Development of the Canadian Marginalization Index: a new tool for the study of inequality/Elaboration de l'indice de marginalisation Canadien: un nouvel outil d'etude des inegalites.

Matheson, Flora I. ; Dunn, James R. ; Smith, Katherine L.W. 等

The use of area-based indicators of socio-economic status has a long history, and as more databases become available and interest in population health research and monitoring grows, so does the use of so-called "area-based deprivation indices" or ABDIs. (1) As a result of this growing interest, the Canadian Marginalization Index (CAN-Marg) was developed from existing area-based research and theory linking neighbourhood marginalization with poor health. Developed by a Toronto-based research team in 2006, the CAN-Marg is a census-based, geographically derived index for use in research that seeks to understand inequalities in health and other social problems related to health among either population groups or geographic areas.

CAN-Marg goes beyond the traditional definitions of exclusion primarily captured by past and current ABDIs. (2-9) Most existing ABDIs have a strong focus on aspects of material deprivation, such as income, car ownership and home ownership. This focus is based on a model that emphasizes economic inequality as paramount. However, as societies in the affluent countries of the world have changed over the past 30-40 years, there are other facets of inequality that may be just as important for health. CAN-Marg, therefore, is a multifaceted index, allowing researchers and policy and program analysts to examine multiple dimensions of marginalization in urban and rural Canada and the effects of those dimensions of inequality on health and other social outcomes. CAN-Marg has been developed with reference to marginalization, which is a process that creates inequalities along multiple axes of social dif ferentiation in Canada. (10) The four dimensions included in the index are residential instability, material deprivation, ethnic concentration and dependency.

CAN-Marg was developed by building initially on a theoretical framework based on previous work in the field of deprivation and marginalization, but in its development we also allowed for the possibility that other forms of marginalization, for example, marginalization by ethno-racial identity, immigration status, life-cycle stage and household composition, could be included empirically. (2-9) With this expanded theoretical conception of the different forms of marginalization in contemporary society, a broader set of potential census indicators were assembled and the index was then empirically derived using principal components factor analysis. (11,12) Our desire was to create an index that captured the nature of between-place variation in a broad set of marginalization indicators and that was independent of any association with health outcomes, so as to avoid any circular or tautological reasoning (i.e., the argument that the index is a good measure for use in health research because it is associated with health). That said, later in this paper we do show its performance in predicting population health outcomes.

The purpose of this paper is to describe CAN-Marg, show its stability across time periods and across geographic areas (e.g., cities and rural areas), and illustrate its association with health and behavioural outcomes. The latter is an important aspect of the project since CAN-Marg itself does not contain health information, yet many researchers have found that marginalization, as measured by area-based composite indices, correlates strongly with measures of health status. (8,9,13-18)

Creating the Canadian Marginalization Index

The selection of neighbourhood characteristics for this analysis was guided by previous research on ABDIs (6,9,17,19) and on contemporary theoretical perspectives on inequality and marginalization in Canadian society. (10,20) The literature on deprivation and residential instability was pivotal as a starting point and provided the input variables for factor analysis. (6,9,10,17,19,20) The initial analysis was conducted using 2001 census tract (CT) data (urban areas). In total, 42 measures from the census data were created for input into factor analysis. Measures with low factor loadings were removed on an iterative basis, after which four factors emerged with 18 CT measures remaining. Factors were constructed using oblique rotation, which allows the factors to co-vary. (11,12,21) We also estimated an orthogonal factor matrix that yielded substantively identical results. We repeated factor analysis using the same 18 census measures for dissemination areas (DAs) (2001 and 2006) and 2006 CTs.

Table 1 shows the indicators associated with each dimension of marginalization. The correlations between the indicators and their respective marginalization dimensions were fairly stable across time and geographic area, the majority being above 0.65. The factor loadings were also fairly stable (most >0.50). Those for proportion below the low income cut-off (see material deprivation) were lower in 2006 (<0.50). Historically, income has been a primary component of deprivation indices, so there is a theoretical rationale for its placement in the deprivation dimension. (3,4,19) In general, however, the factor loadings and eigenvalues were similar within dimensions by year and geographic area, and the percentage variance explained by the four dimensions ranged from 70% to 80% (Table 2). Residential instability was the dominant dimension with the highest eigenvalues of the four dimensions, followed by material deprivation, dependency and ethnic concentration.

Factor loadings were used to compute factor scores, which allowed us to create a separate continuous index for each of the four dimensions. Each dimension is an asymmetrically standardized scale. For the purposes of this paper, the factor scores are used as quintiles. CAN-Marg quintiles were generated by ordering DAs according to increasing marginalization and allocating an equal number of DAs to each quintile. These were created by sorting the marginalization data into five groups, ranked from 1 (least marginalized) to 5 (most marginalized). Each group contains a fifth of the geographic units. For example, if an area has a value of 5 on the material deprivation scale, it means it is in the most deprived 20% of areas in Canada. (22)

Associations between marginalization and health/behavioural outcomes

We next examined the relation between the four dimensions of the 2006 CAN-Marg Index and 18 health and behavioural outcomes using multilevel modeling with DAs as the area-level unit of analysis. Individual-level data were derived from 2 cycles (cycle 3.1 and cycle 2007/08) of the Canadian Community Health Survey (CCHS), a cross-sectional nationally representative survey that provides detailed information on health determinants and outcomes for individuals. (23) The CCHS 2007/2008 respondents had 2006 DAs in the data set, whereas cycle 3.1 (2005) had 2001 DAs. In this cycle, 2001 geographic areas were converted to 2006 geographic areas using the postal code-DA assignment in cycle 2007/08 and the Postal Code Conversion File Plus (version 5H), which provides an algorithm to assign census geographic areas on the basis of postal codes. (24)

Statistical approach

Descriptive analyses were weighted using the combined sampling weight provided by Statistics Canada. Prevalence rates were weighted and shown with weighted sample frequencies. Our approach to evaluating the association between marginalization and CCHS health and behavioural outcomes was to use multilevel modeling, which takes into account the potential for clustered observations within geographic areas and allowed us to assess the extent to which each outcome varies across geographic areas. (25,26) We performed a series of multi-level logistic random intercept regression models. All analyses were conducted using the SAS procedure NLMIXED, version 9.3 (SAS Institute Inc., Cary, NC, US). Data publication guidelines of Statistics Canada were followed throughout the analysis. Ethics approval was obtained from the St Michael's Hospital Research Ethics Review Board.

