Geographic and sociodemographic factors in the etiology of juvenile neoplasms: implications for leukaemia clusters of several factors including density, background radiation, infections and immunisations

This paper presents an analysis of data from the Oxford Survey of Childhood Cancers (OSCC) by the recently invented method of Fears and Brown, as modified (for correct variances of relative risk estimates) by Breslow and Zhao. The analysis has shown that, for OSCC cases, negative associations with population density, stillbirths and infant deaths were sufficiently strong for these alone to account for all but one of the leukaemia clusters so far reported in Britain as 'black hole' effects (the single exception being the Sellafield cluster). Finally, strong positive associations with regional levels of background radiation were found which have reinforced an earlier impression that these inevitable exposures are important causes of childhood cancer. (orig./MG)