Young adults' excess mortality: individual reality or yet another heterogeneity's ruse?

Adrien Remund, University of Geneva

Young adults’ excess mortality (YAEM) is a long known demographic fact, but its causes remain unclear. The common perception of YAEM presents it as an inevitable phenomenon stemming from the endogenous development of adolescents and the natural turmoil associated with puberty. Although this point of view has been challenged by psychologists and anthropologists, it has retained credit, not only in the eyes of the public, but also amongst scholars from other fields. This endogenous hypothesis corresponds in demographic terms to what Vaupel calls a level 1 explanation, i.e. that "the observed change is produced by a corresponding change at the individual level". However, other mechanisms could cause a hump in the force of mortality. Notably, YAEM could be produced by unobserved heterogeneity under the condition that the overall population is composed of at least two subpopulations displaying diverging forces of mortality. This would correspond to a level 2 explanation, namely that the force of mortality observed at the population level is an artifact of a change in the structure of the population. In order to test the validity of this heterogeneity hypothesis, I developed a discrete frailty model capable of generate YAEM as observed in real populations. Due to the highly nonlinear nature of this model, alternative optimization techniques are tested, among which Differential Evolution algorithms are the most efficient. The success of this procedure suggests that the heterogeneity hypothesis is plausible since YAEM follows from its premises. Further work on individual records should allow to test whether these premises are true and thus whether the heterogeneity hypothesis is sound. Social inequalities in certain causes of death such as (traffic) accident are likely candidates to generate such heterogeneity.