Unintentional injury hospitalizations and socio-economic status in areas with a high percentage of First Nations identity residents
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.
Evelyne Bougie, Philippe Finès, Lisa N. Oliver and Dafna E. Kohen
Injuries are a leading cause of death among the First Nations population in Canada and have been identified as a serious public health issue in terms of health care costs and diminished quality of life.Note1
Although relatively few studies of injuries among the Aboriginal population have been conducted,Note2 recent population-based research shows high rates of morbidity and mortality due to injuries in the First Nations population. Karmali et al.Note3 reported that the incidence of severe fatal and non-fatal injuries in the Calgary Health Region was higher among individuals with Registered Indian status, compared with other residents. Using survey data, TjepkemaNote4 reported higher non-fatal injury rates among the provincial off-reserve Aboriginal population, compared with the non-Aboriginal population. Linkages of census and mortality data show higher rates of mortality due to injuries over the 1991-to-2001 period for Status and non-Status Indians, compared with the non-Aboriginal population.Note5,Note6
National studies of hospitalizations due to injuries among the First Nations population have been even scarcer because of the lack or inconsistent reporting of Aboriginal identity information on hospital administrative records. Hospital data are important because discharge records contain information about injuries severe enough to warrant hospital admission. Currently, three provinces (British Columbia, Saskatchewan and Manitoba) report population-based rates of hospitalization due to injuries. Although each provincial database uses different methods to identify First Nations patients,Note7 these data show higher injury hospitalization rates among First Nations people living on- and off-reserve in Western Canada, compared with the general population.
To overcome the general lack of Aboriginal identity data on hospital records, some researchers have adopted an area-based approachNote8 and examined injury hospitalizations in geographic areas with a relatively high percentage of Aboriginal identity residents. For instance, Fantus et al.Note9 found higher all-cause injury hospitalization rates in northern Ontario’s First Nations communities, compared with similarly located non-Aboriginal communities. Alaghehbandan et al.Note10 reported higher all-cause unintentional injury hospitalization rates among children and adolescents in high-percentage Aboriginal identity communities in Newfoundland and Labrador, compared with children and adolescents in lower-percentage Aboriginal identity communities. At the national level (excluding Quebec), Oliver and Kohen,Note11 Finès et al.,Note12 and Carrière et al.Note13 found higher rates of unintentional injury hospitalizations among children, youth and adults in high-percentage First Nations identity areas relative to low-percentage Aboriginal identity areas.
An important limitation of many of these area-based studies is that they did not investigate the role of area characteristics, such as socio-economic conditions or remote location, which may help explain the elevated injury rates in high-percentage Aboriginal identity areas. An exception is Carrière et al.Note13 who demonstrated that adjusting for housing in need of major repair and for rural location reduced (but did not eliminate) the difference in all-cause injury hospitalization rates between areas with relatively high and low percentages of Aboriginal identity residents.
Population studies have shown patterns in injury hospitalizations and mortality by neighbourhood income.Note14-18 Evidence also suggests that individuals living in rural areas may be at an increased risk for injury-related co-morbidities because of reduced access to health care and emergency facilities and because of greater distances travelled in motor vehicles for commodities and services.Note1,Note18-20
Compared with the Canadian population overall, the First Nations population tends to have lower educational attainment, lower employment rates, and lower-quality housing,Note21-23 all of which may be associated with increased injury risk.Note1,Note24 Thus, the goal of this analysis is to investigate whether the higher unintentional injury hospitalization rates observed in areas with a high percentage of First Nations identity residents are associated with area socio-economic conditions and remote location.
Using a geographic approach and six years of national hospitalization data, this study: 1) describes the extent to which unintentional injury hospitalization rates varied based on area socio-economic conditions and location relative to an urban core; 2) describes variations in these associations by age group and sex; and 3) compares unintentional injury hospitalization rates in areas with a high percentage of First Nations identity residents with rates in areas with a low percentage of Aboriginal identity residents with similar socio-economic conditions and location relative to an urban core.
This is an ecological study reporting results for geographic areas. Because Aboriginal identifiers were not available on hospital records, hospitalizations of Aboriginal people could not be ascertained. The associations should, therefore, be interpreted as applying to geographic areas only.
Data and methods
The data are from the Discharge Abstract Database (DAD) for 2004/2005 to 2009/2010 and the 2006 Census. The DAD contains discharge records for all hospital separations in Canada except Quebec.Note25 For each separation, information is available on the patient’s age, sex, residential postal code, diagnosis codes, and dates of admission and discharge.