RESULTS

Table 3 shows odds ratios for each health and behavioural outcome by quintile for each dimension of CAN-Marg. A quintile value of 5 reflects the greatest magnitude of marginalization (i.e., the most marginalized) and a quintile of 1 (the reference category) the least magnitude for each dimension (i.e., ethnic concentration, residential instability, material deprivation, dependency).

Table 3 shows that health and behavioural outcomes differ, depending on how marginalization is conceived of and measured. In most cases, living in areas with higher residential instability was significantly associated with greater health/behavioural problems such as binge drinking, smoking, disability, chronic diseases, low physical activity, and poor self-rated physical/mental health. Two exceptions were flu shots in the previous year and being overweight; in these cases, higher residential instability was associated with a lower likelihood of being overweight and a greater likelihood of having had a flu shot.

With the exception of self-perceived stress, living in areas with higher material deprivation was significantly associated with worse physical and mental health outcomes. Higher material deprivation was also associated with being overweight, being physically inactive, smoking, binge drinking, disability, and a lower likelihood of having had a flu shot in the previous year.

Living in areas with higher dependency was positively associated with being overweight, with disability, an inactive lifestyle, and having had a flu shot in the previous year. There was a greater risk of reporting chronic physical health problems and poor self-rated physical and mental health, and a lower risk of reporting binge drinking, self-perceived stress, smoking, and having a mood/anxiety disorder.

Ethnic concentration was associated with different outcomes from the other three dimensions in many cases. Living in areas with higher ethnic concentration was significantly associated with better health outcomes and more healthy behaviours. This was noted particularly for binge drinking, smoking, disability, chronic physical and mental health, and self-rated physical health. Only in the cases of high self-perceived stress and poor self-rated mental health was living in areas of higher ethnic concentration associated with poorer health.

Residential instability, material deprivation and ethnic concentration were negatively related to strong sense of community belonging, whereas dependency showed the reverse, higher dependency being associated with a greater sense of community belonging.

DISCUSSION

Because health inequalities are large in Canadian society and in many cases avoidable, there is an urgent need for accessible, census-based empirical tools in health inequalities research, like ABDIs that reflect contemporary patterns of social differentiation. The ABDIs developed in the past reflected what were important axes of social differentiation and marginalization of the time. In Canada, patterns of immigration, language, ethnicity, household size and structure, etc., are all associated with health and/or health care outcomes and are profoundly different than they were when the early ABDIs were developed. Consequently, we developed CAN-Marg to fill the need for an empirically derived, theoretically informed tool that is current with contemporary social forces and marginalization processes.

Empirical demonstrations of CAN-Marg in this and in previous studies using the tool reinforce the premise that underpinned its development: (13-15,27-30) different processes of marginalization and axes of social differentiation relate to health in ways that differ from those associated with the enduring construct of material deprivation, reinforcing the need for a tool like CAN-Marg.

Acknowledgements: The development of the Canadian Marginalization Index was supported by funding from the Social Sciences and Humanities Research Council of Canada (Standard Research Grant No. 410-2009-1894), the Canadian Institutes of Health Research, the Population Health Improvement Research Network (PHIRN--http://www.rrasp-phirn.ca/) and the Chair in Research on Urban Neighbourhoods, Community Housing and Health (CRUNCH--www.crunch.mcmaster.ca). J. Dunn is supported by a Chair in Applied Public Health award from the Canadian Institutes of Health Research and the Public Health Agency of Canada. The study was supported by the Keenan Research Centre in the Li Ka Shing Knowledge Institute of St. Michael's Hospital, Toronto, and the Ontario Ministry of Health and Long-Term Care. The opinions, results and conclusions are those of the authors and are independent of the funding and supporting agencies. No endorsement by the Ministry or supporting agencies is intended or should be inferred.

Conflict of Interest: None to declare.

REFERENCES

(1.) Schuurman N, Bell N, Dunn J, Oliver L. Deprivation indices, population health and geography: An evaluation of the spatial effectiveness of indices at multiple scales. J Urban Health 2007;84(4):591-603.

(2.) Townsend PDN. Inequalities in Health: The Black Report. Harmondsworth, UK: Penguin, 1982.

(3.) Krieger N, Williams DR, Moss NE. Measuring social class in US public health research: Concepts, methodologies, and guidelines. Annu Rev Public Health 1997;18(1):341-78.

(4.) Broadway MJ, Jesty G. Are Canadian inner cities becoming more dissimilar? An analysis of urban deprivation indicators. Urban Stud 1998;35(9):1423-38.

(5.) Frohlich N, Mustard C. A regional comparison of socioeconomic and health indices in a Canadian province. Soc Sci Med 1996;42(9):1273-81.

(6.) Carstairs V, Morris R. Deprivation and health. BMJ 1989;299(6713):1462.

(7.) Jarman B. Underprivileged areas: Validation and distribution of scores. BMJ1984;289(6458):1587-92.

(8.) Morris R, Carstairs V. Which deprivation? A comparison of selected deprivation indexes. J Public Health Med 1991;13(4):318-26.

(9.) Crampton PSC, Sutton F. NZDep91 Index of Deprivation Instruction Book. Wellington: Health Services Research Centre, New Zealand, 1997.

