Survival analysis on mortality data at oldest ages: First results on longevity pioneers in France

Studying mortality at extreme old ages has been very challenging, mostly because data of good quality are sparse. Decades of hard work of many research teams offered a new type of data on deaths at oldest ages where validated information at individual level are at disposal, which allows us to make use of methods that could not be of use otherwise. Following Barbi et al. (2018), we adopt herein the same analysis, using proportional hazard model on up-to-date French data on deaths at age 105 onwards, to study the evidence for the existence of a plateau of human mortality in France, as it was stated to be proven in Italy. As results, we find a statistically significant and positive Gompertz slope parameter, suggesting that mortality keeps increasing after age 105 instead of being constant. We also find significant effect of sex but no cohort effect on mortality of French semi-supercentenarians