The cancer mortality and incidence experience of workers at British Nuclear Fuels plc, 1946-2005.

The aim of this study was to estimate cancer mortality and incidence risk associated with external radiation exposure in the BNFL cohort of nuclear workers and to determine if these risks are modified by potential for internal exposure. The cohort comprised 64,956 individuals who were employed at the four study sites between 1946 and 2002, followed up to 2005. External radiation exposures as measured by personal dosimeters (generally 'film badges') were available for 42,431 individuals classified as 'radiation workers'. Poisson regression models were used to investigate cancer mortality and incidence in relation to cumulative external radiation exposure using relative risk models. The cohort showed the expected 'healthy worker' effect. This analysis found an increased risk of all cancers associated with external occupational radiation exposure (ERR/Gy = 0.34 90% CI: 0.07; 0.64), with significant excess risks observed for all solid cancers (ERR/Gy = 0.29 90% CI: 0.02; 0.59) and leukaemia excluding CLL (ERR/Gy = 2.60 90% CI: 0.28; 7.01). The overall cancer risk estimates are consistent with values used by national and international bodies in setting radiation protection standards. The slopes of the dose response relationships for all cancer mortality and incidence were found to be significantly less steep for workers exposed to both external radiation and potentially to internal radiation (ERR/Gy = 0.09 90% CI: -0.17; 0.39) when compared to those workers only exposed to external radiation (ERR/Gy = 1.14 90% CI: 0.49; 1.89). Analyses of individual cancer types indicate that this overall result is mainly driven by that for digestive cancers and in particular cancers of the oesophagus. Categorical analyses also revealed that the difference in the dose response relationship between the two groups is only apparent for those exposed to cumulative external doses in excess of 200 mGy. Such differences have also been observed for non-cancer mortality outcomes in this cohort. Further work is required to explain these differences; for example, whether they may result from confounding by internal organ dose or lifestyle factors associated with socio-economic status.