(10.) MacLeod CM, Eisenberg A. The normative dimensions of equality. In: Green DA, Kesselman JR (Eds.), Dimensions of Inequality in Canada. Vancouver, BC: UBC Press, 2006;33-64.

(11.) Kim JO, Mueller CW. Factor Analysis: Statistical Methods and Practical Issues. Beverly Hills, CA: Sage Publications, 1978.

(12.) Kim JO, Mueller CW. Introduction to Factor Analysis: What It Is and How To Do It. Beverly Hills: Sage Publications, 1978.

(13.) Matheson FI, White HL, Moineddin R, Dunn JR, Glazier RH. Drinking in context: The influence of gender and neighbourhood deprivation on alcohol consumption. J Epidemiol Community Health 2012;66(6):e4.

(14.) Urquia ML, Frank JW, Glazier RH, Moineddin R, Matheson FI, Gagnon AJ. Neighborhood context and infant birthweight among recent immigrant mothers: A multilevel analysis. Am J Public Health 2009;99(2):285-93.

(15.) Matheson FI, Moineddin R, Dunn JR, Creatore MI, Gozdyra P, Glazier RH. Urban neighborhoods, chronic stress, gender and depression. Soc Sci Med 2006;63(10):2604-16.

(16.) McLoone P, Boddy FA. Deprivation and mortality in Scotland, 1981 and 1991. BMJ1994;309(6967):1465-70.

(17.) Townsend P, Phillimore P, Beattie A. Health and Deprivation: Inequality and the North. New York, NY: Croom Helm, 1988.

(18.) Eames M, Ben-Shlomo Y, Marmot MG. Social deprivation and premature mortality: Regional comparison across England. BMJ1993;307(6912):1097-102.

(19.) Pampalon R, Raymond G. A deprivation index for health and welfare planning in Quebec. Chronic Dis Can 2000;21(3):104-13.

(20.) Grabb EG, Guppy LN (Eds.). Social Inequality in Canada: Patterns, Problems, and Policies, 5th ed. Scarborough, ON: Prentice-Hall Canada, 2009.

(21.) Braithwaite RL, Treadwell HM, Arriola KR. Health disparities and incarcerated women: A population ignored. Am J Public Health 2005;95(10):1679-81.

(22.) Matheson FI, Dunn JR, Smith KL, Moineddin R, Glazier RH. Canadian Marginalization Index User Guide. Hamilton, ON: McMaster University, 2011. Available at: http://www.crunch.mcmaster.ca/documents/CAN-Marg_user_ guide_1.0_FINAL_MAY2012.pdf (Accessed August 13, 2012).

(23.) Beland Y. Canadian Community Health Survey--Methodological overview. Health Rep 2002;13(3):9-14.

(24.) Wilkins R, Khan S. PCCF + Version 5H User's Guide: Automated Geographic Coding Based on the Statistics Canada Postal Code Conversion Files Including Postal Codes through October 2010. Ottawa: Health Statistics Division, Statistics Canada, 2011. Catalogue no. 82F0086-XDB.

(25.) Hox JJ. Applied Multilevel Analysis. Amsterdam, Netherlands: TT-Publikaties, 1995.

(26.) Raudenbush SW, Sampson R. Assessing direct and indirect effects in multilevel designs with latent variables. Sociol Methods Res 1999;28(2):123-53.

(27.) Matheson FI, Moineddin R, Glazier RH. The weight of place: A multilevel analysis of gender, neighborhood material deprivation, and body mass index among Canadian adults. Soc Sci Med 2008;66(3):675-90.

(28.) Matheson FI, White HL, Moineddin R, Dunn JR, Glazier RH. Neighbourhood chronic stress and gender inequalities in hypertension among Canadian adults: A multilevel analysis. J Epidemiol Community Health 2009;64(8):705-13.

(29.) Matheson FI, LaFreniere MC, White HL, Moineddin R, Dunn JR, Glazier RH. Influence of neighborhood deprivation, gender and ethno-racial origin on smoking behavior of Canadian youth. Prev Med 2011;52(5):376-80.

(30.) White HL, Matheson FI, Moineddin R, Dunn JR, Glazier RH. Neighbourhood deprivation and regional inequalities in self-reported health among Canadians: Are we equally at risk? Health & Place 2011;17(1):361-69.

Flora I. Matheson, PhD, [1,3,5] James R. Dunn, PhD, [1-3,5] Katherine L.W. Smith, MHSc, [1] Rahim Moineddin, PhD, [3,4] Richard H. Glazier, MD, PhD [1,3-5]

Author Affiliations

[1.] Centre for Research on Inner City Health, Keenan Research Centre, Li Ka Shing Knowledge Institute of St. Michael's Hospital, Toronto, ON

[2.] Department of Health, Aging and Society, McMaster University, Hamilton, ON

[3.] Institute for Clinical Evaluative Sciences, Toronto, ON

[4.] Department of Family and Community Medicine, University of Toronto, Toronto, ON

