Health Reports
Hospitalization for ambulatory care sensitive conditions among urban Métis adults

by Gisèle M. Carrière, Mohan B. Kumar and Claudia Sanmartin

Release date: December 20, 2017

Ambulatory care sensitive conditions (ACSCs) are potentially preventable, but if untreated, can result in high use of health care services.Note 1Note 2 In fact, hospitalization rates for ACSCs are used as an indirect measure of the adequacy and accessibility of primary health care.Note 1Note 3Note 4Note 5Note 6 The people most at risk for ACSC hospitalizations tend to be older; have poorer health, lower socioeconomic status, and comorbidities; be regular smokersNote 3Note 7Note 8; and live in rural areas.Note 9

An elevated risk of avoidable hospitalizations among the Aboriginal population has been reported in Australia,Note 8 but analyses by Aboriginal identity have not been conducted for Canada at the national level. This is a concern because the prevalence of potentially treatable conditions such as diabetes, asthma, and cardiovascular disease has been found to be higher among Aboriginal people.Note 10Note 11Note 12Note 13 Moreover, evidence indicates a trend toward a rise in the prevalence of some ACSCs among Aboriginal populations in Canada and elsewhere.Note 8Note 14

Several factors suggest that Métis may be more likely than non-Aboriginal people to be hospitalized for ACSCs. In 2006, 14% of Métis reported having asthma, and 7% reported diabetes; the corresponding figures for the total Canadian population were 8% and 4%.Note 10 In 2012, daily smoking, an established risk factor for ACSC hospitalization,Note 3 was reported by 26% of Métis aged 12 or older,Note 15 compared with 15% of the total population.

The likelihood of receiving primary care and of ACSC hospitalizations may be related to where people live.Note 9Note 16Note 17Note 18Note 19 While Aboriginal people overall are more likely than non-Aboriginal people to live in rural and remote areas,Note 20 Métis tend to reside in urban areasNote 20; thus, the availability of primary health care may be similar for Métis and non-Aboriginal people.

An Ontario analysis showed that although Métis disproportionately suffer from diabetes,Note 13 no differences were detected for visits to primary care physicians and specialists, or for retinopathy screening, compared with the total Ontario population.Note 13Note 16 However, among Métis with diabetes, hospitalizations were more likely than among Ontarians overall.Note 21 Similar patterns were reported in Manitoba.Note 22 Neither study examined ACSC hospitalizations specifically. Moreover, national information about the health service use of Métis is generally lacking.Note 23

The objective of this analysis is to determine if Métis are more likely than non-Aboriginal people to be hospitalized for ACSCs and whether differences persist after adjustment for socioeconomic and geographic factors. As well, comorbidity among Métis hospitalized for an ACSC is compared with that of their hospitalized non-Aboriginal counterparts.

The study is based on a linkage of the 2006 Census of Population with the Discharge Abstract Database.Note 24 This enables identification of ACSC hospitalizations by Aboriginal identity.

Data and methods

Data sources

The 2006 Census (long-form) was linked to the Discharge Abstract Database (DAD) for all Canadian jurisdictions, excluding Quebec.Note 24 The long-form questionnaire collected detailed information, including Aboriginal identity. The long form was completed by about 20% of the non-institutional population, and was administered to all residents of Indian reserves and settlements, and many remote and northern communities.Note 25

The DAD is a census of acute care hospital discharges in all provinces and territories (excluding Quebec), provided annually to Statistics Canada by the Canadian Institute for Health Information.Note 26 The DAD contains demographic, administrative, and clinical data for about 3 million hospital discharges each year. Acute care hospital discharge records for fiscal years 2006/2007 through 2008/2009 were used for the linkage on which this analysis was based.

Among long-form census respondents who identified as Métis, 78.4% were eligible for linkage to the DAD; between 5.7% and 6.4% (depending on the year) linked to at least one acute care hospitalization. Among non-Aboriginal respondents, 94% were eligible for linkage, and between 5.0% and 5.4% linked to at least one hospitalization.Note 24 Methodological details, including criteria applied to each data source to determine linkage eligibility, are available elsewhere.Note 24

The linkage was approved by Statistics Canada’s Executive Management BoardNote 27 and is governed by the Directive on Record Linkage.Note 28 Statistics Canada ensures respondent privacy during linkage and subsequent use of the linked files. Only employees directly involved in the linkage process have access to the unique identifying information (such as name and sex) and do not access health-related information. When a linkage is complete, an analytical file is created from which the identifying information has been removed. This de-identified file is provided to researchers for analysis.

Study sample

The study cohort consisted of 2006 Census respondents aged 18 to 74 in Census Metropolitan Areas (CMAs) or in zones strongly or moderately influenced by CMAs who reported Métis as a single Aboriginal identity and those who did not identify as Aboriginal (“non-Aboriginal”). Aboriginal identity was derived from the question: “Is this person an Aboriginal person, that is, North American Indian, Métis or Inuit (Eskimo)?” The Aboriginal identity population consists of people who identified with at least one of the following groups: North American Indian, Métis or Inuit, and/or a Treaty Indian and/or or a Registered Indian as defined by the Indian Act of Canada, and/or members of an Indian band or First Nation. Non-Aboriginal people are those who did not report an Aboriginal identity.

