Introduction, findings, and conclusions

Introduction

Health disparities between Aboriginal and non-Aboriginal populations have been documented in Canada^1-7 and internationally,^8,9 the consequences of which includepremature mortality among those with Aboriginal origins.^1,10-13 While efforts have been made to report health indicators that are comparable for Aboriginal and non-Aboriginal populations,¹⁴ and indices of community well-being have been developed,¹⁵ deficiencies in health surveillance for Aboriginal people persist.^2,16

Survey data are often used to examine the health status of Aboriginal people, but because of limited coverage and small samples, information is rarely available by First Nations, Métis and Inuit identity. Studies tend to report pan-Aboriginal health characteristics, and thereby, may mask differences across Aboriginal groups. As well, the extent to which different health determinants contribute to health status may vary by Aboriginal identity group.

Hospital discharge records are indicators of serious morbidity and include populations not regularly covered by national health surveys. However, hospital records do not have consistent information on the Aboriginal identity of patients. Some jurisdictions have attempted to develop more comprehensive health surveillance for Aboriginal people using linked administrative records,^3,17 but employing those methods for analysis at the national level is currently not feasible.

This report combines 2001 Census data with hospital discharge records to compare hospitalization of residents of dissemination areas with a relatively high versus a low percentage of Aboriginal people¹² Hospitalization rates for high-Aboriginal dissemination areas are calculated by predominant Aboriginal identity: First Nations, Métis or Inuit.

While it is important to measure and document differences in hospitalization rates between Aboriginal and non-Aboriginal groups, it is equally important to understand factors associated with these differences. Socio-economic characteristics account, to some degree, for the higher prevalence of poor health in Aboriginal populations.⁵ This analysis uses an ecological approach, based on the assumption that the social, cultural and environmental characteristics of a neighbourhood can influence health status.¹⁸ The goal is to determine if the differences persist to a greater or lesser extent when both individual characteristics and area-level socio-economic conditions are taken into account.

Findings

Distribution of dissemination areas by high-/low-Aboriginal classification

According to the 2001 Census, 2,066 dissemination areas (DAs) (5.1% of all DAs) were high-Aboriginal, that is, at least 33% of the population reported an Aboriginal identity (Appendix A, Table A). The percentage of Aboriginal people in these DAs ranged from 33% to 100% (Figure 1). The total population of these DAs was 471,130 (AppendixA, Table B).

Table A
Distribution of 2001 dissemination areas, by high†- and low-Aboriginal classification, predominant Aboriginal identity and jurisdiction, Canada, 2001

Table B
Weighted distribution of population in high†- and low-Aboriginal dissemination areas, by predominant Aboriginal identity and jurisdiction, Canada, 2001

Figure 1
Location of high†- Aboriginal dissemination areas, by percentage Aboriginal, Canada, 2001/2002

The predominant Aboriginal identity was First Nations in 1,862 of these DAs; Métis in 135; and Inuit in 69. A quarter (25%) of the population in DAs with predominantly First Nations identity were in Manitoba, followed by Saskatchewan (23%), British Columbia (16%), Alberta and Ontario (each 13%) (data not shown). Saskatchewan (29%), Alberta (28%) and Manitoba (20%) had the largest percentages of the population in DAs with predominantly Métis identity . Predominant Inuit identity DAs were located primarily in Nunavut, with a smaller percentage in the Northwest Territories and Newfoundland and Labrador (Figures A, B and C).

Figure A
Distribution of high¹-Aboriginal predominantly First Nations dissemination areas, by proportion Aboriginal, Canada, 2001/2002

Figure B
Distribution of high¹-Aboriginal predominantly Métis dissemination areas, by proportional Aboriginal, Canada, 2001/2002

Figure C
Proportion of high¹-Aboriginal predominantly Inuit dissemination areas, by proportional Aboriginal, Canada, 2001/2002

Hospitalization rates elevated

Hospitalization rates of residents of high-Aboriginal DAs were almost twice those of residents of low-Aboriginal DAs (Figure 2). Rates were highest for DAs where the predominant Aboriginal identity was First Nations: 1,764 hospitalizations per 10,000 population, compared with 925 for low-Aboriginal DAs. For DAs where the predominant identity was Métis, the rate per 10,000 was 1,452, and for predominantly Inuit DAs, 1,230.

