Supercentenarians: slower ageing individuals or senile elderly?

Abstract

Although the increase in the number of centenarians is well documented today, at least in some countries, this is still not the case for people having reached the age of 110 years or more: the supercentenarians. The supercentenarians emerged in the mid-1960s. Their numbers have regularly increased since the mid-1970s. The current prevalence of known supercentenarians in countries involved in the database is approximately five to six times more than in the mid-1970s. In roughly 20 years the maximum age observed has increased by about 10 years from 112 to 122 years. The annual probability of death at age 110 is as low as 0.52 with the validated data (n=106) or with the exhaustive and validated data (n=73). The probabilities of death stagnate between 110 and 115 years, and all the computed probabilities fall below the ceiling of 0.6. Our results are compatible with the last extrapolations of mortality trajectories using a logistic or a quadratic model.

Introduction

Beyond the age of 100 years, death rates fall far below the Gompertz trajectory (Vaupel et al., 1998, Thatcher et al., 1998, Thatcher, 1999b). This paper is a first attempt to answer the question of whether the supercentenarians (110-year olds) are a model of slow ageing or whether, alternatively, they are frail people whose resistance to environmental hazards is very weak thus providing us with a new measure of the quality of the environment that humans are now bringing about (Robine, 2001).

The international database on supercentenarians, established in Rostock and Montpellier at the beginning of 2000, is supplied with two types of data: (i) data gathered on the Internet by Louis Epstein and Robert Young from the USA; and (ii) data specifically collected for this project in Canada by Bertrand Desjardins and Robert Bourbeau (University of Montreal), in Belgium and the Netherlands by Dany Chambre and Michel Poulain (University of Louvain), in England and Wales by Roger Thatcher (Thatcher, 2001) in France by France Meslé, Jacques Vallin (INED) and J.-M. Robine (INSERM), in Japan by Yasuhiko Saito (Nihon University), in Italy by Graziella Caselli and Viviana Egidi (Universita di Roma ‘La Sapienza’), and in Denmark, Finland, Norway and Sweden by Bernard Jeune (University of Odense). More details on these data are provided in Appendix A. In August 2000 the data contained 169 records with at least the year of birth and the year of death. Table 1 shows the age distribution of this new segment of the population.

Although the increase in the number of centenarians is well documented today, at least in some countries (Jeune and Vaupel, 1995), this is still not the case for people having reached the age of 110 years or more: the supercentenarians. Case validation is extremely important for these individuals (Jeune and Vaupel, 1999). Most of the data used in this analysis have been carefully validated, especially those concerning the oldest old having reached the age of 115 years or more (Laslett, 1994, Wilmoth et al., 1996, Robine and Allard, 1998, Desjardins, 1999).

Out of the 169 records already collected in the database, we have excluded eight individuals, all allegedly born before 1840 but with poor evidence of the date of birth. We have also excluded eight individuals acknowledged as false cases such as the famous Canadian, Pierre Joubert, born in 1701 (Charbonneau, 1990), or with incomplete data such as the Japanese man, Shigechiyo Izumi, born in 1865 and judged to be a false case by the Japanese scholars (letter to Jacques Vallin), leaving a sample of 153 records with minimal requirements for analysis. Table 2 shows the distribution of the supercentenarians according to their country of birth. The table also indicates the maximum age reached in each country and the total sex ratio. Among the supercentenarians, the sex ratio is one male for 10.8 females. Because there is evidence that the male supercentenarians are better identified and that they are relatively more numerous among false cases the number of females per male may actually be even greater. It is noteworthy that the sex ratio is much lower among the 16 excluded cases, confirming an assumed male specific exaggeration. Working under these minimal criteria, the database is currently gathering data from 17 countries although the degree of completeness varies considerably from one country to another.

Adding up the numbers of supercentenarians in the countries with exhaustive and validated data and dividing the total by the total population in these countries in 1999 (Levy, 1999), we have computed a crude ratio of supercentenarians per million inhabitants in the combined population of the seven countries with exhaustive and validated data. An estimation of the expected cumulative number of known supercentenarians (dead and alive) in each country is calculated by multiplying this ratio by the current population in each country. These calculations give an estimation of the degree of completeness of the data for the countries with incomplete data (see Table 2).