1. Introduction

The present book is intended to be an additional contribution to the description of the mortality of the oldest-old in the postwar era, begun with "Development of Oldest-Old Mortality, 1950-1990" by the same author. While the first publication was essentially a time series study of the momentous changes that took place in the mortality of old people in advanced countries, their timing and magnitude, their acceleration or deceleration or absence of either, in each of the 28 countries examined, the present one is a life table analysis of the same countries in the same period, less precise on timing and more sharply focused on age. Instead of annual fluctuations in 5-year or broader age groups, dealt with in the former, the present study examines mortality by single years of age. To do it successfully, the data are combined into 10-year periods in order to secure meaningful numbers of observations at ages where they are few. This has, in fact, allowed calculation of fairly robust death rates up to age 109 for an aggregate of several countries.

        In the period 1950-1990, the developed countries experienced an unprecedented and entirely unexpected decline in old age mortality as documented on a broad front in the above-mentioned publication. Division of this period into four decades and preparation of life tables for each of them yields only a rough time series but provides a sharper view of the levels and changes in mortality by age. As data for the 1950s are unavailable for some countries and since the change from then up to 1960s was relatively small, the dimensions and nature of the changes that have so far occurred can perhaps best be seen when comparing the 1960-70 and 1980-90 life tables. This will be frequently done.

        Both studies spring from the Kannisto-Thatcher Oldest-Old Database which is incorporated in the Odense Archive of Population Data on Aging. All the countries comprised in the database, 28 of which have been included in both these studies, are industrialized and can be considered in the global context to be low-mortality countries because the life expectancy at birth for both sexes combined in all of them is at least 69 and in most cases 75 years. This allows, however, significant differences in mortality levels, and the results will in some connections be presented separately for what are termed low, medium and high mortality groups of countries, so classified according to level of oldest-old mortality. It should be stressed that all of these are nevertheless advanced countries with relatively low death rates.

        The geographic limitation of the database to essentially low-mortality countries was by no means planned but was reluctantly accepted for the reason that we have not found reliable data on age at death among the in the national statistics of developing countries.

        For the calculation of any kind of death rate, both the number of deaths and the size of the population at risk are needed. It may be surprising but it is nevertheless a fact that in most of the advanced countries of the world, including all the larger ones, the numbers of population at very high ages are not well known. Censuses regularly produce figures which at ages approaching or exceeding one hundred years are obviously overstated and, when examined over time, incompatible with data on deaths. This realization, together with the finding that the age information on death certificates was in many of the same countries quite accurate, prompted Paul VINCENT to pioneer his method of extinct generations (Vincent, 1951) in which the populations at risk are obtained by summing up the deaths in each cohort beginning with the oldest. Today, more than 40 years later, it is still the method of choice for the study of mortality at very high ages in all countries except the few which maintain reliable permanent population registers.

      A third study which is currently under preparation, will use the same database to attempt to draw and define the trajectory of mortality as far up as is at present possible (Thatcher et al., forthcoming).

Updated by V. Castanova, 1 November 1999