Potential years of life lost at ages 25 to 74 among Métis and non-Status Indians, 1991 to 2001

Warning View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.










by Michael Tjepkema, Russell Wilkins, Sacha Senécal, Éric Guimond and Christopher Penney

As a result of a complex set of social, economic and environmental circumstances,1 Aboriginal peoples experience a disproportionate burden of disease, compared with other Canadians.2,3  Life expectancy, the most basic of health indicators, is considerably shorter for Status Indians (First Nations registered under the Indian Act of Canada)4,5 and for people living in the Inuit-inhabited areas of Canada (80% of whom are Inuit).6  But methodological challenges limit the amount of mortality information available about Métis and non-Status Indians.1 In fact, relative to their population size, these two Aboriginal groups have been under-represented in health research in general.7,8

Mortality data for Métis and non-Status Indians are difficult to generate.  Aboriginal ancestry, identity or status is not routinely recorded on death registrations.  And because most Métis and non-Status Indians do not live in areas where they constitute a high percentage of the total population, their mortality patterns cannot be studied indirectly with an area-based approach.6,9  However, the 1991 to 2001 Canadian census mortality follow-up study has made it possible to examine a wide range of mortality indicators for Métis10 and non-Status Indians.

Mortality studies typically include all ages, and are, therefore, dominated by deaths at older ages.  Results of such studies may reveal only part of the picture, especially for Aboriginal groups, who tend to have a high proportion of younger people.  Premature mortality (defined here as deaths before age 75) and potential years of life lost (PYLL) before age 75 highlight the loss to society as a result of early deaths.11  Premature mortality and PYLL rates are elevated among Status Indians,12-14 and injuries are an important contributor to these high levels,13,14 but PYLL has not been calculated for Métis and non-Status Indians.

The first objective of the present study was to examine PYLL at ages 25 to 74 by cause of death among Métis and non-Status Indians, compared with non-Aboriginal adults.

The second objective was to assess the influence of socio-economic factors on disparities in premature mortality.  Aboriginal peoples tend to rank less favourably than other Canadians on most measures of socio-economic status.15,16 Unlike other mortality studies, which incorporate few if any socio-economic factors, the census mortality follow-up study has made it possible to include many such variables. 

Methods

Data sources

The 1991 to 2001 Canadian census mortality follow-up study is a probabilistically linked cohort consisting of a 15% sample (n = 2,735,152) of the non-institutionalized population aged 25 or older, all of whom were enumerated via the 1991 census long-form questionnaire.17 This cohort was tracked for mortality from June 4, 1991 through December 31, 2001.

Names were not captured on the census database, but they were needed for linkage to the mortality database.  Consequently, creation of the mortality follow-up cohort required two probabilistic linkages.  First, eligible census respondents were linked to a nominal list (name) file (abstracted from 1990 and 1991 tax-filer data and then encrypted) using common variables such as date of birth, postal code, and spousal date of birth (if applicable); 80% of eligible respondents were successfully matched.  Second, the census plus encrypted names were matched to the Canadian Mortality Database.18  Based on 1991 deaths (which could be identified independently in the Canadian Mortality Database and/or the name file), ascertainment of deaths from 1991 to 2001 of cohort members overall was estimated to be at least 95% among those reporting any Aboriginal ancestry, Registered Indian status, or membership in an Indian Band or First Nation.

Eligibility

People enumerated by the 1991 census long-form questionnaire who had reached age 25 by census day were eligible to be part of the cohort. The long form, which was usually given to 1 in 5 households, was administered to all residents of Indian reserves, many remote and northern communities, and non-institutional collective dwellings.  However, 78 Indian reserves, representing about 38,000 people, were either not enumerated or incompletely enumerated,19 and thus, were not part of the cohort. As well, the 1991 census missed 3.4% of Canadian residents; these individuals were more likely to be young, mobile, low income, of Aboriginal ancestry,20 or homeless.

Owing to the exclusion of institutional residents and non-tax-filers, life expectancy of the cohort at age 25 was 1 year longer for men and 2 years longer for women, compared with the 1995 to 1997 life tables for all Canada. This bias would apply equally to Aboriginal and non-Aboriginal respondents and should not appreciably affect relative differences between the two groups.