Métis identity in this study is based on the date of the 2006 Census (May 16). “Ethnicity mobility” as it pertains to MétisNote 29 should be taken into account if the findings of this analysis are compared with those for other periods.

Métis tend to be urban-dwellersNote 20―in 2006, 69% lived in large cities (CMAs) or smaller urban areas. However, among urban-dwellers, Métis were twice as likely as non-Aboriginal people to live in small urban centres (41% versus 20%).Note 20 For the present study, the 2006 Census Standard Area Classification (SAC) TypeNote 30 was used to classify census subdivisions (CSDs) according to whether they were a component of one of the following: CMA; census agglomeration (CA); CMA-influenced-zone; or CA-influenced zone.Note 30Note 31 In areas outside CMAs and CAs, the Statistical Area Type is defined by characteristics of the CSD based on commuting flows to work in CMAs or CAs, which determine if a CSD is a “metropolitan influenced zone.”Note 30

To control for potential variations due to unmeasured effects of access to primary care on the likelihood of ACSC hospitalization, this analysis used cohort members who resided in the levels of SAC Type spanning the range from CMA to non-CMA/non-CA areas with either strong or moderate influence from nearby metropolitan areas. Respondents in areas with weak or unknown metropolitan influence, or missing Statistical Area Type information, were not included. The cohort comprised 2.86 million census respondents, 36,700 of whom were Métis (Table 1).

In 2011, Statistics Canada changed the standards for urban versus rural areas; researchers using more recent vintages of data should consult this revised standard.Note 31


ACSC hospitalizations were those with a “most responsible diagnosis” of diabetes, chronic obstructive pulmonary disease (COPD), asthma, angina, grand mal status and other epileptic convulsions, heart failure and pulmonary edema, or hypertension,Note 32 coded according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Canada (ICD-10-CA) (Appendix Text Table A).Note 33 Dichotomous variables were created to indicate if an individual experienced at least one hospitalization for each condition, as well as for any type of ACSC.

Among those hospitalized for ACSCs, Charlson Index comobiditiesNote 34Note 35 were determined, based on up to 24 diagnoses other than the “most responsible diagnosis” listed on each hospital record (Appendix Text Table B contains ICD-10-CA coding definitions).


Urban/Rural residence,Note 36Note 37 income, education, and employmentNote 38Note 39 are strongly associated with health. These covariates, derived from linked census information, were used to adjust the regression models.

Residence was defined as within CMAs, CAs, or strongly or moderately CMA-influenced areas.

Household income quintile was derived at the economic family level or directly for unattached individuals. Total after-tax income from all sources, from all members of each household was summed, adjusted for household size, and divided into quintiles. To minimize regional income differences, income quintile thresholds were estimated for each province/territory using the distribution of after-tax income in that province/territory. Individuals were assigned to the quintile in which their household income fell.

Educational attainment was the highest level of formal education. Two levels were defined: 1) secondary graduation or more (registered apprenticeship certificate, other trades certificate or diploma, college, CEGEP or other non-university certificate or diploma, university), or 2) less than secondary graduation.

Labour force status pertained to the week before the census date (reference week). Three levels were defined: employed (worked in reference week); unemployed (looking for work, available for work, but did not work); and not in labour force (not working, not looking for work, not available for work).

Because research has shown associations between living arrangements and health outcomes,Note 40Note 41 models were adjusted for household living arrangements.

Statistical methods

Age-standardized hospitalization rates (ASHRs) per 100,000 population and 95% confidence intervals were calculated by Métis and non-Aboriginal identity for ACSCs overall (at least one) and separately for each ACSC. Hospital records for 2006/2007 through 2008/2009 were pooled. ASHRs were computed by dividing hospitalizations of Métis and non-Aboriginal people (numerator) by the unweighted person estimates from the study cohort for the same identity group (denominator) multiplied by three. Sampling weights account for the survey design and the under- or over-representation of people with certain characteristics. Because census weights were not adjusted for linkage eligibility, the long-form census weights are not representative of the linked cohort. Therefore, this analysis is based on an unweighted linked study cohort.

The direct method was used to standardize to the age structure of the 2006 Census Aboriginal population. The standard errors to create 95% confidence intervals for ASHRs were derived using methods described by Spiegelman.Note 42

Rate ratios (RRs) with 95% confidence intervals were calculated for Métis relative to non-Aboriginal adults.

To determine if factors such as socioeconomic status relate to differences between Métis and non-Aboriginal people in the likelihood of an ACSC hospitalization, multivariate logistic regression models estimated the odds, using the total urban adult non-Aboriginal component of the cohort as the reference group. Five sequential models were estimated, with adjustments for covariates added in the following order: 1) age, sex, and province or territory of residence; 2) SAC Type; 3) per person total household income quintile; 4) educational attainment; and 5) labour force status and living arrangements.

Analyses were completed using SAS version 9.3.


