Predictive value of cancer statistics

The value of routinely collected cancer statistics for hypothesis generation and for risk prediction is examined in relation to age, place, migration, time, anatomical sub-site, morphology, and occupation. Current descriptive epidemiological data of this nature reflect the carcinogenic exposures of the past 40 years, and are thus useful for generating hypotheses. While they do not permit prediction of risk associated with a new substance entering the environment, once etiology for a given cancer is known, the likely course of events unless preventive action is taken can be forecast on the basis of existing statistics.