The International Classification of Diseases, 10th Revision (ICD-10-CA)Note26 diagnosis codes were used to classify injuries based on the external cause. The injury episodes extracted for analysis pertained only to unintentional injuries—those for which there was no intent to harm, on the part of the victim or anyone else. Adverse effects due to drugs or medical care were excluded. Data quality reports indicate high accuracy of ICD-10-CA on separation records.Note25
Patients transferred between facilities have multiple separation records for the same injury episode. To avoid double-counting, any patient who was discharged and re-admitted to another hospital on the same day was counted as a single episode. For injury hospitalizations with multiple separation records, the ICD codes on the first record were used.
The data represent the number of injury episodes rather than the number of individuals.
Because the DAD does not contain information on patients’ Aboriginal identity, a geographic method—geozonesNote8,Note27—was used to identify dissemination areas (DAs) with a relatively high percentage of residents who identified as Aboriginal in the 2006 Census. DAs are the smallest geographic unit for which census information is available nationally. They are composed of one or more neighbouring dissemination blocks, with a population of 400 to 700.
This study uses DAs as a proxy for neighbourhoods. Following previous work,Note11-13 DAs with less than 33% of residents reporting an Aboriginal identity were classified as areas with a low percentage of Aboriginal identity residents. DAs where at least 33% of residents reported an Aboriginal identity were classified as areas with a high percentage of Aboriginal identity residents, and then further classified as First Nations, Métis, or Inuit areas based on the predominant Aboriginal group. Only DAs identified as high-percentage First Nations identity or low-percentage Aboriginal identity were retained for this analysis. Residential postal codes on the hospital separation record were used to determine the patient’s DA of residence via the PCCF+.Note28 The availability and accuracy of postal codes on separation records was high, with more than 99% of records successfully assigned to a DA. Because hospital separation records for Quebec contain only the first three digits of the postal code, they were excluded from this study.
In most DAs (38,869) in Canada (excluding Quebec), fewer than 33% of residents reported an Aboriginal identity in the 2006 Census. Of the DAs where at least 33% of residents reported an Aboriginal identity, 1,929 were predominantly First Nations. Owing to small populations, global non-response, or incompletely enumerated Indian Reserves, a number of DAs had insufficient census information and were excluded from this analysis. This left 1,288 DAs in high-percentage First Nations areas (total population of 419,699) and 38,700 DAs in low-percentage Aboriginal areas (total population of 23,217,988) in the analysis (Figure 1).
In 2006, an average of 80% of residents of high-percentage First Nations areas reported an Aboriginal identity (74% single North American Indian identity, 4% single Métis identity, 0.2% single Inuit identity, and 1% multiple or other Aboriginal identities). An average of 3% of residents of low-percentage Aboriginal areas reported an Aboriginal identity.
Socio-economic status and remote location
A composite measure of socio-economic status (SES) was chosen to classify DAs as relatively “high” or “low” SES. The Community Well-Being (CWB) Index, devised by Aboriginal Affairs and Northern Development Canada, consists of seven census-based indicators: income per capita; percentage of population aged 20 or older with at least a secondary school diploma; percentage of population aged 25 or older with at least a university degree; labour force participation and employment among those aged 20 to 65; percentage of population living in crowded conditions; and percentage of population living in dwellings in need of major repair. The CWB Index combines these indicators into a single score ranging from 0 (low socio-economic level) to 100 (high socio-economic level). Information on the CWB Index, its validation, and its relevance to First Nations populations is available elsewhere.Note29,Note30
Microdata from the 2006 Census were used to calculate a CWB Index score for each DA. A CWB Index score was also calculated for each province/territory. To compare injury hospitalization rates of DAs with similar socio-economic conditions, DAs were then classified as relatively “high” or “low” SES. Sample size issues precluded the creation of SES terciles or quintiles. To account for variations in socio-economic conditions across Canada, SES levels were constructed within each province/territory. A DA was classified as low (or high) SES if its CWB Index score was below (equal to or above) its provincial/territorial CWB Index median cut-point value. The majority of DAs in high-percentage First Nations identity areas (1,250 or 97%) were classified as low SES; 38 (3%) were classified as high SES. In low-percentage Aboriginal identity areas, 18,650 DAs (48%) were classified as low SES, and 20,050 (52%) were classified as high SES (Appendix Table A).
