Forecasting, interventions and selection: the benefits of a causal mortality model (Q6593140)

To improve forecasting accuracy, the paper focuses on integrating epidemiological information into mortality. By employing techniques from causal mediation theory, the selection effect, that is usually absent in studies on cause-of-death elimination, is measured. After a discussion about causal mortality models, the analysis aims to determine the impact of interventions and their consequences on mortality by examining how risk mechanisms at the individual level are transferred to populations, in relation to the potential outcomes. The introduction of a generic causal mortality model focuses on when explicit modeling of the feedback mechanism is necessary. A method that uses causal mediation theory is proposed to divide the total effect of an intervention into two components, one directly attributable to the action and the other due to selection. Finally, an application to US data explains the direct and indirect effects of cause-of-death elimination. Included in the Appendix are some technical aspects related to the study.