Figure 2
Unadjusted hospitalization rates of residents of high†- and low-Aboriginal dissemination areas, by predominant Aboriginal identity, Canada excluding Quebec, 2001/2002

Patients from high-Aboriginal DAs tended to be younger than those from low-Aboriginal DAs (Figure 3). The median age of patients from high-Aboriginal DAs was 36 years, compared with 53 years for patients from low-Aboriginal DAs. Patients’ median age ranged from 26 years for those from predominantly Inuit DAs to 42 years for those from predominantly Métis DAs. These medians reflect the younger age profile of the Aboriginal population, attributable in part to higher fertility and premature mortality.

Figure 3
Age distribution of hospitalizations of residents of high†- and low-Aboriginal dissemination areas, by sex and predominant Aboriginal identity, Canada excluding Quebec, 2001/2002

Differences greater when adjusted for age and sex

Given the young age profile of the Aboriginal population, and that serious morbidity requiring hospitalization is more likely at older ages, it is necessary to standardize hospitalization rates to remove the impact of these differences.

The age-/sex-standardized hospitalization rate for residents of high-Aboriginal DAs was more than twice that of residents of low-Aboriginal DAs (Figure4). This difference could, in part, reflect the higher fertility rate of women in high-Aboriginal DAs. However, even when birth-related hospitalizations were excluded, differences in the standardized hospitalization rates persisted (AppendixA, Figure D). Rates were highest for predominantly First Nations DAs: 2,100 hospitalizations per 10,000 population, compared with 925 for low-Aboriginal DAs. In contrast to the crude rates, standardized hospitalization rates for predominantly Métis (1,676) and Inuit (1,677) DAs were almost the same.

Figure 4
Age-/Sex-standardized† hospitalization rates of residents of high‡- and low-Aboriginal dissemination areas, by predominant Aboriginal identity, Canada excluding Quebec, 2001/2002

Figure D
Hospitalization rates† (with suppressions‡) of residents of  high§- and low-Aboriginal dissemination areas, by predominant Aboriginal identity, Canada excluding Quebec, 2001/2002

Hospitalization rates for residents of high-Aboriginal DAs exceeded those for low-Aboriginal DAs in every province and territory (Figure 5). High-Aboriginal DA rates were highest in New Brunswick (2,914 hospitalizations per 10,000 population), followed by Alberta (2,747). At 827 hospitalizations per 10,000 population, Nunavut had the lowest rate for high-Aboriginal DAs (827).

Figure 5
Age-/Sex-standardized† hospitalization rates of residents of high‡- and low- Aboriginal dissemination areas, by jurisdiction, Canada excluding Quebec, 2001/2002

Rate ratios for high- versus low-Aboriginal DAs ranged from 1.3 in Yukon to 2.7 in Alberta (Table 1). High rate ratios in some provinces—for example, Alberta, and Ontario—may partly be because the predominant identity in many of these provinces’ high-Aboriginal DAs was First Nations. Lower rate ratios may indicate that regional solutions have reduced disparities in serious morbidity between Aboriginal and non-Aboriginal populations, or conversely, a lack of access to care.

Table 1
Age-/Sex- standardized† hospitalization rates, rate differences and relative rate ratios of residents of high‡- and low-Aboriginal dissemination areas, by jurisdiction, Canada excluding Quebec, 2001/2002

General disease groups

Among residents of high-Aboriginal DAs, age-/sex- standardized hospitalization rates for respiratory diseases were more than three times those of residents of low-Aboriginal DAs; rates for injuries, poisonings and consequences of other external causes were 2.6 times higher (Figure 6). Residents of high-Aboriginal DAs were just over twice as likely to be hospitalized for a mental health disorder and 1.8 times as likely to be admitted for a circulatory disease, compared with residents of low-Aboriginal DAs.

Figure 6
Age-/Sex-standardized† hospitalization rates of residents of high‡- and low-Aboriginal dissemination areas, by predominant Aboriginal identity and International Shortlist for Hospital Morbidity Tabulation heading category, Canada excluding Quebec, 2001/2002

Residents of predominantly First Nations DAs had the highest hospitalization rates for all general disease groups shown in Figure 6, except respiratory conditions, for which residents of predominantly Inuit DAs had the highest rates.