Analytical techniques

The cohort was divided into ten one-year follow-up periods (June 4, 1991 to June 3, 1992; June 4, 1992 to June 3, 1993; and so on) and one seven-month period (June 4, 2001 to December 31, 2001).  Age was transformed from age at baseline (June 4, 1991) to age at the beginning of each year of follow-up.  Deaths and person-years at risk were calculated separately for each follow-up period, and then pooled by five-year age group (determined at the beginning of each year of follow-up).  Deaths before age 75 were considered premature.  The number of potential years of life lost (PYLL) was calculated by multiplying the number of deaths in each age group by the mean number of PYLL for the same age group.  For example, the death of someone aged 25 to 29 would contribute 47.5 PYLL before age 75.

To calculate rates of PYLL, the number of person-years at risk (up to age 75) was determined for each five-year age group and the rates were age-standardized to the Aboriginal population.  The Aboriginal age distribution was based on cohort members who indicated an Aboriginal ancestry, registration under the IndianAct of Canada, or membership in an Indian Band or First Nation.  Confidence intervals for the age-standardized rates were produced from variances derived using the Spiegelman method.21

Premature mortality (Cox models)

For each cohort member, person-days of follow-up were calculated from baseline (June 4, 1991) to the date of death, emigration (known only for 1991), end of study (December 31, 2001), or until that individual reached age 75.  Because exact date of birth was not available on the analysis file, age in completed years (as of June 4 of each follow-up year) was used to derive age at death and person-years of follow-up.

Cox proportional mortality hazard ratios were used to estimate the effect of socio-economic factors on the disparity in premature mortality among Métis and non-Status Indians, compared with non-Aboriginal adults.  The variables included were age, marital status (married/common-law, not married), single parent (yes, no), educational attainment (less than secondary graduation, secondary graduation, postsecondary diploma, university degree), income adequacy quintile, labour force status (in, out), crowding (more than one person per room; yes, no), home ownership (yes, no), major dwelling repairs needed (yes, no), and urban population size (1 million or more; 500,000 to 999,999; 100,000 to 499,999; 10,000 to 99,999; and less than 10,000).  Definitions of the variables are available in a previously published report17 or the census dictionary.22  All models were sex-specific.  The base model (Model 1) controlled only for age.  Models 2 to 7 controlled for age and one other variable.  The full model (Model 8) controlled for age and all other variables simultaneously.  Differences in excess mortality (1 minus the hazard ratio) comparing the full model to the base model were interpreted as estimates of the effect of the socio-economic variables on the disparities.

Cause of death

The cause of death of cohort members who died during the study period had previously been coded to the World Health Organization's InternationalClassification of Diseases, NinthRevision (ICD-9)23 for deaths occurring from 1991 through 1999, and to the Tenth Revision (ICD-10)24 for deaths occurring in 2000 or 2001. Deaths were also grouped by Global Burden of Disease categories, which underscore human development rather than the body system,25 and by risk factors, namely, smoking-related,26 alcohol-related26 and drug-related diseases,27 or deaths potentially amenable to medical intervention (for instance, deaths due to cerebrovascular diseases, hypertension, breast cancer and pneumonia/influenza).

Definitions

The 1991 census did not collect information on self-identification with an Aboriginal group (North American Indian, Métis, or Inuit).  For this analysis, Métis and non-Status Indians were defined based on two census questions reflecting ancestry and Registered Indian status.

  1. To determine ancestry, respondents were asked to which ethnic or cultural group(s) their ancestors belonged.22  From a list of 15 groups, including North American Indian, Métis and Inuit/Eskimo, respondents were instructed to check as many as applicable.
  2. Registered Indian status was determined by a direct question:  "Is this person a Registered Indian as defined by the Indian Actof Canada?" (yes, no).

In this study, respondents were considered Métis if they reported a single Métis ancestry (no other ancestries) or two or more Aboriginal ancestries with one being Métis.  People were considered non-Status Indians if they reported a single North American Indian ancestry, but were not registered under the Indian Act of Canada.

Results

Characteristics differ

The demographic and socio-economic characteristics of Métis and non-Status Indian cohort members differed from those of non-Aboriginal members (Appendix Table A).   Métis and non-Status Indians were younger and less likely to be legally married, to have completed secondary school, to be employed, and to be homeowners.  They were more likely to be in the two lowest income adequacy quintiles, to live in crowded conditions, and to live in a dwelling needing major repairs.  In 1991, nearly 7 out of 10 Métis cohort members were residents of Manitoba, Saskatchewan or Alberta, and about 7 out of 10 non-Status Indian cohort members were residents of Quebec, Ontario or British Columbia.

