Improving Estimates of Mortality at Older Ages

Spencer Pease, Institute for Health Metrics and Evaluation, University of Washington
Austin E. Schumacher, University Of Washington

Child and adult mortality have been focuses of global health research for decades, and as such have generally broad data availability and robust estimation methods. Old age mortality, in comparison, lacks available data beyond high quality vital registration systems and has limited available methodology for estimation. As life expectancies rise and population distributions shift towards older ages, producing accurate estimates of mortality in these older ages is a topic of increasing interest. We develop and compare parametric and iterative models for extending available vital registration data on old age mortality (ages 60 plus) to higher open age intervals where old age data is nonexistent or unreliable, with the goals of creating reliable estimates of old age mortality out to a consistent open age interval and improving upon existing methods currently in use in the Global Burden of Diseases, Injuries and Risk Factors Study (GBD) mortality estimation process. Performance of these methods are assessed using a model validation framework comparing extended mortality rates produced from artificially collapsed data to the true mortality rates. A secondary analysis assesses the performance of these methods on mortality in the years 2020-2021 during the height of the COVID-19 pandemic.

Keywords: Population Ageing, Mortality and Longevity, Data and Methods

See extended abstract.