Study cohort

Urban Métis study cohort members were slightly younger than their non-Aboriginal counterparts, with median ages of 40 and 44, respectively One-quarter (24%) of Métis cohort members lived in Ontario, compared with about half (54%) of those who were non-Aboriginal. Just over half (56%) of Métis were in CMAs, and 14% were in zones having moderate metropolitan influence; among non-Aboriginal people, the corresponding figures were 75% and 7%. Métis were more likely than non-Aboriginal people to be in the two lowest income quintiles and to have less than secondary graduation. Similar percentages of each group were employed and lived alone.

Hospitalization for ACSC

From 2006/2007 through 2008/2009, 8.7% (n = 30,345) of hospitalizations of study cohort members were related to ACSCs. Although Métis made up 1% of the total cohort, they accounted for 2% of ACSC hospitalizations.

The ACSC-related ASHR among Métis was more than twice that of non-Aboriginal people: 393 compared with 184 per 100,000 population (RR 2.14; CI: 1.96 to 2.33) (Table 2). The most pronounced difference was diabetes-related hospitalizations, with rates almost three times higher for Métis: 110 versus 40 per 100,000 (RR 2.75; CI: 2.31 to 3.28). ASHRs among Métis were more than twice as high for COPD-related conditions (RR 2.36; CI: 2.07 to 2.70) and for asthma (RR 2.35; CI: 1.48 to 3.72). Hospitalization rates for all other types of ACSCs were also higher among Métis.

Characteristics of individuals hospitalized for ACSC

Regardless of whether they were Métis or non-Aboriginal, individuals who experienced at least one ACSC hospitalization tended to be older and male; had lower incomes and less education; were not in the labour force; and lived alone (Table 3). Métis and non-Aboriginal people in zones with only moderate metropolitan influence were more likely to have an ACSC hospitalization than were their counterparts in CMAs.

The prevalence of Charlson Index comorbidities among ACSC patients was higher for Métis than for non-Aboriginal patients (Table 4), but differences were not significant, possibly due to lack of power resulting from small sample size.

Logistic regression models

Logistic regression models tested whether differences in geographic, demographic, and socioeconomic factors accounted for the higher rates of ACSC hospitalization among Métis. For three outcomes―at least one ACSC hospitalization of any type, a hospitalization related to diabetes, or a hospitalization related to COPD―Métis had higher odds (Table 5). Age-, sex-, and province/territory-adjusted odds for at least one ACSC hospitalization were almost twice as high among Métis as among non-Aboriginal people; for diabetes-related or COPD-related hospitalizations, odds were more than twice as high. Further adjustment for SAC Typereduced the odds for Métis, but they remained significantly higher. Additional adjustment for household income quintile greatly reduced, but did not eliminate, differences between the two groups, as did adjustment for educational attainment. Inclusion of employment status and living arrangements yielded slight reductions. Yet even when all adjustments were applied, significant differences between the two groups remained―for Métis, the adjusted odds of a diabetes-related hospitalization were twice those for non-Aboriginal people, and the odds of a COPD-related hospitalization or any kind of ACSC hospitalization were about 1.5 times higher.


At ages 18 to 74, Métis were significantly more likely than non-Aboriginal people to be hospitalized for an ACSC. Demographic, geographic, and socioeconomic characteristics account for some, but not all, of these differences.

Restricting the cohort to residents of metropolitan or metropolitan influenced zones was an attempt to control for unmeasured effects of primary care (supply of physicians in urban areas). Residence within, rather than outside, a CMA was related to lower odds of ACSC hospitalization. A rising gradient in the distribution of ACSC hospitalization was evident across SAC Type from CMA to moderate metropolitan influence. However, area of residence did not fully account for differences in ACSC hospitalizations between the two groups. Métis may face barriers to primary health care similar to those reported by Aboriginal people generally.Note 43Note 44Note 45Note 46

Adjustment for socioeconomic characteristics narrowed gaps between the two groups, but significant differences persisted. These differences may be due to risk factors such as daily smoking and poorer self-reported health, which are more prevalent among Métis,Note 10Note 15Note 47 but which were not available in the linked census–DAD data.

Results of the present analysis suggest a higher prevalence of serious comorbidities among Métis hospitalized with at least one ACSC. Such differences may partly explain the persistent association between ACSC hospitalization and Métis identity after all adjustments. Comorbidity among Métis ACSC patients suggests the existence of more serious illnesses, or that conditions compounded to complicate and further increase the likelihood of ACSC hospitalizations. This is consistent with research citing comorbidity as a possible explanation for the higher likelihood of hospitalization among Métis with diabetes relative to the overall population.Note 21


Métis living in CMAs or in CMA influenced areas are more likely than non-Aboriginal adults in the same types of areas to experience ACSC hospitalizations. A higher prevalence of comorbid conditions among Métis may account for some of the difference, but this requires additional investigation. Research using other data sources is needed to assess the role of comorbid chronic conditions, primary health care use, and health behaviours in the association between Métis identity and the likelihood of ACSC hospitalization.


Date modified: