The aim of the thesis is to study three potential sources of bias in the analyses on the cancer risk associated with chronic external exposure to low doses of ionizing radiation in the CEA-AREVA-EDF cohort. This cohort includes 59 004 nuclear workers from those companies who were badge-monitored for external exposure to radiation.
The first part of this thesis addresses potential internal exposure to radionuclides among the workers. Given the lack of available individual data on internal contaminations, a flag attributing for each worker and for each year a potential contamination risk based on workstations was defined and added to the analyses on cancer mortality risk associated with external radiation exposure.
The second part of this work focuses on non-occupational exposures to radiation. Nuclear workers are exposed to radiation from environmental and medical sources, in particular from radiological examinations carried out in an occupational health framework. Scenarios characterising non-occupational exposures to radiation were constructed and their impact on the cancer mortality risk associated with occupational exposure was examined.
The third part of this thesis addresses dosimeters recording thresholds. Below those thresholds, dose quantification is deemed too imprecise. Historically, below the threshold doses were recorded as null, resulting in systemic dose underestimation. A comprehensive review of the dosimetry practices in the cohort was accomplished and led to the conception of scenarios aiming to impute below the threshold.