The remoteness of a DA relative to an urban core was measured using the Metropolitan Influence Zone indicator. This indicator assigns a category to municipalities outside a census metropolitan area (CMA) or a census agglomeration (CA) based on the percentage of the employed population who commute to work in a CMA/CA. A CMA/CA is an area consisting of one or more neighbouring municipalities situated around a major urban core. The majority of DAs in high-percentage First Nations identity areas (831 or 65%) were located outside a CMA/CA with weak or no metropolitan influence; 114 (9%) were outside a CMA/CA with strong or moderate metropolitan influence; and 343 (27%) were inside a CMA/CA. Few DAs in low-percentage Aboriginal identity areas (3,534 or 9%) were located outside a CMA/CA with weak or no metropolitan influence; 4,565 (12%) were outside a CMA/CA with strong or moderate influence; and 30,601 (79%) were inside a CMA/CA (Appendix Table B).
Age-standardized injury hospitalization rates (ASHRs) were calculated over the six-year period (2004/2005 to 2009/2010), and were standardized to the age structure of the 2006 Aboriginal identity population in five-year age groups. The denominator used to calculate the rates was from the 2006 Census, which corresponds to the midpoint of the years of hospitalization data, and was multiplied by six to account for the six years of data. For the small number of DAs that lacked the detailed age and sex data required for age standardization, age and sex were estimated from total population counts or population estimates of incompletely enumerated Indian Reserves.
ASHRs per 10,000 person-years at risk were calculated for three age groups (0 to 19, 20 to 44, and 45 or older), by sex and by selected DA characteristics (SES level and location relative to an urban core) in high-percentage First Nations and low-percentage Aboriginal identity areas.
Four sets of rate ratios (RRs) were calculated. The first set compared rates in low-SES DAs to rates in high-SES DAs separately for high-percentage First Nations and low-percentage Aboriginal DAs. The second set compared rates in high-percentage First Nations DAs with rates in low-percentage Aboriginal DAs with similar SES. The third set compared rates in urban-core DAs with rates in remote DAs separately for high-percentage First Nations and low-percentage Aboriginal DAs. The fourth set compared rates in high-percentage First Nations DAs with rates in low-percentage Aboriginal DAs with similar location relative to an urban core. All 95% confidence intervals were calculated based on the assumption of log normality.Note31 Two rates were considered to differ significantly if their 95% confidence intervals did not overlap.
From 2004/2005 through 2009/2010, there were a total of 27,887 hospital separations for unintentional injury among residents of high-percentage First Nations identity areas, and 814,313 among residents of low-percentage Aboriginal identity areas (excluding Quebec). In high-percentage First Nations identity areas, 57% of hospitalizations were for males; 42% were for people aged 45 or older; 34% were for 20- to 44-year-olds; and 24% were for 0- to 19-year-olds. In low-percentage Aboriginal identity areas, 50% of hospitalizations were for males; 66% were for people aged 45 or older; 20% were for 20- to 44-year-olds; and 14% were for 0- to 19-year-olds (Table 1).
Almost without exception, ASHRs for unintentional injury were significantly higher in low-SES DAs, compared with high-SES DAs (Table 2, Figure 2). This was true both for high-percentage First Nations and low-percentage Aboriginal identity DAs. One exception occurred in high-percentage First Nations DAs, where ASHRs for males and females aged 0 to 19 did not significantly differ between high- and low-SES DAs.
Age-standardized unintentional injury hospitalization rates (ASHR) per 10,000 person-years, high-percentage First Nations identity and low-percentage Aboriginal identity Dissemination Areas (DAs), by age group, sex, and DA socio-economic status and location, Canada (excluding Quebec), 2004/2005 to 2009/2010
When DAs with the same SES were compared, ASHRs in low-SES DAs were 1.9 to 3.5 times greater in high-percentage First Nations identity DAs than in low-percentage Aboriginal identity DAs (Table 2, Figure 3). Trends were similar in high-SES DAs—ASHRs were 2.1 and 3.2 times greater in high-percentage First Nations DAs than in low-percentage Aboriginal DAs. The largest disparity between high-percentage First Nations and low-percentage Aboriginal DAs was among females aged 20 to 44 in low-SES DAs (RR = 3.5).
Location relative to an urban core
ASHRs for unintentional injuries were significantly higher in remote DAs (outside an urban core with weak or no metropolitan influence) than in urban-core DAs (Table 2 and Figure 4). This was true for high-percentage First Nations and low-percentage Aboriginal identity DAs. This pattern prevailed among all age and sex groups.
