Occupational Exposure to Ionizing Radiation: Models for Policy Making

摘要 据估计，每年有150万美国工人在职业中暴露于电离辐射。的 在美国，每年约有12万名工人接受辐射照射监测 能源部设施，而其他数十万人在与医疗机构合作时受到辐射 辐射源、核能发电和工业过程，如射线照相和食品 辐射。与许多其他已确立的职业致癌物形成对比的是，已从 随着时间的推移，美国工作场所受到辐射照射的工人数量不仅没有减少，反而增加了 随着辐射在医学和其他工业环境中的新用途的出现。我们对此的理解 辐射对健康的影响来自多种来源，包括实验室研究、研究 医疗辐射和对原子弹幸存者的研究。然而，对工人的流行病学研究认为 一个特殊的地方，因为这样的研究允许直接评估不需要外推的证据 从细胞到有机体，物种之间，或跨种群和暴露条件。 直到最近，对辐射工作者的流行病学研究往往导致不准确的风险估计 经常跨越空值的置信度区间。因此，日本原子性肺炎的流行病学分析 炸弹幸存者一直是辐射防护标准的主要量化基础。然而， 最近的流行病学研究表明，集合队列数据产生的辐射风险估计相对较接近 可信区间。加强对当代放射工作者的保护基础，并改进 对于过去暴露的工人的薪酬决定，我们建议使用最先进的统计分析方法 参数g公式方法应用于最近作为主要数据的一部分汇编的数据 汇集在英国、法国和美国受雇的核工作人员数据的国际努力。 具体地说，我们建议评估：1)辐射效应的时间修饰物(自暴露以来的时间、在 暴露和达到的年龄)；2)辐射效应按癌症类型的变化；以及3)非癌症患者的辐射效应 癌症是致死原因。此外，我们还提出了贝叶斯方法来评估4)剂量和剂量率效应； 5)由于结果分类错误而产生的偏差。因为对流行病学发现的因果解释是 通过可重复性和一致性的证据加强，我们将评估得出的结果的一致性 来自这些国际同龄人。建议的方法使我们能够最大限度地减少偏见，包括健康的工人 幸存者偏见，正式结合来自不同研究的信息，并利用这些最近汇集的核 工人队列数据。这一研究项目的发现预计将对 了解职业性辐射暴露的影响。这项工作将通过以下方式解决Nora的优先事项 提高对辐射致癌的认识，强化政策依据 建议。 1

ABSTRACT There are an estimated 1.5 million U.S. workers occupationally exposed to ionizing radiation each year. Of this number, approximately 120,000 workers are monitored annually for radiation exposure at United States Department of Energy facilities, while hundreds of thousands of others are exposed while working with medical sources of radiation, nuclear power generation, and industrial processes such as radiography and food irradiation. In contrast to many other established occupational carcinogens, which have been removed from US workplaces over time, the number of radiation-exposed workers has not diminished, but rather has grown with the emergence of new uses of radiation in medicine and other industrial settings. Our understanding of the health effects of radiation exposure comes from a variety of sources, including laboratory research, studies of medical irradiation, and studies of atomic bomb survivors. However, epidemiological studies of workers hold a special place because such studies allow direct evaluation of evidence that does not require extrapolation from cells to organisms, between species, or across populations and exposure conditions. Until recently, epidemiological studies of radiation workers tended to result in imprecise risk estimates with confidence intervals that often spanned the null. Consequently, epidemiological analyses of Japanese atomic bomb survivors have served as the primary quantitative basis for radiation protection standards. However, recent epidemiological studies that pool cohort data have yielded radiation risk estimates with relatively tight confidence intervals. To strengthen the basis for protection of contemporary radiation workers, and to improve compensation decisions for workers exposed in the past, we propose state-of-the-art statistical analysis using parametric g-formula methods applied to data that recently have been assembled as part of a major international effort to pool data for nuclear workers employed in the United Kingdom, France, and USA. Specifically, we propose to assess: 1) temporal modifiers of radiation effects (time-since-exposure, age-at- exposure, and attained age); 2) variation in radiation effects by type of cancer; and, 3) radiation effects on non- cancer causes of death. In addition, we propose Bayesian methods to evaluate 4) dose and dose-rate effects; and, 5) bias due to outcome misclassification. Because a causal interpretation of epidemiological findings is strengthened by evidence of reproducibility and consistency, we will assess the consistency of results derived from these international cohorts. The proposed methods allow us to minimize bias, including healthy worker survivor bias, formally combine information from different studies, and leverage these recently pooled nuclear worker cohort data. The findings of this research project are expected to have substantial impact on understanding of the effects occupational radiation exposures. The work will address NORA priorities by improving understanding of cancer caused by radiation and strengthening the basis for policy recommendations. 1