Transfer of risk coefficients across populations

The variation of lifetime risk projections for a Canadian population caused by the uncertainty in the choice of method for transferring excess relative risk coefficients between populations is assessed. Site-specific projections, varied by factors up to 3.5 when excess risk coefficients of the BEIR V relative risk models were transferred to the Canadian population using an additive and multiplicative method. When the risk from all cancers are combined, differences between transfer methods were no longer significant. The Canadian projections were consistent with the ICRP-60 nominal fatal cancer risk estimates. (author)