The old age life tables published in this book were constructed from cohort observations, specified by country, sex, age and year, each with the number exposed-to-risk and the number of deaths, and straddling two calendar years. About 50,000 such double observations were used.
The first life table presented here (Table 4) summarizes the contemporary human longevity in advanced countries. The greatest interest may be centred on the very oldest of whom substantial numbers are now on record. However, they contribute very little to the entire old age experience. Of the 6.27 years of life remaining for an 80-year-old man, only 0.01 year (T100 = 0.0107) is lived above age 100, and of the 7.83 years of a woman, 0.03 year. Continuing to play with numbers and calculating survival with greater precision we find that, divided equally between all 80-year-old women, the time to be lived above age one-hundred-and-ten amounts to 26 minutes for each. A closer examination of this and other life tables is the subject of the following chapters.
Slightly more than one hundred decennial life tables for individual countries are given in Annex Table 2 and a few tables for aggregates in Annex Table 3. For reasons of space, the columns Lx and Tx are omitted from these since, if needed, the user can calculate them. Any resulting inconvenience is regretted.
When probabilities of dying cannot be calculated above age 99, as is the case with several countries of the database, the life expectancy at age 100 needs to be estimated. Basing on evidence from countries with complete data, the following approximation was found reasonable:
and was applied to "short" life tables for the calculation of:
Comparison with official life tables
The validity of the life tables produced from the Odense database was checked by comparing them with official life tables of the same countries. Available for comparison purposes was a large file in the United Nations Statistical Division in New York, containing life expectancies at various ages in a very large number of countries around the world since the beginning of the century. Among them, were life expectancies at age 80 according to nearly one thousand life tables from 125 countries. About two hundred e80 values for each sex in 26 countries are listed in Annex Table 4 and were used in the comparison.
Although regular demographic statistics are usually not reliable enough for the calculation of death rates or life expectancies after age 90 or 100, they are in most Western countries good for calculating it at age 80 with sufficient precision. This is because only 10 to 15 percent of the lifetime after age 80 is lived after 90, and less than one percent after age 100. The other side of the coin is that the comparison cannot validate any calculations regarding centenarian life.
Comparison of life expectancies at age 80 in official life tables with values derived from Odense Data is nevertheless a meaningful operation regarding at least octogenarians and may corroborate the validity of both sources or reveal weaknesses in one or the other.
It should be borne in mind that makers of official life tables are expected to treat their data with caution and have considerable freedom in adjusting the results to what they believe to be a good approximation of reality. In so doing, they frequently use smoothing or extrapolation rather freely or apply a theoretical model of their choice. All in all, there is a good deal of subjective choice involved regarding the values at high ages. The Odense life tables, on the other hand, are constructed adhering strictly to the official data on deaths and leave room for subjective input only in the estimation of current survivors. We have made extensive corrections to official data on survivors at ages approaching or exceeding 100 years but they have relatively little impact at age 80.
In cases where age overstatement was considered serious, population data even for octogenarians were corrected in Odense to conform with survival data but when also the latter suffered from age exaggeration, mortality remained artificially low and life expectancy too high.
A comparison was carried out
graphically for 26 countries in the period 1950-1990 and is shown in Figure
4 on seven pages. The official data are expressed with an 0 in the single year or
central year of the life table period. The basic data are given Annex
Table 4.
The following interpretation is suggested.
a) Good agreement between the two sources:
| Austria | Germany, West |
| Belgium | Japan |
| Denmark | Netherlands |
| England & Wales | Norway |
| Finland | Scotland |
| France | Sweden |
| Switzerland |
b) Good agreement but insufficient evidence:
| Australia |
| Estonia |
c) Good agreement but some volatility:
| Iceland, due to small numbers |
d) Official life tables more cautious suggesting awareness of age overstatement in deaths:
| Ireland, in all periods |
| Italy in 1950s only but strongly |
| New Zealand, until the most recent |
| Portugal, until about 1980 |
e) Official life tables too optimistic, inconsistent with data on deaths:
| Czechoslovakia, all periods with few exceptions |
| Germany (East), in two tables of the 1960s; |
| otherwise good agreement |
| Hungary, in about half of the cases |
| Poland, in female tables |
f) Likelihood of age overstatement in deaths:
| Spain, suggested by more cautious official tables in 1950 and 1960 and by high values in the 1980s. |
g) Evidence insufficient
| Luxembourg |
Of the countries constituting the prime group of thirteen, eleven are in group a) (good agreement), one in c) (good agreement but volatility) and one, Italy, in d) (age overstatement in 1950s). Age overstatement is suggested in the six countries of groups d) and f). Other evidence suggests it as present also in Estonia and Poland.
The findings corroborate generally very well the results of the data quality assessment in Chapter 2.