Uncovering adult modal age at death in populations with grouped data

Carlo G. Camarda, Institut National d'Études Démographiques (INED)
Nadine Ouellette, Institut National d'Études Démographiques (INED)

The adult modal age at death is a measure of the most frequent length of life among adults. It has been demonstrated to be an important lifespan measure in longevity research, capable of shedding light on some specific aspects of old-age mortality that are not necessarily captured well by other widely used measures of old-age survival. The current P-spline method for estimating the modal age at death is flexible and highly effective but the model requires population size and mortality data that are detailed by single years of age. For several countries and regions of the world, these data are solely available for broader age groupings. In this paper, we introduce a generalized version of the earlier method based on the Penalized Composite Link Model to estimate the modal age at death in instances where population estimates and/or mortality data are not provided by single years of age. We start by illustrating the new method and assessing its performance using data from the Human Mortality Database. We then uncover sex-specific trends in the modal age at death between 1996 and 2010 in Brazil, a country where population estimates and mortality data are readily available by 5-year age groups only from the WHO Mortality Database and Latin American Human Mortality Database. We offer concluding remarks about the vast range of possible applications and extensions of the newly proposed method.