暴露进行病例交叉分析，并使用兄弟对照进行长期暴露分析，以尽量减少混杂偏倚。传统流行病学研究设计与创新方法的结合将减少通常与基于注册表的分析相关的混杂偏倚。为了检验残留混杂的可能性，第三个目标将使用广义加性模型来检验纳入环境PM2.5暴露并调整已知风险因素后残留风险的时空格局。拟议研究的结果将提供关于婴儿毛细支气管炎和中耳炎的新信息，这两个结果有强烈的迹象表明环境因素，并且在易感婴儿人群中患病率越来越高。对于像环境PM2.5这样广泛的接触，即使很小的关联也会对公众健康产生重大影响。

DESCRIPTION (provided by applicant): The potential association between ambient PM2.5 exposure and the risk of infant bronchiolitis and otitis media is a significant public health concern. This concern is supported by growing toxicologic and epidemiologic evidence that ambient PM2.5 exposure increases potentiation of the disease process. Infant bronchiolitis and otitis media may have serious long-term consequences for the affected infants and the healthcare related costs are substantial. However, available studies are sparse and have important limitations. This study will address many of these weaknesses and make significant contributions to understanding the relationship between ambient PM2.5 exposure and infant bronchiolitis and otitis media. The first aim is to improve PM2.5 exposure assessment compared to earlier studies by supplementing data from ground monitors with remote sensing data from satellites. Instead of linking infant ZIP code centroids to data from the closest monitor, advanced statistical modeling will be used to predict the PM2.5 concentration at the infants' geocoded addresses. This will reduce exposure misclassification compared to existing studies. The second aim is to integrate the PM2.5 exposure with data from a large longitudinal Massachusetts (MA) birth cohort linked to hospital data and the MA Birth Defects Registry. With more than 40,000 cases of inpatient and outpatient infant bronchiolitis and 125,000 cases of inpatient and outpatient otitis media, analyses will have sufficient power to detect small effects as well as examine vulnerable subpopulations including infants born premature or with pre- existing respiratory or cardiac conditions and/or congenital anomalies. The linked PELL dataset allows for case-crossover analyses for short-term exposures and analyses of long-term exposures using sibling controls to minimize confounding bias. The combination of traditional epidemiologic study designs with innovative methods will reduce confounding bias often associated with registry-based analyses. To examine the possibility of residual confounding, the third aim will use generalized additive models to examine the space- time pattern of residual risk after including ambient PM2.5 exposure and adjusting for known risk factors. The results of the proposed study will provide new information on infant bronchiolitis and otitis media, two outcomes for which there is a strong indication of an environmental component and for which there is increasing prevalence in the vulnerable infant population. For an exposure as widespread as ambient PM2.5, even small associations can have a significant public health impact.