Table A Selected characteristics of Métis, non-Status Indians and non-Aboriginal men and women, non-institutional cohort members aged 25 to 74, Canada, 1991Table A Selected characteristics of Métis, non-Status Indians and non-Aboriginal men and women, non-institutional cohort members aged 25 to 74, Canada, 1991

Age distribution of deaths

Of the deaths of cohort members that occurred during the 1991 to 2001 period, 71% among Métis and 76% among non-Status Indians were at ages 25 to 74, compared with 48% of the deaths among non-Aboriginal people (data not shown).  And of those cohort members in this age range who died, Métis and non-Status Indians tended to be younger.   For example, about 70% of Métis and non-Status Indian men and two-thirds of Métis and non-Status Indian women who died were younger than 65, compared with around half of the non-Aboriginal decedents.

Distribution of adult PYLL

The percentage distribution of potential years of life lost (PYLL) by various causes of death differed by Aboriginal ancestry and sex.  Non-communicable (chronic) diseases ranked first, accounting for 57% and 64% of total PYLL for Métis and non-Status Indian men, respectively, and for 77% and 70%, respectively, for Métis and non-Status Indian women.  Nonetheless, these percentages were below the corresponding figures for non-Aboriginal adults (Table 1).  Cardiovascular diseases were the largest non-communicable disease subcategory for men (23% for Métis; 26% for non-Status Indians); for women, malignant neoplasms (cancer) were the largest subcategory (36% for both Métis and non-Status Indians).

Table 1 Distribution of potential years of life lost (PYLL) by cause of death at ages 25 to 74 for Métis, non-Status Indian and non-Aboriginal men and women, non-institutional cohort members, Canada, 1991 to 2001Table 1 Distribution of potential years of life lost (PYLL) by cause of death at ages 25 to 74 for Métis, non-Status Indian and non-Aboriginal men and women, non-institutional cohort members, Canada, 1991 to 2001

Injuries accounted for much higher percentages of  PYLL among the Aboriginal groups than non-Aboriginal adults.  For men, injuries made up 35% of Métis PYLL and 30% of non-Status Indian PYLL, compared with 16% of non-Aboriginal PYLL.  For women, the corresponding figures were 15%, 23% and 9%.

Rates of PYLL

Age-standardized rates of PYLL were much higher for Aboriginal than non-Aboriginal adults (Table 2). Compared with non-Aboriginal adults, PYLL rate ratios for Métis and non-Status Indian cohort members were around twice as high.  Among Métis men, rate ratios were highest in the younger age groups and lowest at ages 55 to 74.  Among non-Status Indian men, the rate ratio peaked at ages 45 to 54 and was lowest at ages 65 to 74.  The pattern was less clear for Aboriginal women, among whom rate ratios were relatively high at ages 65 to 74.

Table 2 Deaths, rates of potential years of life lost (RPYLL)* and rate ratios (RR) for Métis and non-Status Indians compared with non-Aboriginal men and women, by age group, non-institutional cohort members, Canada, 1991 to 2001Table 2 Deaths, rates of potential years of life lost (RPYLL)* and rate ratios (RR) for Métis and non-Status Indians compared with non-Aboriginal men and women, by age group, non-institutional cohort members, Canada, 1991 to 2001

Causes of death

To obtain a complete picture of mortality patterns, it is necessary to examine both relative and absolute inequalities in causes of death.  If a cause of death is rare, the relative inequality can be quite high but account for a negligible number of deaths.  On the other hand, a common cause of death can account for a large number of deaths (and be a significant contributor to absolute inequality), even if the relative risk is only slightly elevated.  Thus, measuring relative (rate ratios) and absolute inequalities (rate differences) between Aboriginal and non-Aboriginal adults highlights specific causes that might be important in developing public health programs.

For most causes,  PYLL rate ratios were elevated among Métis and non-Status Indians.  Métis men had particularly high rate ratios for rheumatic heart disease, hypertensive heart disease, unintentional injuries, and violence (Table 3).  Non-Status Indian men had high rate ratios for neuropsychiatric conditions including alcohol use disorders, digestive diseases such as cirrhosis of the liver, and road traffic accidents.

Table 3 Age-standardized rate ratios (RRs) and rate differences (RDs) for potential years of life lost at ages 25 to 74 for Métis and non-Status Indian men compared with non-Aboriginal men, by cause of death, non-institutional cohort members, Canada, 1991 to 2001Table 3 Age-standardized rate ratios (RRs) and rate differences (RDs) for potential years of life lost at ages 25 to 74 for Métis and non-Status Indian men compared with non-Aboriginal men, by cause of death, non-institutional cohort members, Canada, 1991 to 2001

PYLL rate ratios for Métis and non-Status Indian women were elevated for almost all causes.  Among Métis women, rate ratios were especially high for respiratory infections, leukemia, alcohol use disorders, hypertensive heart disease, chronic obstructive pulmonary diseases, cirrhosis of the liver, and unintentional injuries, notably poisoning (Table 4).  Among non-Status Indian women, rate ratios were high for communicable diseases, breast cancer, cervix uteri cancer, cerebrovascular disease, cirrhosis of the liver, and unintentional injuries, especially  poisoning.

Table 4 Age-standardized rate ratios (RRs) and rate differences (RDs) for potential years of life lost at ages 25 to 74 for Métis and non-Status Indian women compared with non-Aboriginal women, by cause of death, non-institutional cohort members, Canada, 1991 to 2001Table 4 Age-standardized rate ratios (RRs) and rate differences (RDs) for potential years of life lost at ages 25 to 74 for Métis and non-Status Indian women compared with non-Aboriginal women, by cause of death, non-institutional cohort members, Canada, 1991 to 2001

Rate ratios for alcohol- and drug-related diseases were higher for Métis and non-Status Indians of both sexes than for their non-Aboriginal counterparts; the rate ratio for deaths due to diseases amenable to medical intervention was high among non-Status Indian women.  The rate ratio for smoking-related diseases was significantly high for non-Status Indian men, but not women or for Métis of either sex.

Absolute inequalities as indicated by rate differences measure "excess" PYLL.  Among men, excess PYLL for all-cause mortality was 6,040 per 100,000 person-years at risk for Métis, and 5,496 for non-Status Indians (Table 3).   About two-thirds of the excess PYLL among Métis men was due to injuries (55% unintentional, 11% intentional), and one-quarter was due to non-communicable diseases, notably, cardiovascular disease (data not shown).  Among non-Status Indian men, injuries accounted for 48%, and non-communicable diseases, 45% of excess PYLL.

Excess PYLL was 3,005 per 100,000 person-years at risk for Métis women and 5,710 for non-Status Indian women (Table 4).  More than half (57%) of the excess PYLL in both groups was due to non-communicable diseases (data not shown).  Injuries accounted for 27% of excess PYLL for Métis women and 29% for non-Status Indian women; and communicable diseases, 13% and 12%, respectively.  Results differed more for specific causes of death; for example, breast cancer was a major contributor to excess PYLL for non-Status Indian women (22%), but not for Métis women.

Socio-economic factors

Compared with non-Aboriginal men, the age-adjusted hazard ratio for dying before age 75 was 1.52 for Métis and 1.76 for non-Status Indians (Table 5, Model 1).  However, socio-economic factors (such as education, income, housing, and labour force status) were important contributors to premature mortality.  Six additional models (Models 2 to 7) were run, each adjusting for age and one other socio-economic variable.  Except for Model 7, which controlled for geographic variables, the hazard ratios were attenuated compared with Model 1, suggesting that each variable had an effect on the disparity in premature mortality.  In Model 8, which controlled for age and all the socio-economic factors simultaneously, the hazard ratios were reduced from 1.52 (Model 1) to 1.11 for Métis men and from 1.76 (Model 1) to 1.28 for non-Status Indian men.  Among women, the corresponding reductions in hazard ratios were from 1.99 to 1.57 for Métis and from 2.27 to 1.83 for non-Status Indians.

Table 5 Hazard ratios for death before age 75 among Métis and non-Status Indians compared with non-Aboriginal cohort members, controlling for selected demographic, economic, housing and geographic factors, by sex, non-institutional cohort members aged 25 to 74, Canada, 1991 to 2001Table 5 Hazard ratios for death before age 75 among Métis and non-Status Indians compared with non-Aboriginal cohort members, controlling for selected demographic, economic, housing and geographic factors, by sex, non-institutional cohort members aged 25 to 74, Canada, 1991 to 2001

Discussion

The PYLL rates for Métis and non-Status Indian members of the census mortality follow-up cohort were about twice those of non-Aboriginal members.  Because this is the first study of PYLL among Métis and non-Status Indians, direct comparisons with earlier research are not possible.  However, the estimates are consistent with calculations of PYLL (at ages 1 to 74) for Status Indians in Manitoba12 and British Columbia.13  Slightly lower rate ratios in the present study could reflect the exclusion of persons younger than 25, among whom PYLL rate disparities between Aboriginal and non-Aboriginal people are greatest.28  As well, a companion article that examined PYLL among Status Indian members of the cohort reported rate ratios more than twice those of  non-Aboriginal adults (see "Potential years of life lost at ages 25 to 74 among Status Indians, 1991 to 2001" in this issue).