[5.] Dalla Lana School of Public Health, University of Toronto, Toronto, ON

Correspondence: Flora I. Matheson, Centre for Research on Inner City Health, St. Michael's Hospital, 30 Bond St., Toronto, ON M5B 1W8, Tel: 416-864-6060, ext. 77482, Fax: 416-864-5485, E-mail: mathesonf@smh.ca Table 1. Factor Loadings (FL) and Correlations (R) for the Four Dimensions of CAN-Marg Census Tracts 2001 2006 FL R FL R Residential Instability Proportion living alone 96 0.94 96 0.95 Proportion of youth population 95 0.88 79 0.66 aged 5-15 * Crowding: Average number of 93 0.90 92 0.89 persons per dwelling * Proportion multi-unit housing 80 0.87 81 0.88 Proportion of the population 75 0.89 71 0.87 that is married/common-law * Proportion of dwellings that 71 0.87 69 0.87 are owned * Proportion of residential 61 0.60 56 0.54 mobility (same house as 5 years ago) Material Deprivation Proportion 25+ without 90 0.77 89 0.74 certificate, diploma or degree Proportion of lone-parent 82 0.85 82 0.84 families Proportion government 77 0.87 52 0.49 transfer payment Proportion unemployment 15+ 72 0.78 61 0.72 Proportion below low income 65 0.86 49 0.78 cut-off Proportion of homes needing 56 0.54 68 0.68 major repair Dependency Proportion of seniors (65+) 88 0.90 89 0.91 Dependency ratio 91 0.86 92 0.89 (0-14 + 65+)/(15-64) Labour force participation 78 0.83 80 0.84 rate (aged 15 and older) * Ethnic Concentration Proportion of 5-year recent 95 0.93 93 0.92 immigrants Proportion of visible minority 97 0.93 96 0.90 Dissemination Areas 2001 2006 FL R FL R Residential Instability Proportion living alone 89 0.89 90 0.90 Proportion of youth population 74 0.71 70 0.65 aged 5-15 * Crowding: Average number of 87 0.83 86 0.82 persons per dwelling * Proportion multi-unit housing 85 0.87 85 0.87 Proportion of the population 80 0.87 78 0.85 that is married/common-law * Proportion of dwellings that 80 0.86 78 0.85 are owned * Proportion of residential 68 0.63 60 0.55 mobility (same house as 5 years ago) Material Deprivation Proportion 25+ without 76 0.72 76 0.72 certificate, diploma or degree Proportion of lone-parent 52 0.59 51 0.58 families Proportion government 70 0.80 70 0.79 transfer payment Proportion unemployment 15+ 70 0.69 64 0.63 Proportion below low income 51 0.71 39 0.60 cut-off Proportion of homes needing 57 0.49 57 0.52 major repair Dependency Proportion of seniors (65+) 88 0.91 89 0.92 Dependency ratio 79 0.69 84 0.77 (0-14 + 65+)/(15-64) Labour force participation 78 0.79 76 0.78 rate (aged 15 and older) * Ethnic Concentration Proportion of 5-year recent 88 0.84 85 0.83 immigrants Proportion of visible minority 96 0.89 94 0.87 * Reverse coded Table 2. Eigenvalues and Variance Explained for the Four Dimensions of CAN-Marg Marginalization Census Dissemination Dimension Tracts Areas 2001 2006 2001 2006 Residential Instability Eigenvalue 7.92 7.03 6.31 6.02 Proportion of variance explained 0.44 0.39 0.35 0.33 Material Deprivation Eigenvalue 2.77 2.75 2.96 1.95 Proportion of variance explained 0.15 0.15 0.16 0.11 Dependency Eigenvalue 2.31 1.60 2.09 2.99 Proportion of variance explained 0.13 0.09 0.12 0.17 Ethnic Concentration Eigenvalue 1.55 2.20 1.48 1.54 Proportion of variance explained 0.09 0.12 0.08 0.09 Table 3. Mixed Regression Model Results from Merged CAN-Marg and Combined CCHS 3.1 and 2007/08, Canada Outcome Quintile Odds Ratios (confidence intervals) Health Behaviours Residential Material Instability Deprivation Binge drinking Q2 1.07 (1.02,1.11) 0.99 (0.95,1.03) (>5 drinks Q3 1.06 (1.02,1.11) 1.01 (0.97,1.05) [greater than Q4 1.06 (1.02,1.11) 1.01 (0.97,1.05) or equal to] Q5 1.09 (1.05,1.14) 1.06 (1.02,1.11) once/month) Overweight Q2 1.15 (1.11,1.18) 1.12 (1.09,1.15) (BMI * [greater Q3 1.13 (1.10,1.17) 1.19 (1.15,1.23) than or equal to] Q4 1.09 (1.06,1.13) 1.24 (1.20,1.27) 25 kg/[m.sup.2]) Q5 0.93 (0.91,0.96) 1.35 (1.31,1.39) Flu shot in Q2 1.09 (1.05,1.14) 1.01 (0.97,1.05) past year Q3 1.10 (1.06,1.14) 0.98 (0.94,1.01) Q4 1.17 (1.13,1.21) 0.97 (0.94,1.01) Q5 1.17 (1.13,1.21) 0.89 (0.86,0.92) Current smoker Q2 1.24 (1.19,1.30) 1.35 (1.29,1.41) Q3 1.40 (1.34,1.46) 1.61 (1.54,1.68) Q4 1.57 (1.51,1.64) 1.89 (1.81,1.97) Q5 1.74 (1.67,1.82) 2.44 (2.35,2.54) Inactive Q2 1.09 (1.05,1.13) 1.16 (1.12,1.19) Q3 1.13 (1.10,1.17) 1.29 (1.25,1.33) Q4 1.18 (1.14,1.22) 1.44 (1.40,1.49) Q5 1.32 (1.27,1.36) 1.62 (1.57,1.67) Disability/activity Q2 1.21 (1.17,1.25) 1.16 (1.12,1.20) limitation Q3 1.31 (1.26,1.35) 1.25 (1.21,1.30) (sometimes/often) Q4 1.41 (1.36,1.46) 1.32 (1.28,1.37) Q5 1.49 (1.44,1.54) 1.47 (1.42,1.52) Health Outcomes Asthma Q2 1.03 (0.98,1.08) 1.04 (0.99,1.10) Q3 1.08 (1.03,1.14) 1.05 (1.00,1.10) Q4 1.12 (1.07,1.18) 1.14 (1.08,1.19) Q5 1.20 (1.14,1.26) 1.23 (1.17,1.28) Hypertension Q2 1.25 (1.20,1.31) 1.14 (1.10,1.19) Q3 1.37 (1.31,1.42) 1.28 (1.23,1.33) Q4 1.37 (1.32,1.43) 1.37 (1.32,1.42) Q5 1.39 (1.33,1.44) 1.47 (1.41,1.52) Diabetes Q2 1.27 (1.19,1.35) 1.36 (1.28,1.45) Q3 1.38 (1.30,1.47) 1.58 (1.49,1.68) Q4 1.42 (1.34,1.51) 1.67 (1.57,1.77) Q5 1.52 (1.44,1.61) 1.97 (1.86,2.08) Heart disease