Space-Time Clustering of Non-Hodgkin's Lymphoma Using Residential Histories

描述（由申请人提供）： 拟议的研究项目的主要目标是产生有价值的见解，非霍奇金淋巴瘤（NHL）的病因，通过调查在美国的一个大型，多中心，以人群为基础的研究中的病例和对照的当前和过去的居住地的时空聚类。尽管全世界每年诊断出超过30万例NHL，约占所有癌症的2.8%，但对NHL的病因学知之甚少。作为一个整体，已建立和推定的风险因素仅占每年发生的NHL病例总数的一小部分，需要新的技术来解决这个谜。我们的研究团队最近开发了Q统计和时空信息系统（STISTM）技术，使时空聚类分析成为可能。用于分析癌症集群的其他方法通常忽略了居住地的流动性，并且几乎只适用于诊断时或死亡时居住地的静态空间点分布。此外，很少有空间技术充分考虑已知的风险因素和协变量。我们的方法解决了每一个这些需求，通过利用住宅历史的参与者表示为一个生命线，从而评估在任何时候在生活过程中的住宅历史的情况下，相对于控制的住宅历史的时空聚类。此外，在空间随机性的广泛使用（但往往不合适）零假设的地方，Q统计可以根据他/她已知的风险因素和协变量将每个人的概率纳入病例。 该项目将应用创新的Q-统计和STISTM技术来识别NHL在病例的居住历史相对于对照的居住历史的生命过程中的任何时刻的空间聚集。这将是有史以来第一次，允许流行病学家评估可能的地理和时间集群NHL发病率使用当前和过去的住所。该研究是对一个大型的、仔细收集的多中心数据集（1321例病例，1057例对照）的二次分析，其目的是产生与NHL相关的潜在病因学因素的新假设，以便在未来的研究中进行研究。

DESCRIPTION (provided by applicant): The primary objective of the proposed research project is to generate valuable insights concerning the etiology of Non-Hodgkin's lymphoma (NHL) by investigating space-time clustering of current and past residences of cases and controls in a large, multi-center, population-based study in the United States. Despite more than 300,000 cases of NHL diagnosed each year worldwide, accounting for approximately 2.8% of all cancers, relatively little is known about the etiology of NHL. The established and presumptive risk factors, taken as a whole, account for only a small proportion of the total NHL cases that occur annually, and novel techniques are needed to solve this enigma. Our research team has recently developed Q-statistics and Space-Time Information System (STISTM) technology that enable space-time cluster analyses. Other methods for analyzing cancer clusters typically ignore residential mobility, and almost exclusively work with static spatial point distributions of place-of-residence at time of diagnosis or time of death. In addition, few spatial techniques adequately account for known risk factors and covariates. Our approach addresses each of these needs by utilizing the residential history of the participants represented as a life-line, and thus evaluates space-time clustering at any moment in the life-course of the residential histories of the cases relative to the residential histories of the controls. In addition, in place of the widely used (but often inappropriate) null hypothesis of spatial randomness, Q-statistics can incorporate each individual's probability of being a case based on his/her known risk factors and covariates. This project will apply the innovative Q-statistics and STISTM technology to identify spatial clustering of NHL at any moment in the life-course of the residential histories of the cases relative to the residential histories of the controls. This will, for the first time ever, allow epidemiologists to assess possible geographic and temporal clusters of NHL incidence using current and past residences. The proposed research is a secondary analysis of a large, carefully collected pre-exiting multi-center dataset (1321 cases, 1057 controls), with the intended purpose of generating new hypotheses about potential etiologic factors associated with NHL to be investigated in future studies.