Over the last two decades residential exposure to extremely low frequency magnetic fields (ELF MF) has been associated with childhood leukaemia relatively consistently in epidemiological studies, though causality is still under investigation. We aimed to estimate the cases of childhood leukaemia that might be attributable to exposure to ELF MF in the European Union (EU27), if the associations seen in epidemiological studies were causal. We estimated distributions of ELF MF exposure using studies identified in the existing literature. Individual distributions of exposure were integrated using a probabilistic mixture distribution approach. Exposure-response functions were estimated from the most recently published pooled analysis of epidemiological data. Probabilistic simulation was used to estimate population attributable fractions (AF P) and attributable cases of childhood leukaemia in the EU27. By assigning the literature review-based exposure distribution to all EU27 countries, we estimated the total annual number of cases of leukaemia attributable to ELF MF at between~50 (95% CIs: −14, 132) and~60 (95% CIs: −9, 610), depending on whether exposure-response was modelled categorically or continuously, respectively, for a non-threshold effect. This corresponds to between~1.5% and~2.0% of all incident cases of childhood leukaemia occurring annually in the EU27. Considerable uncertainties are due to scarce data on exposure and the choice of exposure-response model, demonstrating the importance of further research into better understanding mechanisms of the potential association between ELF MF exposure and childhood leukaemia and the need for improved monitoring of residential exposures to ELF MF in Europe.

No abstract

There is substantial public and scientific interest in the health consequences of exposures to ionising radiation in nuclear installations, in particular at nuclear fuels enrichment and production facilities. In this issue of Journal of Radiological Protection, McGeoghegan and Binks report on the follow-up of a cohort of over 19 000 uranium fuel and uranium hexafluoride production workers employed at the Springfields uranium production facility between 1946 and 1995. Workers included in the study were followed, on average, for almost 25 years. Findings from this study present potentially important information about mortality and cancer morbidity in populations receiving low-dose protracted exposures to ionising radiation. At uranium fuel production facilities such as Springfields, inhalation of airborne uranium dust may represent an important potential source of radiation exposure. Workers in these facilities therefore have two main possible sources of radiological exposure to tissues of the whole body: external gamma-ray exposure—which results in fairly uniform distribution of dose—and internal depositions that deliver radiation doses primarily to the lung and lymphatic system. If the uranium dust is solubilised, exposure may also result in other tissues such as the liver, the kidney and the bone. Tumours occurring in these organs are therefore, a priori, of particular interest in epidemiological studies of workers at uranium production facilities. Assessment of past internal exposure of nuclear workers is often complicated by methodological difficulties of internal dosimetry, as well as by inadequate historical information with which to accurately quantify internal radiation doses. These exposure measurement problems pose significant problems for epidemiology: the inability to accurately classify workers by level of internal radiation exposure may lead to confounding of the analyses of radiation-cancer associations, since workers with significant dose from internal contamination are often persons with substantial external exposure. It is important to consider these concerns with regards to the findings reported by McGeoghegan and Binks, as analyses in this paper are restricted to external doses only. In this paper, standardised mortality and morbidity ratios (SMRs), which did not rely upon exposure classification, were reported. For the full cohort, deficits in mortality were observed in a number of broad categories of cause of death, including all causes, all cancers, and cardiovascular disease. This conforms to expectations about health-related selection and survivorship in relatively well-paid industrial cohorts [1-3]. SMRs for these broad categories of cause of death were lower for radiation workers than for non-radiation workers, suggesting stronger processes of health-r...