Q2 1.30 (1.22,1.39) 1.19 (1.12,1.27) Q3 1.49 (1.40,1.59) 1.36 (1.28,1.45) Q4 1.55 (1.45,1.64) 1.48 (1.39,1.57) Q5 1.77 (1.66,1.88) 1.65 (1.55,1.74) COPD * (age 30+) Q2 1.23 (1.08,1.42) 1.20 (1.05,1.37) Q3 1.30 (1.14,1.49) 1.25 (1.10,1.42) Q4 1.34 (1.17,1.53) 1.35 (1.18,1.53) Q5 1.65 (1.45,1.89) 1.39 (1.22,1.57) Emphysema (age 30+) Q2 1.29 (1.11,1.52) 1.35 (1.15,1.60) Q3 1.53 (1.32,1.78) 1.74 (1.50,2.04) Q4 1.66 (1.43,1.93) 2.08 (1.79,2.41) Q5 2.17 (1.87,2.51) 2.40 (2.07,2.77) Chronic bronchitis Q2 1.35 (1.22,1.48) 1.20 (1.09,1.33) Q3 1.55 (1.41,1.70) 1.59 (1.45,1.74) Q4 1.69 (1.54,1.85) 1.77 (1.61,1.94) Q5 1.99 (1.82,2.18) 2.28 (2.09,2.48) Mood Q2 1.12 (1.07,1.18) 1.13 (1.08,1.19) disorder/anxiety Q3 1.21 (1.16,1.28) 1.17 (1.11,1.23) Q4 1.35 (1.28,1.42) 1.23 (1.17,1.29) Q5 1.73 (1.64,1.81) 1.45 (1.38,1.52) Self-reported Health/Perceptions Self-perceived Q2 0.91 (0.88,0.94) 0.98 (0.95,1.02) stress (quite a Q3 0.91 (0.87,0.94) 0.98 (0.94,1.01) bit/extremely) Q4 0.94 (0.90,0.97) 0.97 (0.94,1.01) Q5 1.04 (1.00,1.07) 0.93 (0.90,0.96) Self-rated health Q2 1.26 (1.20,1.32) 1.31 (1.25,1.37) (poor/fair) Q3 1.44 (1.37,1.51) 1.57 (1.50,1.64) Q4 1.59 (1.52,1.67) 1.85 (1.77,1.93) Q5 1.87 (1.79,1.96) 2.35 (2.25,2.45) Self-rated mental Q2 1.16 (1.08,1.24) 1.20 (1.12,1.29) health (poor/fair) Q3 1.21 (1.13,1.29) 1.32 (1.23,1.41) Q4 1.45 (1.36,1.55) 1.52 (1.43,1.63) Q5 1.76 (1.65,1.88) 1.87 (1.76,2.00) Sense of community Q2 1.00 (0.96,1.04) 0.98 (0.94,1.02) belonging Q3 1.00 (0.96,1.04) 0.98 (0.94,1.01) (strong/somewhat Q4 0.91 (0.88,0.95) 0.94 (0.90,0.97) strong) Q5 0.67 (0.64,0.69) 0.97 (0.93,1.00) Outcome Quintile Odds Ratios (confidence intervals) Health Behaviours Dependency Ethnic Concentration Binge drinking Q2 0.86 (0.83,0.90) 1.05 (1.01,1.09) (>5 drinks Q3 0.82 (0.79,0.86) 1.06 (1.03,1.11) [greater than Q4 0.80 (0.77,0.84) 1.06 (1.02,1.10) or equal to] Q5 0.70 (0.67,0.73) 0.79 (0.75,0.82) once/month) Overweight Q2 1.04 (1.01,1.07) 0.92 (0.89,0.94) (BMI * [greater Q3 1.12 (1.09,1.15) 0.84 (0.81,0.86) than or equal to] Q4 1.22 (1.18,1.25) 0.78 (0.76,0.80) 25 kg/[m.sup.2]) Q5 1.29 (1.25,1.33) 0.64 (0.62,0.66) Flu shot in Q2 1.22 (1.18,1.27) 0.99 (0.96,1.02) past year Q3 1.34 (1.30,1.39) 0.95 (0.92,0.99) Q4 1.44 (1.39,1.49) 0.87 (0.84,0.90) Q5 1.81 (1.75,1.87) 0.86 (0.83,0.90) Current smoker Q2 0.93 (0.89,0.97) 0.98 (0.94,1.02) Q3 0.97 (0.93,1.01) 1.00 (0.96,1.04) Q4 0.98 (0.94,1.02) 0.99 (0.96,1.03) Q5 0.93 (0.89,0.97) 0.83 (0.79,0.86) Inactive Q2 1.01 (0.98,1.05) 0.94 (0.91,0.97) Q3 1.06 (1.03,1.10) 0.91 (0.88,0.94) Q4 1.11 (1.08,1.15) 0.89 (0.86,0.92) Q5 1.23 (1.19,1.27) 1.02 (0.99,1.06) Disability/activity Q2 1.12 (1.09,1.16) 0.95 (0.92,0.98) limitation Q3 1.24 (1.20,1.28) 0.90 (0.87,0.93) (sometimes/often) Q4 1.37 (1.33,1.42) 0.81 (0.78,0.84) Q5 1.67 (1.62,1.72) 0.71 (0.69,0.74) Health Outcomes Asthma Q2 0.96 (0.91,1.00) 1.06 (1.02,1.11) Q3 0.97 (0.93,1.02) 1.06 (1.02,1.11) Q4 0.98 (0.93,1.02) 1.09 (1.05,1.14) Q5 0.96 (0.92,1.01) 0.97 (0.92,1.02) Hypertension Q2 1.38 (1.33,1.44) 0.90 (0.87,0.93) Q3 1.64 (1.57,1.70) 0.78 (0.75,0.81) Q4 1.93 (1.86,2.00) 0.65 (0.63,0.68) Q5 2.51 (2.42,2.60) 0.60 (0.58,0.62) Diabetes Q2 1.30 (1.22,1.39) 0.93 (0.89,0.98) Q3 1.51 (1.42,1.61) 0.85 (0.81,0.89) Q4 1.85 (1.74,1.96) 0.72 (0.68,0.76) Q5 2.17 (2.05,2.29) 0.73 (0.69,0.77) Heart disease Q2 1.47 (1.38,1.58) 0.89 (0.85,0.94) Q3 1.87 (1.75,2.00) 0.80 (0.76,0.84) Q4 2.27 (2.13,2.42) 0.62 (0.59,0.66) Q5 3.12 (2.94,3.31) 0.55 (0.51,0.58) COPD * (age 30+) Q2 1.24 (1.08,1.43) 1.02 (0.92,1.13) Q3 1.35 (1.17,1.55) 1.02 (0.92,1.14) Q4 1.45 (1.27,1.66) 0.88 (0.78,0.99) Q5 1.71 (1.51,1.94) 0.72 (0.63,0.83) Emphysema (age 30+) Q2 1.34 (1.14,1.57) 0.88 (0.78,0.98) Q3 1.48 (1.26,1.73) 0.90 (0.80,1.01) Q4 1.76 (1.52,2.05) 0.79 (0.70,0.90) Q5 2.20 (1.91,2.53) 0.64 (0.55,0.75) Chronic bronchitis Q2 1.10 (1.01,1.21) 0.95 (0.88,1.02) Q3 1.28 (1.17,1.40) 0.92 (0.85,0.99) Q4 1.42 (1.30,1.54) 0.83 (0.77,0.90) Q5 1.61 (1.49,1.75) 0.80 (0.74,0.88) Mood Q2 0.92 (0.88,0.97) 1.08 (1.03,1.13) disorder/anxiety Q3 0.95 (0.90,1.00) 1.15 (1.10,1.20) Q4 0.96 (0.92,1.01) 1.18 (1.12,1.23) Q5 0.97 (0.93,1.01) 1.09 (1.04,1.14) Self-reported Health/Perceptions Self-perceived Q2 0.94 (0.91,0.97) 1.07 (1.03,1.10) stress (quite a Q3 0.89 (0.86,0.92) 1.18 (1.14,1.21) bit/extremely) Q4 0.81 (0.79,0.84) 1.25 (1.21,1.29) Q5 0.73 (0.71,0.76) 1.31 (1.27,1.36) Self-rated health Q2 1.17 (1.12,1.23) 0.92 (0.88,0.95) (poor/fair) Q3 1.37 (1.31,1.44) 0.85 (0.81,0.88) Q4 1.54 (1.47,1.61) 0.78 (0.75,0.82) Q5 1.95 (1.87,2.03) 0.83 (0.80,0.87) Self-rated mental Q2 0.95 (0.89,1.01) 0.98 (0.93,1.04) health (poor/fair) Q3 0.99 (0.93,1.06) 1.01 (0.95,1.07) Q4 1.01 (0.95,1.07) 1.07 (1.01,1.14) Q5 1.06 (0.99,1.12) 1.23 (1.15,1.30) Sense of community Q2 1.14 (1.10,1.19) 0.87 (0.84,0.90) belonging Q3 1.21 (1.17,1.26) 0.80 (0.78,0.83) (strong/somewhat Q4 1.32 (1.27,1.37) 0.67 (0.64,0.69) strong) Q5 1.48 (1.43,1.54) 0.60 (0.57,0.62) Note: Q1 is the reference category for all outcomes. * BMI = body mass index; COPD = chronic obstructive pulmonary disease.

