The applied measurement of mortality: a critique and a revision

William L J Xu-Doeve, Senior Partner, ANRC Consulting

In countries with adequate mortality statistics, the common applied approach to measuring mortality is based on establishing age-specific empirical annual occurrence / exposure rates Mx: the number of recorded events of death in an age group during a year, divided by a measure of the population of that age and at risk during the year. In a next step, these mortality rates are converted to probabilities of dying qx, which form the basis of a life table characterizing the empirically observed mortality regime. Scientifically this approach is less than optimal. In particular, the occurrence / exposure rates Mx are fundamentally flawed on mathematical, methodological, and empirical grounds. This paper explains and examines these flaws. Using conventionally available data on deaths and well-established mathematical methods, it demonstrates how these flaws can be avoided, and how even more informative results, such as instantaneous mortality rates, can be obtained in a routine manner.