9 Mortality in metropolitan areas ○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○○ Heather Gilmour and Jane F. Gentleman Abstract Objectives This article examines differences in all causes mortality ig cities generally have a bad reputation. Pollution, rates and rates for the leading causes of death (heart disease, cancer and cerebrovascular disease) by noise, high stress levels, lack of space, and a hectic census metropolitan area (CMA). Data source pace can all take a toll on an individual’s health. The data are from the Canadian Vital Statistics Data B Base maintained by Statistics Canada. Yet if mortality rates are any indication, the health of urban- Analytical techniques dwellers varies sharply, depending on which city they call Annualized age-standardized mortality rates were calculated for Canada and for each CMA for the three- home. year period from 1994 to 1996. Differences between the CMA rates and the national rate were examined. Regional variations in mortality rates have been used as Main results Mortality rates tend to be high in CMAs in the Atlantic evidence of the importance of the social and physical provinces and Québec and low in CMAs in the Prairies 1 and British Columbia. Ontario contains CMAs with environment to public health. The spatial distribution of some of the highest mortality rates in Canada, as well as others whose rates are among the lowest. The mortality rates may suggest the need for case detection and pattern of mortality for specific causes also differs within CMAs: a CMA may have a high death rate for one treatment programs, services and facilities. As well, to some cause, but a low rate for another. degree, geographically based data indicate what is achievable. Key words cause of death, death rate, urban health, heart That is, particularly low mortality in one area suggests that diseases, neoplasms, cerebrovascular disease improvements are feasible in regions where rates are Authors 2 Heather Gilmour (613-951-6610; [email protected]) elevated. is with the Canadian Centre for Justice Statistics at Statistics Canada, Ottawa K1A 0T6. Jane F. Gentleman Geographic variations in death rates have long been (301-436-7085; [email protected]), formerly with Statistics Canada, is currently with the Division of Health recognized. Decades ago in the United States, because of Interview Statistics, National Center for Health Statistics, Hyattsville, Maryland. high cerebrovascular mortality rates, parts of the South became known as the “stroke belt.”3 In England and Wales, a gradient in mortality rates for most causes has been observed: high in the north and west, low in the south and east.4 Canada, too, has a geographical gradient in Health Reports, Summer 1999, Vol. 11, No. 1 Statistics Canada, Catalogue 82-003 10 Urban mortality mortality, with higher rates in the Atlantic provinces The demographic, socioeconomic and physical and Québec than in the Prairie provinces and British characteristics of CMAs differ. In 1996, populations Columbia.1 ranged from 125,600 in Thunder Bay to 4.3 million To a considerable extent, this east-to-west gradient in Toronto. Immigrants made up substantial shares may be strongly influenced by mortality rates in of the populations of Toronto and Vancouver.5 In census metropolitan areas (CMAs). CMAs are large Winnipeg, Saskatoon and Regina, Aboriginal people urban centres having at least 100,000 inhabitants in constituted a larger proportion of residents than their central core. In 1996, 62% of Canadians lived was the case in other CMAs.6 The industries that in Canada’s 25 CMAs, and CMAs accounted for 57% form the economic base of each CMA vary as well. of deaths that occurred in the 1994-1996 period. For example, Calgary has long been the Methods Data sources (see Definitions). However, each CMA contains neighbourhoods Mortality data are from the Canadian Vital Statistics Data Base, whose social, economic and health characteristics vary widely. Thus, which compiles vital statistics submitted by the offices of vital high or low mortality rates in specific parts of a CMA may be masked statistics in each province and territory. Intercensal population by the rates in the rest of the CMA.7 estimates by age, sex and census metropolitan area (CMA) for Because CMA boundaries must respect the administrative 1995 were used to calculate the three-year average rates. The boundaries of census subdivisions (CSD), some CMAs include CSDs decedent’s place of residence, not the place where the death with large amounts of sparsely settled territory, and only the occurred, was used to determine death rates for each CMA. population closest to the urban core has a close relationship with 8 Analytical techniques that core. For most diseases, incidence rates provide the best measure of The 1991 population of Canada (all ages) was used as the standard risk.7 It is unclear how reliably mortality rates can be used as a population for calculation of age-standardized mortality rates. All measure of risk of disease in particular CMAs.7 mortality rates were