Il y a longtemps qu'on utilise des indicateurs regionaux du statut socioeconomique; plus il y a de bases de donnees disponibles, plus on s'interesse a la recherche et a la surveillance en sante des populations, et plus on utilise ce qu'on appelle les <<indices regionaux de defavorisation>> ou IRD (1). En raison de cet interet croissant, nous avons cree l'indice de marginalisation canadien (CAN-Marg) a partir de recherches et de theories regionales existantes qui reliaient la marginalisation de certains quartiers avec la mauvaise sante. Mis au point par une equipe de recherche torontoise en 2006, le CAN-Marg est fonde sur le Recensement et derive geographiquement; il sert aux chercheurs qui etudient les inegalites de sante et d'autres problemes sociaux lies a la sante entre des segments demographiques ou des zones geographiques.

Le CAN-Marg va au-dela des definitions classiques de l'exclusion principalement saisies par les IRD actuels et passes (2-9). La plupart des IRD existants sont tres axes sur les aspects materiels de la defavorisation : le revenu, la possession d'une voiture, la propriete d'une maison. Ils s'inspirent en cela d'un modele oU les inegalites economiques sont primordiales. Toutefois, les societes des pays riches ont evolue depuis 30 ou 40 ans; d'autres facettes des inegalites peuvent etre tout aussi importantes pour la sante. Le CAN-Marg est donc un indice multidimensionnel qui permet aux chercheurs et aux analystes de politiques et de programmes d'examiner plusieurs dimensions de la marginalisation dans les zones urbaines et rurales du Canada et leurs effets sur la sante et sur divers resultats sociaux. Le CAN-Marg a ete elabore en rapport avec la marginalisation, laquelle est un processus qui cree des inegalites selon plusieurs axes de differenciation sociale au Canada (10). Les quatre dimensions incluses dans l'indice sont l'instabilite residentielle, la defavorisation materielle, la concentration ethnique et la dependance economique.

