Cancer Clustering for Residential Histories

摘要：本项目将开发一种新的、综合的方法来评估病例对照数据的聚类，该方法考虑了居住史、危险因素、协变量以及癌症诱发和潜伏期。迄今为止，癌症地理研究的两个主要缺陷是缺乏一种适当的方法来调查生命过程中的群集，并且风险因素/协变量未包括在分析中。这个项目克服了这些限制。将根据最近邻关系矩阵集开发居住历史的本地、全球和重点测试，这些矩阵集反映了案例和控制的居住地址的时空几何变化。将定义解释暴露与疾病表现之间的潜伏期的暴露轨迹，以及使用不同持续时间的暴露窗口。由此衍生的几种方法将用于评估密歇根州东南部正在进行的膀胱癌病例对照研究中居住史的聚类，并用于确定加利福尼亚州马林县的局部乳腺癌过度。由于人类是可移动的，这些新方法相对于忽略居住历史，而只依赖于诊断或死亡时的居住地点的方法是重大的进步。主要的创新是创造了分析和模拟病例和控制的居住历史的方法，以确定研究人群本身以及与假定的危害（如致癌物的点源释放）相关的地理过度癌症风险。与居住史相关的癌症聚类：本项目将开发的技术和软件将提供更简洁和准确的癌症病例聚类描述，该描述考虑了风险因素、协变量、居住史、癌症潜伏期、诊断时间和假定发生致癌暴露的暴露窗口。据我们所知，这个项目的技术和软件将是第一个在一个单一的、全面的框架内解决所有这些因素的。

DESCRIPTION (provided by applicant): Cancer Clustering for Residential Histories Abstract: This project will develop a new, comprehensive approach for evaluating clustering in case-control data that accounts for residential histories, risk factors, covariates and cancer induction and latency periods. To date, two of the major deficiencies of geographic studies of cancer are that they lack an appropriate methodology for investigating clusters over the life course and risk factors/covariates are not included in the analysis. These limitations are overcome by this project. Local, global and focused tests for residential histories will be developed based on sets of matrices of nearest neighbor relationships that reflect the changing space-time geometry of the residential addresses of cases and controls. Exposure traces that account for the latency between exposure and disease manifestation, and that use exposure windows of varying duration will be defined. Several of the methods so derived will be applied to evaluate clustering of residential histories in an ongoing case-control study of bladder cancer in south eastern Michigan, and to identify local excesses of breast cancer in Marin County, California. Because humans are mobile, these new methods are a significant advance over approaches that ignore residential histories and instead rely only on place of residence at time of diagnosis or death. The major innovation is the creation of methods for analyzing and modeling the residential histories of cases and controls to identify geographic excesses of cancer risk, both in the study population itself as well as in relation to putative hazards such as point-source releases of carcinogens. Cancer Clustering for Residential Histories Relevance: The techniques and software to be developed in this project will provide a more concise and accurate description of clustering of cancer cases that accounts for risk factors, covariates, residential history, cancer latency, time of diagnosis, and the exposure windows during which causative exposures are hypothesized to have occurred. To our knowledge the techniques and software from this project will be the first to address all of these factors within a single, comprehensive framework.