ABSTRACT
This chapter continues the discussion of the conventional age-period-cohort (APC) accounting model but focuses on empirical applications. It first describes a three-step procedure that combines descriptive and statistical APC analysis using the Intrinsic Estimator through studies of recent cancer incidence and mortality trends by sex and race in the United States. There is evidence of nonparallelism among age patterns by time period in rates of both incidence and mortality for most cancer sites, indicating cohort variations. The chapter examines the A, P, and C effects on cancer incidence and mortality rates based on the results of model comparison analysis using the best-fitting models for each site, outcome, and demographic group. The APC model-based forecasting methods showed substantial advantages over the simple data extrapolation methods in accuracy of point forecasts and assessments of uncertainty in these forecasts through prediction intervals.