Specific diseases and conditions

Hospitalization rates were ranked to identify the ten most common causes. These rankings should be interpreted cautiously, because some are based on small numbers of hospitalizations and small differences in rank order. Those causes ranking high for residents of high-Aboriginal DAs were pneumonia, diabetes mellitus, acute bronchitis and bronchiolitis, cholelithiasis, and early or threatened labour (Table 2).

Table 2
Unadjusted top 10 frequency ranking of ICD-9 diagnosis† as most responsible cause of hospitalization, by predominant Aboriginal identity in high‡-Aboriginal dissemination areas, Canada excluding Quebec, 2001/2002

Pneumonia (organism unspecified) was the most common cause of hospitalization among residents of predominantly First Nations and Métis DAs; acute bronchitis/bronchiolitis was the most frequent among residents of predominantly Inuit DAs. Diabetes mellitus ranked high as a cause of hospitalization for residents of First Nations and Métis DAs, but not Inuit DAs. The high ranking of poisoning caused by medication was unique to Inuit DAs.

Provincial/Territorial differences in the most common causes of hospitalization were evident (Table 3). Pneumonia ranked first among residents of high-Aboriginal DAs in Ontario and the Prairies, and second in British Columbia and the Northwest Territories. By contrast, for residents of low-Aboriginal DAs in these jurisdictions (except the Northwest Territories), pneumonia ranked fourth to ninth. Although Ontario, Manitoba and Alberta did not have any predominantly Inuit DAs, the high ranking of respiratory conditions in these provinces may, in part, be because 1% to 3% of their total hospital discharges were out-of-province patients from Inuit identity DAs in other jurisdictions (Appendix A, Table C). Among residents of high-Aboriginal DAs in the Atlantic provinces, birth-related conditions and respiratory conditions were leading causes of hospitalization. For residents of high-Aboriginal DAs in the Northwest Territories, dental diseases and disorders and respiratory conditions were among the most common causes.

Table 3
Unadjusted top 5 frequency ranking of ICD-9 diagnosis† as most responsible cause of hospitalization in high‡-Aboriginal dissemination areas, by jurisdiction, Canada excluding Quebec, 2001/2002

Table C
Distribution of hospitalizations, by high†- and low-Aboriginal dissemination area, and in high-Aboriginal dissemination areas by predominant Aboriginal identity, Canada excluding Quebec, 2001/2002

Hospitalization rates for conditions known to be highly prevalent among Aboriginal people (ischemic heart disease,^3,4,20,21 mental disorders,^5,22 and asthma^21,23,24) and conditions associated with premature death such as specific types of injury, including suicide or homicide, were examined separately (Table 4). Hospitalization rates for intentional injury inflicted by others was nine times higher among residents of high- than of low-Aboriginal DAs. For intentional self-harm by means other than poisoning and for substance-abuse-related mental disorders, the rates were six times higher, and for self-harm by poisoning (for example, consumption of tranquilizers or barbiturates), four times higher (Table 4).

Table 4
Unadjusted and age-/sex- standardized† hospitalization rates, for selected conditions,‡ by residence in high§- and low-Aboriginal dissemination area, Canada excluding Quebec, 2001/2002

Hospitalization rates for acute bronchitis or bronchiolitis and for pneumonia and influenza were five and four times higher for residents of high- than of low-Aboriginal DAs, respectively. Rates for heart failure/pulmonary edema and asthma among residents of high-Aboriginal DAs were twice those for residents of low-Aboriginal DAs. Differences were narrower for acute myocardial infarction and ischemic heart disease. Reflecting the higher prevalence of HIV among Aboriginal people,²⁵ the rate for HIV was 1.4 times higher among residents of high- than of low-Aboriginal DAs. Differences were marginal for selected cancers, although cancer-related hospitalizations are likely underestimated (see Limitations).

Differences reduced by adjustment for area-level factors

Age-/Sex-standardization adjusts hospitalization rates for demographic differences in Aboriginal and non-Aboriginal populations. However, disparities between these populations’ hospitalization rates likely result from differences in a much broader range of factors such as socio-economic status, employment and housing (Appendix A, Figure E).

