Modeling the Age Pattern of Human Mortality: Mathematical and Tabular Representations of the Risk of Death

Abstract

The age pattern of vital events is one of the oldest and most well studied topics in demography. Mortality is no exception with the introduction of the life table as a central demographic tool in the mid 17th century. Over the past 50 years, population analysts have developed mathematical equations to parsimoniously describe the shape of human mortality across age allowing for succinct time-series comparisons, but many of these models can only describe restricted portions of the age range and at times can be difficult to fit due to the necessarily large number of parameters needed to describe the age pattern of mortality. Furthermore, owing to estimation difficulties or data limitations, these models are rarely applied to African populations. The first part of this dissertation is devoted to investigating the sex-age-specific changes in mortality in a prospectively monitored rural population in South Africa during the course of the HIV epidemic. These changes are quantified by fitting the eight parameter Heligman-Pollard model of age-specific mortality using a robust Bayesian estimation method. Fitting a model like the Heligman-Pollard or calculating summary mortality indicators such as life expectancy at birth, requires a complete set of age-specific mortality rates. Absent high-quality vital registration systems that could collect age-specific mortality data, many countries rely on indirect techniques like <italic>model life tables</italic> -- models that produce a complete set of mortality rates from age restricted mortality indicators such as the probability of death before age five. Existing model life table systems are outdated and unable to produce certain mortality patterns like those with very low levels of child mortality or high adult mortality resulting from HIV. The second portion of the dissertation is devoted to generating a model life table system for both high and low-to-middle income countries by identifying similar age patterns of mortality in a collection of 844 life tables from the Human Mortality Database and 329 life tables from the INDEPTH network of Health and Demographic Surveillance Sites spread throughout Africa and parts of Asia.