The contribution of familial component in Sardinian longevity: measuring unobserved heterogeneity

The aim of this contribution is to investigate further the possible determinants of the exceptional level of longevity detected among Sardinian population in the Blue Zone (Poulain et al. 2004). In particular the focus of this study is on contribution on individual survival of the familial component, which is operationalized considering a set of variables. The village of Villagrande Strisaili is chosen for detailed analysis as for this municipality it is available a family reconstructed database build following survival of newborns from 1866 to 1915 until 2006 (VILD, Salaris 2009). The analysis of the data investigates the role of familial variables on longevity, estimating mortality trajectories aware of the changes occurring in the composition of the birth cohorts due to selective mortality at early ages (Vaupel and Yashin 2001). Unobserved heterogeneity and the existence of subpopulations within the population of Villagrande Strisaili are therefore considered by means of multilevel and frailty models for survival data (Wienke 2003, Skrondal and Rabe-Hesketh 2004). The analysis considers that there is dependency between single units as different individuals are at the same time members of the same family, and this is particularly true in a village as Villagrande where the rate of geographical endogamy in the past was considerably high. Introduction The familial transmission of longevity is a complex subject of study as it has two distinct interpretations. There is in fact: i) a familial genetic component, meaning that the favourable traits for survival are inherited as well as part of the shared family DNA; ii) a familial environmental component, which refers to the influence that specific characteristics of the household, shared familiar environments, daily habits and so on, have on individuals’ survival. Specific traits favourable to longevity may be directly detectable through the analysis of individual data such as blood sample or other biological material, or might also be indirectly investigated looking at survival resemblance, integrating individual information with that of parents and siblings. However, a variable such as family can not be exclusively considered as a genetransmitter, as it also represents the social environment where individuals lived the first part of their lives until they formed a new family. For this reason, disentangle the contribution on survival of the two familial components is not an easy task.