L'indice CAN-Marg s'est initialement construit sur un cadre theorique fonde sur les travaux anterieurs dans le domaine de la defavorisation et de la marginalisation, mais en l'elaborant, nous avons laisse place a la possibilite d'inclure empiriquement d'autres formes de marginalisation, par exemple la marginalisation en fonction de l'identite ethno-raciale, du statut d'immigrant, du stade de la vie et de la composition des menages (2-9). Compte tenu de cette conception theorique elargie des differentes formes de marginalisation dans la societe contemporaine, nous avons reuni un plus grand eventail d'indicateurs possibles tires du Recensement; l'indice a ensuite ete derive empiriquement a l'aide d'une analyse factorielle en composantes principales (11,12). Nous avons voulu creer un indice qui saisit la nature des variations selon le lieu pour un vaste eventail d'indicateurs de marginalisation et qui est independant de toute association avec les resultats sanitaires, afin d'eviter les raisonnements circulaires ou tautologiques (l'argumentation selon laquelle l'indice est un bon indicateur a utiliser dans la recherche en sante parce qu'il est associe a la sante). Cela dit, plus loin dans cet article nous montrons que le CAN-Marg parvient effectivement a predire les resultats de sante d'une population.

Le but de cet article est de decrire le CAN-Marg, d'en montrer la stabilite dans le temps et d'une zone geographique a l'autre (p. ex., entre des villes et des regions rurales) et d'en illustrer l'association avec certains resultats sanitaires et comportementaux. Ce dernier objectif est un aspect important du projet, car le CAN-Marg ne contient aucune information sanitaire en soi, mais de nombreux chercheurs constatent que la marginalisation, telle que mesuree par des indices composites regionaux, presente une forte correlation avec les indicateurs d'etat sanitaire (8,9,13-18).

Creation de l'indice de marginalisation canadien

La selection des caracteristiques du quartier pour une telle analyse est inspiree de la recherche anterieure sur les IRD (6,9,17,19) et de perspectives theoriques contemporaines sur les inegalites et la marginalisation dans la societe canadienne (10,20). Les etudes publiees sur la defavorisation et l'instabilite residentielle ont ete essentielles comme point de depart; elles nous ont fourni les variables d'entree de l'analyse factorielle (6,9,10,17,19,20). L'analyse initiale a ete menee a l'aide des donnees des secteurs de recensement (SR) de 2001 (agglomerations urbaines). En tout, 42 indicateurs ont ete crees a partir des donnees du Recensement pour entrer dans l'analyse factorielle. Les indicateurs de faible saturation ont ete elimines de facon iterative, apres quoi quatre facteurs ont emerge, et 18 indicateurs par SR sont restes. Les facteurs ont ete construits par rotation oblique, ce qui permet leur covariation (11,12,21). Nous avons aussi estime une matrice orthogonale de facteurs qui a donne des resultats identiques en substance. Nous avons repete l'analyse factorielle a l'aide des 18 memes indicateurs pour les aires de diffusion (AD) des Recensements de 2001 et de 2006 et pour les SR de 2006.

Le tableau 1 montre les indicateurs associes a chaque dimension de la marginalisation. Les correlations entre les indicateurs et telle ou telle dimension sont assez stables dans le temps et d'une zone geographique a l'autre, la majorite etant superieures a 0,65. Les saturations des facteurs sont aussi assez stables (>0,50 pour la plupart). Les saturations pour la proportion de gens sous le seuil de faible revenu (voir la defavorisation materielle) etaient cependant inferieures en 2006 (<0,50). Par le passe, le revenu etait une composante principale des indices de defavorisation; il y a donc une justification theorique a le placer sous la dimension <<defavorisation >> (3,4,19). En general toutefois, les saturations et les valeurs propres sont semblables d'une annee et d'une zone geographique a l'autre pour une meme dimension, et la proportion de la variance expliquee pour les quatre dimensions se situe entre 70% et 80% (tableau 2). L'instabilite residentielle est la dimension dominante, avec les valeurs propres les plus elevees des quatre; viennent ensuite, dans l'ordre, la defavorisation materielle, la dependance economique et la concentration ethnique.

Les saturations ont servi a calculer des scores factoriels, lesquels nous ont permis de creer un indice continu distinct pour chacune des quatre dimensions. Chaque dimension est une echelle asymetriquement standardisee. Pour les besoins de l'article, les scores factoriels sont utilises comme quintiles. Les quintiles du CAN-Marg ont ete generes en classant les AD en ordre croissant de marginalisation et en affectant un nombre egal d'AD a chaque quintile. Ces derniers ont ete crees en triant les donnees de marginalisation en cinq groupes, classes de 1 (le moins marginalise) a 5 (le plus marginalise). Chaque groupe contient le cinquieme des unites geographiques. Par exemple, si une region a une valeur de 5 sur l'echelle de defavorisation materielle, cela signifie qu'elle se situe parmi les 20% des regions les plus defavorisees du Canada (22).

Associations entre la marginalisation et les resultats sanitaires ou comportementaux

Nous avons ensuite examine la relation entre les quatre dimensions de l'indice CAN-Marg 2006 et 18 resultats sanitaires et comportementaux a l'aide d'une modelisation multiniveaux en nous servant des AD comme unites regionales d'analyse. Les donnees au niveau des particuliers sont derivees de deux cycles (3.1 et 2007-2008) de l'Enquete sur la sante dans les collectivites canadiennes (ESCC), une enquete transversale representative a l'echelle nationale qui fournit des renseignements detailles sur les determinants de la sante et les resultats sanitaires pour les particuliers (23). Les repondants de l'ESCC 2007-2008 avaient les AD de 2006 dans leur jeu de donnees, tandis que ceux du cycle 3.1 (2005) avaient les AD de 2001. Pour le cycle 3.1, les zones geographiques de 2001 ont donc ete converties en zones geographiques de 2006 a l'aide de l'affectation code postal-AD du cycle 2007-2008 et du Fichier de conversion des codes postaux plus (version 5H), qui contient un algorithme pour affecter les zones geographiques du Recensement d'apres le code postal (24).