Figure E
Percentage of population living in high†- and low-Aboriginal dissemination areas, by predominant Aboriginal identity and selected area characteristics, Canada excluding Quebec, 2001/2002

In fact, when area-level characteristics were taken into account, disparities in hospitalization rates between high- and low-Aboriginal DAs were lessened. Adjusting for rural location and the prevalence of housing in need of major repair in the DA lowered the hospitalization rate among residents of high-Aboriginal DAs (Table 5). Rural residence had the greatest impact, particularly in Inuit-identity DAs, where the adjusted rate ratio dropped to less than half that of low-Aboriginal DAs. By contrast, adjusting for housing conditions in predominantly Inuit DAs had the reverse effect, raising hospitalization rates.

Table 5
Standardized† all-cause acute-care hospitalization rates and rate ratios‡ per 10,000 population for residents of high§-Aboriginal areas, by predominant Aboriginal identity, Canada excluding Quebec, 2001/2002

For most of the selected conditions, adjusting for rural residence or the prevalence of housing in need of major repair had the greatest impact on hospitalization rates (Table 6). For example, hospitalization rates for injury and poisonings among residents of high-Aboriginal DAs fell from 204 per 10,000 population (rate ratio 2.6) to 178 (rate ratio 2.3) when adjusted for rural location, and to 174 (rate ratio 2.2) when adjusted for housing in need of major repair. Similarly, adjusting for these two factors reduced rate ratios for heart failure/pulmonary edema and pneumonia/influenza.

Table 6
Standardized† hospitalization rates per 10,000 population and rate ratios for selected conditions,‡ residents of high§- and low-Aboriginal dissemination areas, by selected characteristics, Canada excluding Quebec, 2001/2002

Adjusting for overcrowded housing lowered the hospitalization rate ratio, particularly for pneumonia/influenza (from 3.9 to 3.1) and for acute bronchitis/bronchiolitis (from 4.8 to 4.2). Adjusting for this characteristic modestly reduced the hospitalization rate ratio for heart failure/pulmonary edema (Table 6).

In general, adjusting for average DA household income quintile had relatively little impact on rate ratios; the greatest reduction was for pneumonia/influenza. And for asthma and acute bronchitis/bronchiolitis, adjusting for household income had the reverse effect, slightly increasing the rate ratio.

Adjusting for DA labour force participation and educational attainment resulted in modest or no change to the rate ratios across all selected disease categories.

Discussion and conclusions

Health disparities between Aboriginal and non-Aboriginal populations in Canada have been extensively documented.^2-7,21 This report provides new information about variations in serious morbidity, as measured by acute-care hospitalization, among residents of high- and low-Aboriginal DAs, and by predominant Aboriginal identity.

Survey data provide estimates of disease prevalence but not severity or progression. Self-reported survey data may also introduce recall bias and be limited by respondents’ ability to accurately describe medical conditions. On the other hand, by definition, acute-care hospitalization indicates relatively severe morbidity. Hospital administrative data may contain more accurate medical information and can give more specifics about the types of morbidity that contribute to the differences in hospitalization rates between residents of high- and low-Aboriginal DAs.

The results of this analysis are consistent with information reported for First Nations or Registered Indian populations using regional hospital discharge data,^3,21 other health service administrative data,²⁰ and international evidence.⁸

The differences in hospitalization rates were greater when standardized by sex and age. This suggests that residents of high-Aboriginal DAs were more likely to experience serious injury and illness at younger ages. The higher hospitalization rate for injury among residents of high-Aboriginal DAs reflects the greater prevalence of serious injury among the off-reserve Aboriginal population.⁶ Higher hospitalization rates for self- and other harm mirror the ranking of such injuries among the leading causes of premature mortality in high-Aboriginal health regions.¹²

Similarly, higher hospitalization rates for circulatory conditions are consistent with the ranking of circulatory diseases as the third leading cause of death in high-Aboriginal health regions.¹² The elevated hospitalization rates may be the outcome of inter-relationships between different diseases over an individual’s life. Pneumonia can be a factor in the development of circulatory disease,²⁷ as can diabetes and hypertension. These circulatory conditions are widely prevalent in the Aboriginal population.^{5,23,24,28-30} As well, poor oral health can place individuals at risk for heart disease-related morbidity later in life.^31,32 And to some extent, higher hospitalization rates for circulatory diseases and respiratory conditions may reflect the higher prevalence of smoking,^5,23,24 higher caloric intake,³⁴ and overweight and obesity,^34,35 in the Aboriginal population.

