Air Pollutants and Cardiovascular Risk: Investigating Thresholds with Pooled Cohorts and Electronic Health Records

项目摘要 长期接触空气污染物被认为是心血管疾病的危险因素； 环境中的细颗粒物(PM2.5)被认为是导致过早死亡的十大全球风险因素之一 发病率主要是由于心血管疾病的影响。在国家和全球一级的公共卫生行动仍然需要更好的 关于污染物和心血管疾病之间联系的性质的信息：是否有接触的门槛， 以下哪项效果不会发生？这种影响是线性的，还是遵循一种更具生物学意义的功能？ 这项建议通过两种方法解决这些关键问题：1)汇集高质量的队列研究； 2)基于电子健康档案(EHR)分析健康系统档案。前者的方法是 流行病学标准，而后者代表“大数据”的未来。两种语言的结合与比较 同一研究计划中的这两种方法为解决另一个关键问题提供了独特的机会 公共卫生问题：“大数据”能提供与传统队列方法相同的答案吗？ 这项建议使用最先进的暴露评估来解决低水平空气污染对健康的影响，精细- 浓度(PM2.5、氮氧化物和臭氧)和高级统计的尺度混合模拟 方法：研究方法。队列允许最大限度地减少由于混合和错误分类而产生的偏差 信息来自美国一组久负盛名的队列，总计近100万名参与者，每个人都有 适当的结果、家庭住址和个人层面的协变量细节。财团队列的特点是 地理和曝光的多样性。本项目还将研究一种非同寻常的特征良好的大规模 综合健康提供系统，Kaiser Permanente North California，详细的EHR超过4.7 百万会员。这项研究的结果将为指导美国的政策提供至关重要的知识 国家和全球。此外，这项建议将直接比较传统队列和新的“大数据” 回答复杂的流行病学研究问题的方法，允许更好地理解 “大数据”取代和/或补充传统方法的能力。

Project Summary Long-term exposure to air pollutants is recognized as a risk factor for cardiovascular disease (CVD); exposure to ambient fine particulates (PM2.5) is considered among the top ten global risk factors for premature death and morbidity primarily due to CVD effects. Public health action on a national and global level still requires better information on the nature of the association between pollutants and CVD: Is there a threshold for exposure, below which effects do not occur? Is the effect linear or does it follow a more biologically likely function? This proposal addresses these critical questions with two approaches: 1) pooling high-quality cohort studies; and 2) analyzing health system records based on electronic health records (EHR). The former approach is the epidemiological standard while the latter represents the “big data” future. The combination and comparison of these two approaches in the same research program affords a unique opportunity to address another key public health question: Can “big data” provide the same answer as traditional cohort approaches? This proposal addresses low-level air pollution health effects using state-of-the-art exposure assessment, fine- scale hybrid modeling of concentrations (PM2.5, oxides of nitrogen, and ozone) and advanced statistical methods. The cohorts permit optimal minimization of bias due to confounding and misclassification by pooling information from a set of well-established cohorts in the US totaling nearly one million participants, each with appropriate outcome, home address, and individual level covariate detail. The consortium cohorts feature geographic and exposure diversity. This project will also study one unusually well-characterized large-scale integrated health delivery system, Kaiser Permanente Northern California, with detailed EHR on more than 4.7 million members. Results of this study will provide critically important knowledge to guide policy in the United States and globally. Further, this proposal will directly compare traditional cohort and new “big data” approaches for answering complex epidemiological research questions, allowing for better understanding of the ability of “big data” to replace and/or supplement traditional approaches.