Abstract
In 1990, Professor Martin Gardner and his colleagues reported the results of the West Cumbria leukaemia and lymphoma case-control study. The most striking finding was a statistical association between relatively high doses of radiation (cumulative doses more than 100 mSv) recorded by film badges worn by fathers while employed at the Sellafield nuclear installation before the conception of their children, and the level of leukaemia in these children. This highly publicised finding led to extensive research being carried out to allow a proper interpretation of this novel association. One such study, which commenced in 1992, was a case-control study of childhood cancer around three USA Department of Energy (DOE) nuclear facilities. The results of this study are available as a USA National Institute of Occupational Safety and Health (NIOSH) report. The study was completed in November 1997 and the results communicated to workers and management at the sites studied in October 1998. However, as yet, the study findings have not been published in a scientific journal and, therefore, the results are not widely known, which is disappointing. Given the limited dissemination of the study findings, they are summarised here.
Dr Lowell Sever and his colleagues from the Battelle Memorial Institute, Seattle, studied cases of leukaemia, non-Hodgkin lymphoma and central nervous system (CNS) tumours diagnosed during 1957-1991 while under 15 years of age and resident in selected counties around Hanford, Idaho National Engineering and Environmental Laboratory, and Oak Ridge. To be eligible for inclusion in the study, cases also had to be born to residents of one of these counties. Two non-overlapping sets of four controls were selected for each case, individually matched on year of birth, county of residence, sex, race and maternal age, and, for the second set only, paternal preconceptional employment at one of the study sites when the father of the affected child was so employed. Controls had to be resident in the counties at the time of diagnosis of the matched case. Parents were linked to employment at the nuclear sites using personnel data and doses were calculated from external dosimetry records: cumulative preconceptional doses for mothers and fathers, and the dose during gestation for mothers. Exposures due to internally deposited radionuclides were also assessed. The methodology was similar to that of a previous case-control study of congenital malformations around Hanford and occupational exposure of parents to radiation (L E Sever et al 1988 Am. J. Epidemiol. 127 226-42).
Of 233 children with cancer included in the study, 28 had a father employed at one of the three sites before conception, compared with 135 out of 932 controls, giving a relative risk (RR) of 0.77 (95% confidence interval (CI): 0.48 to 1.24). The distribution of RRs for paternal preconceptional employment by cancer type and nuclear site is not unusual, although the authors note that for CNS tumours at Hanford the RR is 2.50, but this is based upon 10 cases and is not statistically significant (95% CI: 0.79 to 7.96). For each cancer type, the mean paternal preconceptional dose is lower for cases than for controls. No father of a case child, but four fathers of control children (three of whom were from Hanford), had a dose before conception in excess of 100 mSv, two of these controls being matched to leukaemia cases. Clearly, then, the RRs for the greater than 100 mSv dose category (that produced the associations of note in the study of Gardner et al) are unremarkable. When the authors treated the paternal preconceptional dose as a continuous, rather than categorical, variable, and assumed a linear relationship of relative risk with dose, the RR at 100 mSv is 0.75 (95% CI: <0.75 to 2.1) for all cancers and 0.73 (95% CI: <0.73 to 7.6) for leukaemia. The RR at 100~mSv for leukaemia at Hanford alone is 1.93 (95% CI: <0.1 to 29), which is far from being statistically significant. The RR at 100 mSv for CNS tumours at Hanford is 0.22 (95% CI: <0.22 to 8.1), so that the non-significant association with paternal preconceptional employment is not reflected in an association with dose. RRs for paternal internal exposures, and those for maternal employment and dose, are unremarkable.
The authors of this USA study draw the following conclusion: 'The results of this study, for all the cancer types combined, for leukaemia and lymphoma, both for individual DOE facilities and for the three facilities combined, are consistent with a null hypothesis of no association between paternal preconception exposure and risk from these forms of childhood cancer. Thus, our findings do not support the earlier association observed by Gardner et al for Sellafield. The findings are consistent with subsequent studies that have failed to demonstrate (observe) increased risks associated with preconception doses of radiation received by fathers employed by nuclear facilities. It is our interpretation of these results and the accumulating scientific evidence that there do not appear to be increased risks of leukaemia and non-Hodgkin's lymphoma among children whose fathers have been occupationally exposed to ionising radiation at currently acceptable levels'.
The report of this study, 'Epidemiologic Evaluation of Childhood Leukemia and Paternal Exposure to Ionizing Radiation', Final Report (October 9, 1997), Centers for Disease Control and Prevention Cooperative Agreement U50/CCU012545-01, by L E Sever et al, can be obtained by contacting the NIOSH Office in Cincinnati, OH. It is to be hoped that a paper describing this study, which clearly involved appreciable time and effort, will appear in a scientific journal, and that its proper place in the literature will not be neglected because, as sometimes unfortunately happens, negative results are considered to be uninteresting.