Data: Population and Deaths Acknowledgements

5 Conclusion

The evaluation of the data quality in Canada suggests that there are some problems concerning deaths and population counts over the age of 80 years, confirming at least in part the conclusions of the study made by Kannisto [20]. However, results in section 4.1 show that Canadian data are quite good up to the age of 100 (generation heaping and age misstatement are at levels similar to those found in comparable countries) and that the main problems concern the centenarians (overstatement of age at death and errors in census age declarations).

International comparisons on the basis of two mortality indicators for the age 80 to 100 lead to the same conclusion: Canadian mortality is lower than most European countries. The best match is still with the United States.

So, we can conclude that, although there are some problems with the data, there is some strong evidence of a lower mortality at older ages in Canada. In fact, the main comparisons shown in section 4.2 concern the 80-99 age interval where Canadian data are quite good.

Can we also conclude that a North American mortality profile exists, as suggested by some authors [2] [17] [29]? This profile is characterised by relatively high mortality rates below age 65 and relatively low mortality rates above age 80, when compared to other low-mortality countries. According to our results, Canadian mortality corresponds to the latter part of this definition; however, mortality below age 65 in Canada is much lower than in the United States and more comparable to many European countries. Although Canada and the United States share the same low mortality level for the oldest-old, their mortality profiles are probably different.

Our results show that Canada and the United States also share the same types of error in their data for the oldest-old. This is not a totally surprising result, because in the two countries it is very difficult to ascertain the date of birth due to the absence of birth registers (before 1921 in Canada). However, recent findings from a record linkage study by Hill et al. [16] show that the American data concerning the age at death of older persons are quite reliable, giving more support to the lower level of U.S. mortality rates above age 80 (at least among native whites).

The same kind of matching study would be useful for Canada. In the meantime, a first step in improving estimation of mortality at advanced ages would be for all Canadian provinces to ameliorate the data on deaths; as there was no civil registration before 1921 in Canada, the task is not easy. However, some reconciliation can be done with administrative files (health records, for example). Another avenue would be to estimate mortality using only data from the Province of Quebec, where death registration is very good and age at death can be verified against baptism register before 1926 [5] [11]. Quebec has the added advantage of a universal health insurance program (Medicare) with a file on all beneficiaries. This would enable us to compare data from vital statistics and from the census to this file. These procedures would lead to a better knowledge of the mortality profile at old ages.


Data: Population and Deaths Acknowledgements

logo70.gif (2450 bytes)

Mortality Statistics for the Oldest-Old: An Evaluation of Canadian Data
Robert Bourbeau and André Lebel
© 2000 Max-Planck-Gesellschaft ISSN 1435-9871