Dealing with Problems in Data Quality for the Measurement of Mortality at Advanced Ages in Canada

Abstract

The level and age trajectory of mortality at advanced ages in Canada are not readily and exactly obtained, because of problems with the reliability of data on deaths and on population counts beyond a certain point in the official statistics. There are two ways to ensure nonetheless the termination of the life tables. One consists of finding ways to validate a sufficient number of unbiased high ages at death to produce an accurate measure with the extinct, or almost extinct, generation method. This paper presents the results of a systematic verification of ages at death and a preliminary estimation of centenarian mortality based on observations, which seems to lend credence to a leveling off of mortality rates at the highest ages for females.

Another is to establish convincing evidence as to the pattern of survival at the very highest ages; mathematical techniques can then be used to generate the rates as an extension of mortality at ages 70 to 90 or 100. Historical data were used here to give an insight on what this pattern of survival could be. Contrary to what might have been expected, the progression of mortality remains pretty much exponential until the unavoidable erratic values corresponding to the few extreme observations are reached. This entails that whatever the nature of the selections that would produce a slowing down of the rate of increase of the rates at the highest ages, they did not express themselves conclusively a few centuries ago.