Combined analysis of mortality among nuclear workers

摘要 20世纪60年代，美国原子能委员会发起了一项关于老年人健康和死亡率的研究计划 美国核工人。从一开始就注意到，针对具体设施的队列研究不太可能包括 足够数量的工作人员，以提供统计上准确的流行病学结果；然而，结合数据 跨设施可以让调查人员研究大量核工作人员的关联。我们 建议对5家主要工厂的员工进行美国员工数据的汇集分析，这些数据基于 队列规模、人口和就业历史信息的可获得性以及辐射的完整性 监测数据：爱达荷州橡树岭国家实验室、萨凡纳河遗址、汉福德遗址 实验室和朴茨茅斯海军造船厂。每个队列的生命状态跟踪的最新情况是 最近进行的，使拟议的分析是可行的。拟议的对美国工人的联合研究将 允许我们解决有关辐射风险随暴露时间变化的悬而未决的问题，以及 暴露年龄，以及不同癌症类型辐射风险的差异。我们建议采用最先进的- ART方法，最大限度地减少由于混淆而产生的偏差，敏感性分析，以评估暴露的影响 错误分类，以及评估效果测量修改和队列间的创新回归模型 辐射风险的异质性。此外，对流行病学研究结果的因果解释是 可再现性和一致性的证据加强了这一点。因此，我们建议对 美国核工作者数据得出的结果与英国和法国核数据得出的结果的一致性 工人群体，这两个国家也开发了大型核计划。一种新的分层回归方法 将使用一种方法来评估美国、英国和法国工人在联合调查中发现的一致性 对每个队列采用可比较的分析方法的分析。如果汇集这3个国家/地区的数据是 适当时，将进行国际联合分析，以得出最精确的辐射 到目前为止，已报告的核工作人员风险估计值。拟议的国际联合分析将具有 比以往任何有关核工作人员癌症的流行病学研究的统计能力都高出一倍以上。这个 这一研究项目的发现预计将对有关职业的讨论产生重大影响 辐射风险和辐射防护。

Abstract In the 1960s, the US Atomic Energy Commission initiated a research program on the health and mortality of US nuclear workers. From the outset, it was noted that facility-specific cohort studies were unlikely to include sufficient numbers of workers to provide statistically precise epidemiological findings; however, combining data across facilities could allow an investigator to study associations in a large population of nuclear workers. We propose to conduct a pooled analysis of US worker data for employees of 5 major facilities selected based on cohort size, availability of demographic and employment history information, and completeness of radiation monitoring data: Oak Ridge National Laboratory, Savannah River Site, Hanford site, Idaho National Laboratory, and Portsmouth Naval Shipyard. An update of vital status follow-up of each cohort has been conducted recently, making the proposed analysis feasible. The proposed combined study of US workers will permit us to address outstanding questions regarding variation in radiation risks with time-since-exposure and age-at-exposure, as well as variation in radiation risks by type of cancer. We propose to employ state-of-the- art methods to minimize bias due to confounding, sensitivity analyses to assess the impact of exposure misclassification, and innovative regression models to assess effect measure modification and between-cohort heterogeneity in radiation risks. Moreover, a causal interpretation of epidemiological study findings is strengthened by evidence of reproducibility and consistency. Therefore, we propose to empirically assess the consistency of results derived from US nuclear worker data with results derived from UK and French nuclear worker cohorts, two nations that also developed large nuclear programs. A novel hierarchical regression approach will be used to assess the consistency of findings between US, UK, and French workers in a joint analysis that employs comparable analytical methods for each cohort. If pooling data from these 3 countries is appropriate, an international combined analysis will be conducted in order to derive the most precise radiation risk estimates for nuclear workers reported to-date. The proposed international combined analysis will have more than twice the statistical power of any prior epidemiological study of cancer among nuclear workers. The findings of this research project are expected to have substantial impact on discussions regarding occupational radiation risks and radiation protection.