Some Age-Period-Cohort Models to Study Time Trends in the Incidence of Cancer of the Larynx

Abstract

In this paper, three approaches are described to fit Age-PeriodCohort models to study the temporal variation in the incidence of cancer Osmond and Gardner (1982) estimate the parameters for the full models from the estimates of three two-factor models minimizing the Euclidean distance between the parameters. The concept of drift given by Clayton and Schifflers (1987) is described. Robertson and Boyle explained how individual records could be used tq get estimates of parameters for the full Age-Period-Cohort model with more precision. The methods of Osmond and Gardner (1982) and Decarli and La Vecchia (1987) are conceptually similar. The parametric bootstrap method can be applied to provide Monte Carlo estimates of the standard errors.