Cardiovascular diseases, notably ischemic heart disease, were a major contributor to total PYLL for Métis, non-Status Indian and non-Aboriginal cohort members alike.  Relative inequalities (compared with non-Aboriginal adults) for Métis and non-Status Indians were modestly elevated (ranging from 1.5 to 3.0), but due to the high incidence of cardiovascular disease deaths, they were large contributors to excess PYLL (about 10% for non-Status Indian and Métis men, and 17% to 22% for non-Status and Métis women).  These findings confirm the growing importance of cardiovascular disease in various Aboriginal populations.29,30 They also support studies showing the high prevalence of cardiovascular risk factors such as smoking,31 obesity,32 metabolic syndrome,33,34 hypertension and type 2 diabetes35 in some Aboriginal populations.

Cancer, too, was a significant contributor to total PYLL. For men, relative inequalities among Métis and non-Status Indians men were slightly elevated, which resulted in somewhat higher absolute inequalities.  For women, both relative and absolute cancer inequalities were substantially higher among Métis and non-Status Indians.  Consistent with other research,10,36-39 relative and absolute inequalities varied by cancer subtypes. 

Earlier studies have shown Status Indians to be at greater risk of dying from intentional and unintentional injuries.13,40  In this analysis, injury deaths made up a large share of total PYLL among Métis and non-Status Indian cohort members.  Both absolute and relative inequalities were significantly elevated, with relative risks ranging from 2.5 to 4.3, and injuries contributing sizeable percentages of excess PYLL.   These results illustrate the public health importance of injury prevention for Aboriginal adults.12

Compared with non-Aboriginal adults, the rates of PYLL for alcohol-related diseases were about five times higher for Métis and non-Status Indian men and more than eight times higher times for women. Moreover, these rates do not include deaths (due to injury, for example) in which alcohol may have been a contributing factor.  A British Columbia study of Status Indians also reported a high age-standardized mortality rate for alcohol-related deaths.13

The disproportionate burden of illness and death among Aboriginal peoples is thought to be the product of a wide range of social determinants that are experienced throughout the lifetime.16,41,42  The results of this analysis are similar to other population-based studies demonstrating the importance of socio-economic status as a contributor to health inequalities.10,43  Factors such as education, income, housing, and labour force status were significantly associated with the disparity in premature mortality compared with non-Aboriginal adults.  Nevertheless, these variables did not explain all of the disparity in premature death, suggesting that factors not assessed in this study contribute to the inequality.

Limitations

The results apply to non-institutional census respondents aged 25 or older who filed taxes.  This cohort is healthier than the Canadian population overall, so caution should be exercised when generalizing these results to the entire Métis, non-Status Indian and non-Aboriginal adult populations (which include institutional residents and non-tax-filers).

A question on Aboriginal self-identity was not asked on the 1991 census.  Therefore, this study used the ancestry-based definition.  It is estimated that more than 90% of individuals defined as Métis or non-Status Indians in this study would also self-identify as Aboriginal.  Nonetheless, changes in Aboriginal self-identification over the past 20 years44 mean that care must be taken when comparing these results with those of more recent censuses.

Ascertainment of deaths among Aboriginal cohort members is estimated to be slightly lower than that for the cohort as a whole.  This would be expected to exert a slight downward bias in calculated mortality rates for Métis and non-Status Indians, so the true extent of the disparities could be slightly larger than indicated here.

Due to the small number of cohort members who were non-Status Indian or Métis, confidence intervals for some causes of deaths were wide, thereby limiting the detection of statistically significant differences in PYLL between Métis or non-Status Indians and non-Aboriginal adults.

Conclusion

This study adds to the information about mortality among Métis and non-Status Indians in Canada.  These two Aboriginal groups had significantly higher rates of potential years of life lost, compared with non-Aboriginal adults.  As was the case for non-Aboriginal adults, the largest losses of potential years of life among Métis and non-Status Indians were due to non-communicable (chronic) diseases such as cancers and cardiovascular diseases.  However, injuries were a major contributor to disparities in premature mortality, as were alcohol- and drug-related deaths.  The findings highlight the prevalence of premature mortality due to chronic diseases and the public health importance of injury, alcohol and drug prevention programs.  The results are also in line with other research demonstrating the significant role of socio-economic factors.

Funding

Funding for this study was provided by the Strategic Research and Analysis Directorate of Indian and Northern Affairs Canada.  The 1991 to 2001 census mortality follow-up study was developed with funding from the Canadian Population Health Initiative, part of the Canadian Institute for Health Information.