Component of Statistics Canada Catalogue no. 82-003-X Health Reports Article Income disparities in health- adjusted life expectancy for Canadian adults, 1991 to 2001 by Cameron N. McIntosh, Philippe Finès, Russell Wilkins and Michael C. Wolfson November, 2009 Statistics Canada, Catalogue no. 82-003-XPE • Health Reports, Vol. 20, no. 4, December 2009 55 Income disparities in health-adjusted life expectancy for Canadian adults, 1991 to 2001 • Research Article Income disparities in health- adjusted life expectancy for Canadian adults, 1991 to 2001 by Cameron N. McIntosh, Philippe Finès, Russell Wilkins and Michael C. Wolfson Abstract he dramatic increase in life expectancy in Background Canada and other economically developed Health-adjusted life expectancy is a summary T measure of population health that combines mortality and morbidity data into a single index. nations during the last century stands as testimony This article profi les differences in health-adjusted life expectancy across income categories for a to the success of improvements in public health and representative sample of the Canadian population. 1 Data and methods advances in medical care. But despite these gains Mortality data were obtained from the 1991-2001 Canadian census mortality follow-up study, in longevity, inequalities in health outcomes across which linked a 15% sample of the 1991 adult non-institutional population with 11 years of death records from the Canadian Mortality Data Base. different subpopulations are still pervasive in Canada Information on morbidity was obtained from the 2-6 Health Utilities Index Mark 3 instrument on the and other industrialized countries. 2000/2001 Canadian Community Health Survey. The Sullivan method was used to compute health- adjusted life expectancy for national deciles of population ranked by income. Main results Irrespective of how socio-economic requires summary measures refl ecting the For both sexes, and with few exceptions, a nearly status is defi ned and measured (by effects of differences in both mortality linear gradient across income deciles emerged income, educational attainment or and morbidity across socio-economic for health-adjusted life expectancy at age 25. 13-15 Compared with people in higher-income deciles, occupational prestige), mortality rates categories. Accordingly, a variety those in lower-income deciles had fewer years of show a gradual but systematic increase of summary measures, which integrate health-adjusted life expectancy. These disparities were substantially larger than those revealed by at successively lower levels of the socio- information on mortality and morbidity life expectancy alone. economic hierarchy. Nonetheless, into a single numeric index, have been Interpretation time-series analyses for Canada suggest developed.16-18 These fi ndings highlight the generally worse that, in absolute terms, socio-economic A number of studies in Canada have health-related quality of life of lower-income groups. The results demonstrate that differences in mortality have been calculated summary measures in order assessments of socio-economic disparities in diminishing over recent decades, with to examine broad national patterns of health should include the effects of both mortality 19 and morbidity. the specifi c patterns dependent on sex mortality and morbidity, regional and the cause of death considered.7,8 variations in population health,20 the Keywords Mortality, however, is only one population health impact of specifi c health inequalities, Health Utilities Index, life 21-25 expectancy, socio-economic, Sullivan method aspect of population health. It is now diseases and risk factors, and whether widely recognized that information population health has been improving Authors on morbidity (including disability and over time.7,26,27 Few investigations in Cameron N. McIntosh (1-613-949-4135; [email protected]) is with Public reduced health-related quality of life) the Canadian context have focused on Safety Canada. Philippe Finès (1-613-951-3896; is crucial for monitoring health trends, socio-economic differences in summary [email protected]) and Russell Wilkins setting priorities, and conducting cost- measures of population health, and (1-613-951-5305; [email protected]) are with the Health Analysis Division at Statistics effectiveness analysis of population- those that have done so have relied Canada. Michael C. Wolfson (1-613-951-8216; level interventions. Because morbidity primarily on macro-level indicators [email protected]) is with 2,3,9-12 Statistics Canada, Ottawa, Ontario, K1A 0T6. varies by socio-economic status, a (such as neighbourhood income) rather complete assessment of health disparities than micro-level indicators (such as household income),28-30 thereby almost 56 Health Reports, Vol. 20, no. 4, December 2009 • Statistics Canada, Catalogue no. 82-003-XPE Income disparities in health-adjusted life expectancy for Canadian adults, 1991 to 2001 • Research Article certainly attenuating the association The Canadian Community Health Canadian Community Health Survey, between socio-economic status and Survey income deciles were constructed in the health. One Canadian study used micro- Data on health-related quality of life same manner, except that total household level indicators for analyses of socio- were derived from the 2000/2001 income was used rather than total economic disparities in health, but only Canadian Community Health Survey economic family income. 31 within a single province. (cycle 1.1). The Canadian Community Because of the high proportion of This article examines socio-economic Health Survey is an ongoing cross- missing income data on the Canadian differences in health for a nationally sectional survey that collects Community Health Survey (21.5%, representative sample of the adult information on health status, health compared with 1.4% for the census), population of Canada, using a summary determinants, and health care utilization. the extent of possible bias was addressed measure known as health-adjusted life It is representative of the household by performing two sets of analyses. The 32 expectancy. By weighting years of population aged 12 or older in all fi rst set completely excluded cases with life according to their quality, health- provinces and territories, excluding missing income deciles, and the second adjusted life expectancy converts the residents of institutions, of Indian set was conducted after the missing decile conventional, purely mortality-driven Reserves and Crown Lands, of Canadian information was fi lled in using hot deck 37 life expectancy measure into expected Forces bases (military and civilian imputation. In the latter procedure, equivalent years of full health. This residents) and of certain remote regions, groups of Canadian Community Health study combines mortality data from the and full-time members of the Canadian Survey respondents matched on age, 1991-2001 Canadian census mortality Forces. The response rate for cycle sex and educational attainment were 33 follow-up study with information about 1.1 was 84.7%. For consistency with created. For the cases with missing health-related quality of life from the the Canadian census mortality follow- income deciles in each of these groups, a 2000/2001 Canadian Community Health up study, the present investigation decile value was randomly selected from Survey to estimate health-adjusted life was restricted to respondents aged the pool of complete cases (the “donor” expectancy for different income groups. 25 or older (n = 106,283). More cases). Because the hot deck imputation detailed descriptions of the Canadian did not appreciably alter the results (data Data and methods Community Health Survey design, not shown), all of the analyses presented sample, and interview procedures can here are based on cases with originally Data source be found in other published reports and complete income data from the Canadian The 1991-2001 Canadian census on the Statistics Canada website.34,35 Community Health Survey. mortality follow-up study Death data were obtained from the Defi nitions Analytical techniques 1991-2001 Canadian census mortality Income deciles Health-adjusted life expectancy follow-up study, conducted by Statistics For the census mortality linked fi le, Health-adjusted life expectancy was Canada in collaboration with the deciles (tenths) of population ranked by estimated using a modifi ed version Canadian Population Health Initiative. income adequacy were created. First, of the Sullivan38 method. Chiang’s39 With probabilistic linkage techniques, for each economic family or unattached method was used to calculate abridged a 15% sample (n = 2,735,152) of the individual in the non-institutional census (fi ve-year age group) period life tables, non-institutional population aged 25 or population of all ages, total pre-tax, corresponding standard errors, and 95% older who completed the 1991 census post-transfer income from all sources confi dence intervals for each population long-form questionnaire (the cohort) was was pooled across all economic family subgroup of interest (by age, sex and matched to 11 years of death records members and divided by the weighted income decile). Before computing (June 4, 1991 to December 31, 2001) family size (or “equivalent person the life tables, it was necessary to: (1) from the Canadian Mortality Data Base. unit” scale). An economic family is a transform baseline age to age at the The linked fi le contains information on