Estimating 1a0 and 4a1 in a Life Table: A Model Approach Based on Newly Collected Data.

Abstract

The average age of infant deaths, a10, and the average number of years lived-in the age interval-by those dying between ages 1 and 5, a41, are important quantities allowing the construction of any life table including these ages. In many applications, the direct calculation of these parameters is not possible, so they are estimated using the infant mortality rate-or the death rate from 0 to 1-as a predictor. Existing methods are general approximations that do not consider the full variability in the age patterns of mortality below the age of 5. However, at the same level of mortality, under-five deaths can be more or less concentrated during the first weeks and months of life, thus resulting in very different values of a10 and a41. This article proposes an indirect estimation of these parameters by using a recently developed model of under-five mortality and taking advantage of a new, comprehensive database by detailed age-which is used for validation. The model adapts to a variety of inputs (e.g., rates, probabilities, or the proportion of deaths by sex or for both sexes combined), providing more flexibility for the users and increasing the precision of the estimates. This fresh perspective consolidates a new method that outperforms all previous approaches.