The objective of the project is to develop more accurate models to forecast mortality by cause of death and disability level, within a CoDA framework.
Duration of the Project: 2020 – 2022
The main investigator is Marie-Pier Bergeron-Boucher, postdoctoral researcher at the Interdisciplinary Center on Population Dynamics (CPop). There will also be collaboration with James W. Vaupel, Professor, and Jim Oeppen, Associate Professor at University of Southern Denmark (USD), and Søren Kjærgaard, Ph.D. student at CPop.
The objective of the project is to develop more accurate models to forecast mortality by cause of death and disability level, within a CoDA framework, providing by the end of the project:
- a forecast by cause of death where one cause cannot come to dominate the forecast.
- a forecast by cause which could have predicted actual mortality better than, or at least consistent with, an all-cause forecast, using historical forecasts.
- a forecast by disability level with a high degree of accuracy, assessed using historical forecasts.
Forecasts by cause of death and disability level have been confronted with different methodological problems and the CoDA framework is an innovative approach which addresses some of these. Such methodology could be used to tackle the differences in mortality dynamics between sub-populations. For instance, national demographic data is frequently used to predict future mortality improvements due to the homogeneity and robustness of the data. However, inherent differences in the risk profile of an insured population from the general population create a basis risk. Mortality in an insured population differs from the total population both in terms of the distribution and the dynamic of the cause-specific mortality. Forecasting by cause of death could better capture the mortality dynamic specific to an insured population. Such forecasts also better represent changes in underwriting practices such as those currently experienced by the United States with the development of accelerated underwriting.
* * * * *
First Results of the Project [EN] - Author: Marie-Pier Bergeron-Boucher, Odense 28 October 2020
"Mortality forecasts by age and cause of death - How to forecast both components coherently"
"Mortality forecasts by age and cause of death are important for more efficient spending on, for example, health care and medical technology. However, there is a reluctance in including the cause of death dimension to the forecast, as forecasts by cause are confronted with many methodological problems. While some of these problems have been addressed in the last two decades, an important remaining issue with forecasts by cause is their inconsistence with all-causes forecasts. This problem relates to how changes in mortality by age and cause interact. So how can we forecast this relation in a coherent manner? To address this problem, we use a model framework based on a Compositional Data Analysis (CoDA) approach which model 1) age and cause simultaneously; 2) cause-of-death distribution within each age group; and 3) age-at-death distribution within each cause. We specify multiple models within each of the three frameworks to obtain a better understanding of the age and cause interactions. The results show that forecasting cause-of-death distribution within each age group generally provide the most accurate forecasts and allow for the forecast by cause and for all-cause to be consistent with one-another. Our results highlight that forecasting the age-specific cause-of-death distribution using a damped trend model and linked it to an all-cause forecast provide the most accurate forecasts, while avoiding many of the concerns about forecasting by cause of death". - Marie-Pier Bergeron-Boucher
Second Results of the Project [EN] – Author: Marie-Pier Bergeron-Boucher, Paris 6 April 2022
"Modelling and forecasting healthy life expectancy. A Compositional Data Analysis approach"
“Will the extra years of life gained by the increase in life expectancy be lived in good or bad health? As forecasts support social, economic and medical decisions, as well as individuals' choices, there is a clear rationale for forecasting healthy life expectancy. However, only a limited number of models are available to forecast healthy life expectancy. Some require separate forecasts of transition rates for mortality within different health statuses, and of the incidence rate. We suggest two methods that can simultaneously forecast mortality and health prevalence, based respectively on the Sullivan and the multistate approach to estimate healthy life expectancy. Both forecast models use a Compositional Data Analysis (CoDA) approach, accounting for the correlation between ages and health statuses. The methods are applied to the mortality of Swedish females aged 60 and above. We show that deaths have shifted towards older ages and states that are not severely limited, leading to less years of life with severe disability.”- Marie-Pier Bergeron-Boucher