Mortality and cancer registration experience of the Sellafield workers known to have beeninvolved in the 1957 Windscale accident: 50 year follow-up

Abstract

This paper studies the mortality and cancer morbidity of the 470 male workers involved intackling the 1957 Sellafield Windscale fire or its subsequent clean-up. Workers werefollowed up for 50 years to 2007, extending the follow-up of a previously published cohortstudy on the Windscale fire by 10 years. The size of the study population is small, but thecohort is of interest because of the involvement of the workers in the accident. Significantexcesses of deaths from diseases of the circulatory system (standardised mortality ratio(SMR) = 120, 95%CI = 103–138; 194 deaths) driven by ischaemic heart disease (IHD) (SMR = 133, 95%CI = 112–157, 141 deaths) were found when compared with the population of Englandand Wales but not when compared with the population of Northwest England (SMR = 105, 95%CI = 90–120and SMR = 115, 95% CI = 97–136 respectively). When compared with those workers in post at the time of the firebut not directly involved in the fire the mortality rate from IHD among thoseinvolved in tackling the fire was raised but not statistically significantly (rate ratio(RR) = 1.11, 95%CI = 0.92–1.33). A RR of 1.11 is consistent with an excess relative risk of 0.65 Sv − 1 as reported in an earlier study of non-cancer mortality in the British NuclearFuels plc cohort of which these workers are a small but significant part.There was a statistically significant difference in lung cancer mortality (RR = 2.18, 95%CI = 1.05–4.52) rates between workers who had received higher recorded external doses during thefire and those who had received lower external doses. Comparison of the mortality rates ofworkers directly involved in the accident with workers in post, but not so involved, showedno significant differences overall. On the basis of the use of a propensity score theaverage effect of involvement in the Windscale fire on all causes of death was − 2.13% (se = 3.64%,p = 0.56) thoughthis difference is not statistically significant. The average effect of involvement in the Windscale firewas − 5.53% (se = 3.81,p = 0.15) for all cancersmortality and 6.60% (se = 4.03%, p = 0.10) for IHD mortality though neither figure was statistically significant.This analysis of the mortality and cancer morbidity experience of those Sellafield workersinvolved in the 1957 Windscale fire does not reveal any measurable effect of the fire upontheir health. Although this study has low statistical power for detecting small adverseeffects, due to the relatively small number of workers, it does provide reassurance that nosignificant health effects are associated with the 1957 Windscale fire even after 50 years offollow-up.