In this presentation we will present a model that takes a holistic approach to the estimation of cohort effects in US mortality data. Data has been collected by sex and education level for 51 causes of death for the US for single ages and years from 1989. Previous work has fitted CBD-X age-period-cohort model to these data. Discussion focused on the similarity of the fitted cohort effects for groups of causes of death with, for example, one such group having smoking as the main driver of death rates for causes in the group. We build on this by proposing a single model for all causes of death that has a small number of distinct cohort effects that can be linked to specific health behaviours.

We will discuss to what extent this approach leads to new insights into the cohort effect that can be observed at the all-cause level.

Find the Q&A here: Q&A on 'Mortality Modelling and Long Term Care'