Comparisons of similarly located DAs show that ASHRs in remote DAs were 1.3 to 2.6 times greater in high-percentage First Nations identity areas than in low-percentage Aboriginal identity areas (Table 2 and Figure 5). As well, ASHRs in urban-core DAs were 1.8 to 3.4 times greater in high-percentage First Nations areas than in low-percentage Aboriginal areas. The widest disparity between high-percentage First Nations and low-percentage Aboriginal identity areas was among women aged 20 to 44 living in urban-core DAs (RR=3.4).
This ecological study uses a geographic approach and national hospital data to describe variations in unintentional injury hospitalization rates based on area SES and location relative to an urban core. Unintentional injury hospitalization rates were higher in DAs with low (rather than high) SES, and in DAs in remote areas rather than in urban cores. This was true in both high-percentage First Nations and low-percentage Aboriginal identity areas. These results are consistent with the literature, which shows a socio-economic gradient for many health outcomes including injuries,Note14-18 and a difference in exposure to injury risks between urban and rural environments.Note1,Note18,19,20
However, when DAs with similar SES and location relative to an urban core were compared, the relative risk for unintentional injury hospitalization was greater in those with a high percentage of First Nations identity residents than in DAs with a low percentage of Aboriginal identity residents. This is in line with Carrière et al.,Note13 who showed that adjusting for housing conditions and rural location did not entirely eliminate the difference between high- and low-percentage Aboriginal identity DAs for all-cause injury hospitalizations. In the present study, women aged 20 to 44 in high-percentage First Nations identity areas appeared to be at particular risk of unintentional injury hospitalizations in low-SES DAs, and in DAs located inside an urban core, compared with their counterparts in comparable low-percentage Aboriginal identity areas.
Findings from this study show that socio-economic conditions and remote location accounted for some, but not all, of the differences in unintentional injury hospitalizations between high-percentage First Nations identity and low-percentage Aboriginal identity DAs. Factors not measured in this analysis may play a role in DA-level unintentional injury hospitalization risk. Such factors may include environmental and behavioural risk factors for injuriesNote1, as well as occupational, workplace, and recreational space safety.Note15
Strengths and limitations
This study fills an important health information gap by investigating associations between DA SES, location relative to an urban core, and hospitalizations due to unintentional injuries in high-percentage First Nations and low-percentage Aboriginal identity areas. The use of six years of hospitalization data and the calculation of hospitalization rates by sex and age group are also notable strengths.
Nonetheless, the results should be interpreted in the context of several limitations. Because individual Aboriginal identifiers were not available on hospital records, hospitalizations of Aboriginal people cannot be ascertained. Rather, this is an ecological study reporting results for geographic areas; the associations observed do not necessarily apply at the individual level. Furthermore, the populations of these areas are comprised of Aboriginal and non-Aboriginal identity residents, so the results are not specific to First Nations people but to all residents of those areas. Also, the results cannot be generalized to the First Nations population in Canada, because a portion of this population resides in low-percentage Aboriginal identity areas.
Limitations also stem from the hospitalization data. The data exclude injuries that resulted in death before hospital admission; people who suffered injuries but did not seek medical care; and individuals presenting to emergency rooms, physicians’ offices or clinics. The geographic location where the injury occurred was not available, so the residential postal code was used as a proxy. Assignment of DAs via postal codes is less accurate in rural areas, where residents often use a P.O box instead of a home address, and a single postal code can span several DAs.Note28 Hospital separation records for the province of Quebec were not used in this analysis because they contain only the first three digits of the six-digit postal code.
Other limitations are associated with the DA data. A total of 641 high-percentage First Nations identity DAs and 169 low-percentage Aboriginal identity DAs were excluded from the analysis because of insufficient census information. Sample size and data quality precluded classifying DAs into three or more SES levels; to preserve confidentiality, DAs were dichotomized into high- versus low-SES. Finally, the study included only broad SES indicators at the DA level. Individual-level SES indicators, individual behaviours, and additional area characteristics might have provided a more complete picture, but such measures were not available.
This ecological study offers insights into associations between unintentional injury hospitalizations and DA SES and location relative to an urban core in high-percentage First Nations and low-percentage Aboriginal identity DAs. Research is needed to identify other factors contributing to the higher risk of unintentional injury hospitalization in areas with a high percentage of First Nations identity residents.
The First Nations and Inuit Health Branch (Health Canada) provided financial support for analysis of the data and preparation of the manuscript.