Development of cancer risk estimates from epidemiologic studies

Radiation risk estimates may be made for an increase in mortality from, or for an increase in incidence of, particular types of disease. For both endpoints, two numerical systems of risk expression are used: the absolute risk system (usually the excess deaths or cases per million persons per year per rad), and the relative risk system (usually excess deaths or cases per year per rad expressed as a percentage of those normally expected). Risks may be calculated for specific age groups or for a general population. An alternative in both risk systems is the estimation of cumulative of lifetime risk rather than annual risk (e.g. in excess deaths per million per rad over a specified long period including the remainder of lifespan). The derivation of both absolute and relative risks is illustrated by examples. The effects on risk estimates of latent period, follow-up time, age at exposure and age standardization within dose groups are illustrated. The dependence of the projected cumulative (lifetime) risk on the adoption of a constant absolute risk or constant relative risk is noted. The use of life-table data in the adjustment of cumulative risk for normal mortality following single or annual doses is briefly discussed