age-adjusted using the direct method. Age- The analysis excludes Prince Edward Island, the Northwest adjustment means that the rates are comparable across CMAs, Territories and the Yukon, which have no CMAs. However, any despite local variations in age distribution. The standard population analysis below the provincial/territorial level would be difficult in these was not disaggregated by sex. It is, therefore, possible to compare regions because their death counts are low. age-standardized rates for males with age-standardized rates for The 1991 Census population counts were adjusted for net females. Although mortality rates refer to the total population (from undercoverage and for non-permanent residents. Subsequent age 0), for readability, the terms “men” and “women” rather than investigation by officials at Statistics Canada revealed that the “males” and “females” are used in this article. adjustment overcompensated for the undercount, resulting in figures Comparisons between areas may reflect random variation rather that were too high. Therefore, population figures for 1986 to 1991 than real differences. Confidence intervals were calculated to are being re-estimated by Statistics Canada. Mortality rates in this assess the variation of each CMA’s mortality rate. Two-sided tests article were calculated before the revision at the CMA level, and were performed to identify statistically significant differences thus may be slightly low. However, the impact of such adjustments between the age-adjusted rate for each CMA and the age-adjusted should be small, and the underlying patterns should be similar, even national rate. Because the mortality rates for large CMAs can after revision. influence the national rate, these rates cannot be assumed to be Because the law requires all deaths to be reported, the registration independent of the national rate. To account for the degree of of deaths is considered virtually complete. Nonetheless, there are correlation between a given CMA’s mortality rate and the national differences in diagnostic practices and coding procedures among rate, estimated covariances were calculated between the two rates provinces. Consequently, the specific cause of death category to and were used in the calculation of the variance of the difference which a given death is attributed may vary from one CMA to another. between rates. As well, a small number of late registrations may result in some Limitations underestimation of rates. The data in this analysis should be interpreted with caution. CMAs are defined to represent economically and socially integrated areas Health Reports, Summer 1999, Vol. 11, No. 1 Statistics Canada, Catalogue 82-003 Urban mortality 11 administrative hub of the country’s oil and gas Saint John (New Brunswick) and Halifax (Nova industry.9 With seven degree-granting institutions, Scotia). Halifax is Atlantic Canada’s headquarters for All causes mortality rates in 1994-1996 were above education.10 Ottawa-Hull, which encompasses the the national level in each CMA, except for men in national capital, is the only CMA to cross provincial Halifax (Chart 1, Appendix Table A). In St. John’s, boundaries. this was the result of high mortality for heart disease, The unique character of CMAs extends to the cancer (excluding lung) and cerebrovascular disease death rates of their residents. Even in the same for both sexes, and lung cancer for men (Charts 2 province, differences between CMAs can be to 5). Saint John, too, had high heart disease and pronounced. And within a single CMA, the death lung cancer mortality rates, although rates for other rate for one cause may be well above the national cancers and cerebrovascular disease did not differ level, while the rate for another cause is below it. from the national level. Halifax residents had high This article focuses on three years of data (1994 lung cancer mortality rates, and among women, the to 1996) from the Canadian Vital Statistics Data Base rate for other cancers was also high. However, to analyze mortality patterns in Canada’s 25 CMAs women in Halifax had a low mortality rate for (see Methods and Definitions). Age-standardized cerebrovascular disease. mortality rates for men and women for all causes of death are examined, as well as rates for the three leading causes: heart disease, cancer, and cerebrovascular disease (stroke). Lung cancer, the Definitions leading type of cancer, influences overall patterns All causes contributing to a death are entered on the death of cancer mortality and, therefore, is analyzed certificate in accordance with the Ninth Revision of the International separately. Classification of Diseases (ICD-9).12 A single underlying cause of Death rates are the ultimate outcome of a death is coded. The following ICD-9 codes were used for this multitude of factors: socioeconomic, environmental, article: all causes (001-E999), lung cancer (162), other cancer medical, and lifestyle. This article is a descriptive (140-208, excluding 162), heart disease (391, 392.0, 393-398, analysis only.