Relatively low immunization rates among some Aboriginal populations³⁶ may contribute to higher hospitalization rates for respiratory illnesses. However, the analysis in this report suggests that housing conditions are also relevant. Overcrowding and homes in need of major repair²³ may be conducive to the spread of infectious disease and lead to higher rates of respiratory-related hospitalizations.^37-40

Hospitalization rates varied by DA predominant Aboriginal identity. Rates for residents of predominantly Inuit DAs were lower than those for residents of predominantly First Nations DAs. Several factors may play a role in this difference. The diet of Inuit adults still consists largely of meat and harvested fish1²³ — foods that could have protective health benefits. Inuit may be more likely to travel outside their jurisdiction for acute-care hospitalization, which may result in longer hospital stays. By contrast, residents of less remote high-Aboriginal DAs may “cycle” through hospitals more frequently with shorter stays. It is possible that residents of predominantly Inuit DAs do not have the same degree of access to health services as residents of other high-Aboriginal DAs. The lower hospitalization rates for residents of predominantly Inuit DAs may indicate less access to care, which results in death rather than hospitalization. This possibility could be evaluated in the future with a person-level analysis examining length of hospital stay and hospital distance from the patient’s residence. The increase in hospitalization rates for residents of predominantly Inuit DAs that results from adjusting for housing conditions underscores the potential role of these factors in hospitalized morbidity of the Inuit.

Hospitalization rates were also lower among residents of predominantly Métis DAs, compared with predominantly First Nations DAs. This may be, in part, because nearly 70% of the Métis population lives in urban centres,²⁴ and therefore, may have more access to primary health care services than residents of First Nations DAs.

This report sought to disentangle the relative contribution of socio-economic and geographic factors to differences in hospitalization rates between high- and low-Aboriginal DAs. The selected area factors vary in the degree in which they are associated with hospitalization rates, although rural location and overcrowded housing appear to have the strongest relationships, in particular, for First Nations and Inuit DAs and for those with respiratory-related hospitalizations.

It is less obvious how housing conditions could be involved in cardiac morbidity. Overcrowding may increase the risk of infectious diseases, which, in turn, could affect cardiac health.⁴¹ Housing in need of major repair could contribute to exposure to cold, which may play a role in cardiac pathogenesis.⁴²

The significant association between rural location and hospitalization may be partially because rural residents are generally at greater risk of injury,⁴³ for example, travelling greater distances in motor vehicles. Rural residence may also be a proxy for difficulties accessing preventive primary health care.⁴⁴ Lastly, smoking, obesity and a sedentary lifestyle, which are risk factors for cardiac and respiratory morbidity, are more prevalent among rural residents.⁴⁵

The differences in hospitalization rates between high- and low-Aboriginal DAs are not much reduced when adjusted for average area household income, educational attainment and labour force participation. This underlines the limitations of an ecological approach, given the well-established gradients in health status by such determinants at the individual level. Area-level variables can play a role in morbidity, but they are not as direct an influence as person-level factors like smoking.

Hospitalization rates for DAs with a high percentage of Aboriginal residents, in particular, First Nations, were higher nationally and across jurisdictions. This could signal a need for preventive health care, a lack of access to care, or a greater prevalence of serious morbidity. These high rates also indicate candidate areas and health conditions for targeted interventions, that might eventually reduce disparities in premature mortality between Aboriginal and non-Aboriginal people.

The area characteristic standardization analysis identifies modifiable factors, specifically housing conditions, associated with differences in hospitalizationrates. However, adjusting for area household income, educational attainment and labour force participation did not reduce hospitalization rate differences between high- and low-Aboriginal DAs to the same extent as adjusting for rural locale and housing conditions.

While this report attempts to address existing data gaps, it is based on area-level data. Whether individuals from high-Aboriginal DAs are at greater risk of hospitalization than are people from low-Aboriginal DAs cannot be definitively answered without person-level hospitalization counts that contain information on Aboriginal identity.