Trends and forecasts in cause-specific mortality
Mortality forecasting models are typically limited in that they pertain only to national death rates, predict only all-cause mortality, or do not capture and utilize the correlation among diseases. I have developed a novel Bayesian hierarchical model that jointly forecasts cause- specific death rate...
Main Author: | |
---|---|
Other Authors: | |
Published: |
Imperial College London
2015
|
Online Access: | https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.754642 |
id |
ndltd-bl.uk-oai-ethos.bl.uk-754642 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-bl.uk-oai-ethos.bl.uk-7546422019-02-05T03:24:58ZTrends and forecasts in cause-specific mortalityForeman, Kyle JaysonEzzati, Majid ; Best, Nicky2015Mortality forecasting models are typically limited in that they pertain only to national death rates, predict only all-cause mortality, or do not capture and utilize the correlation among diseases. I have developed a novel Bayesian hierarchical model that jointly forecasts cause- specific death rates for geographic subunits. I examined the model’s effectiveness by applying it to United States vital statistics data from 1982 to 2011 that I prepared using a new cause of death reassignment algorithm. I found that the model not only generated coherent forecasts for mutually exclusive causes of death, but it also exhibited lower out-of-sample error than alternative commonly-used models for forecasting mortality. I then used the model to produce forecasts of US cause-specific mortality through 2025 and analysed the resulting trends. I found that total death rates in the US were likely to continue their decline, but at a slower rate of improvement than has been observed for the past several decades. While death rates due to major causes of death like ischaemic heart disease, stroke, and lung cancer were projected to continue trending downward, increases in causes such as unintentional injuries and mental and neurological conditions offset many of these gains. These findings suggest that the US health system will need to adapt to a changing cause composition of disease burden as its population ages in the coming decade. Forecasting research should continue to consider how to best incorporate and balance the many dimensions of mortality when producing projections.Imperial College Londonhttps://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.754642http://hdl.handle.net/10044/1/61344Electronic Thesis or Dissertation |
collection |
NDLTD |
sources |
NDLTD |
description |
Mortality forecasting models are typically limited in that they pertain only to national death rates, predict only all-cause mortality, or do not capture and utilize the correlation among diseases. I have developed a novel Bayesian hierarchical model that jointly forecasts cause- specific death rates for geographic subunits. I examined the model’s effectiveness by applying it to United States vital statistics data from 1982 to 2011 that I prepared using a new cause of death reassignment algorithm. I found that the model not only generated coherent forecasts for mutually exclusive causes of death, but it also exhibited lower out-of-sample error than alternative commonly-used models for forecasting mortality. I then used the model to produce forecasts of US cause-specific mortality through 2025 and analysed the resulting trends. I found that total death rates in the US were likely to continue their decline, but at a slower rate of improvement than has been observed for the past several decades. While death rates due to major causes of death like ischaemic heart disease, stroke, and lung cancer were projected to continue trending downward, increases in causes such as unintentional injuries and mental and neurological conditions offset many of these gains. These findings suggest that the US health system will need to adapt to a changing cause composition of disease burden as its population ages in the coming decade. Forecasting research should continue to consider how to best incorporate and balance the many dimensions of mortality when producing projections. |
author2 |
Ezzati, Majid ; Best, Nicky |
author_facet |
Ezzati, Majid ; Best, Nicky Foreman, Kyle Jayson |
author |
Foreman, Kyle Jayson |
spellingShingle |
Foreman, Kyle Jayson Trends and forecasts in cause-specific mortality |
author_sort |
Foreman, Kyle Jayson |
title |
Trends and forecasts in cause-specific mortality |
title_short |
Trends and forecasts in cause-specific mortality |
title_full |
Trends and forecasts in cause-specific mortality |
title_fullStr |
Trends and forecasts in cause-specific mortality |
title_full_unstemmed |
Trends and forecasts in cause-specific mortality |
title_sort |
trends and forecasts in cause-specific mortality |
publisher |
Imperial College London |
publishDate |
2015 |
url |
https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.754642 |
work_keys_str_mv |
AT foremankylejayson trendsandforecastsincausespecificmortality |
_version_ |
1718973277398892544 |