Approche statistique

Les analyses descriptives ont ete ponderees a l'aide du poids d'echantillonnage combine fourni par Statistique Canada. Les taux de prevalence ont ete ponderes et presentes avec des frequences d'echantillonnage ponderees. Notre demarche pour evaluer les associations entre la marginalisation et les resultats sanitaires et comportementaux de l'ESCC a consiste a utiliser la modelisation multiniveaux, qui tient compte de la possibilite d'observer des grappes a l'interieur des zones geographiques et nous a permis d'evaluer la mesure dans laquelle chaque resultat varie selon la zone geographique (25,26). Nous avons execute une serie de modeles de regression logistique multiniveaux avec ordonnee a l'origine aleatoire. Toutes les analyses ont ete effectuees a l'aide de la procedure NLMIXED, version 9.3, du SAS (SAS Institute Inc., Cary [Caroline du Nord], Etats-Unis). Les lignes directrices de Statistique Canada sur la publication de donnees ont ete respectees tout au long de l'analyse. L'ethique de la recherche a ete approuvee par le comite de deontologie de l'hopital St. Michael.

RESULTATS

Le tableau 3 presente les rapports de cotes de chaque resultat sanitaire et comportemental par quintile pour chaque dimension du CAN-Marg. Le quintile 5 reflete la marginalisation la plus importante (c.-a-d. le groupe le plus marginalise) et le quintile 1 (la categorie de reference) l'envergure la moins grande pour chaque dimension (concentration ethnique, instabilite residentielle, defavorisation materielle, dependance economique).

Comme on le voit au tableau 3, les resultats sanitaires et comportementaux different selon la facon dont on concoit et dont on mesure la marginalisation. Dans la plupart des cas, le fait de vivre dans des zones oU l'instabilite residentielle est elevee presente une correlation significative avec des problemes de sante ou de comportement plus importants, comme les exces d'alcool, le tabagisme, l'invalidite, les maladies chroniques, l'activite physique faible et la mauvaise sante physique ou mentale autoevaluee. Nous avons note deux exceptions: le vaccin contre la grippe recu au cours de l'annee precedente et le surpoids; dans ces cas, une instabilite residentielle superieure est associee a une probabilite inferieure d'etre en surpoids et a une probabilite superieure d'avoir recu un vaccin contre la grippe.

A l'exception du stress autopercu, le fait de vivre dans des zones oU la defavorisation materielle est elevee presente une correlation significative avec de moins bons resultats de sante physique et mentale. Une defavorisation materielle elevee est aussi associee au surpoids, a l'inactivite physique, au tabagisme, aux exces d'alcool, a l'invalidite et a une probabilite inferieure d'avoir recu un vaccin contre la grippe au cours de l'annee precedente.

Le fait de vivre dans des zones oU la dependance economique est elevee est associe positivement au surpoids, a l'invalidite, a un mode de vie sedentaire et au fait d'avoir recu un vaccin contre la grippe au cours de l'annee precedente. La probabilite de declarer des problemes de sante chroniques et une mauvaise sante physique et mentale autoevaluee est superieure, et la probabilite de declarer des exces d'alcool, un stress autopercu, un tabagisme et un trouble de l'humeur ou un trouble anxieux est inferieure.

La concentration ethnique est associee a des resultats differents de ceux des trois autres dimensions dans bien des cas. Le fait de vivre dans des zones oU la concentration ethnique est elevee presente une correlation significative avec de meilleurs resultats sanitaires et des comportements plus sains. Nous l'avons note en particulier pour les exces d'alcool, le tabagisme, l'invalidite, les problemes de sante physique et mentale chroniques et la sante physique autoevaluee. Ce n'est que dans les cas oU le stress autopercu est eleve et oU la sante mentale autoevaluee est mauvaise que le fait de vivre dans les zones a forte concentration ethnique est associe a une moins bonne sante.

L'instabilite residentielle, la defavorisation materielle et la concentration ethnique sont negativement associees a un fort sentiment d'appartenance communautaire, tandis que la dependance economique presente l'association contraire: une dependance elevee etant associee a un plus fort sentiment d'appartenance communautaire.

DISCUSSION_

Comme les inegalites de sante sont grandes dans la societe canadienne et qu'elles sont evitables dans bien des cas, il existe un urgent besoin d'outils accessibles, fondes sur le Recensement, pour effectuer de la recherche sur le sujet; les indices regionaux de defavorisation (IRD) en sont un exemple, car ils refletent les grandes tendances contemporaines de la differenciation sociale. Les IRD elabores par le passe refletaient des axes de differenciation sociale et de marginalisation qui etaient importants a l'epoque. Au Canada, les grandes tendances de l'immigration, des langues, de l'ethnicite, de la taille et de la structure des menages, etc., sont toutes associees a la sante et/ou aux resultats des soins de sante et sont profondement differentes de ce qu'elles etaient lorsque les premiers IRD ont ete mis au point. C'est pourquoi nous avons elabore le CAN-Marg: il nous manquait un outil derive empiriquement et reposant sur des bases theoriques qui soit au diapason des forces sociales et des processus de marginalisation de notre epoque.

Les demonstrations empiriques du CAN-Marg, dans cet article et dans les etudes anterieures portant sur cet outil, renforcent la premisse qui a sous-tendu son elaboration (13-15,27-30), a savoir que des processus de marginalisation et des axes de differenciation sociale differents sont differemment associes a la sante que le construit de la defavorisation materielle (utilise depuis beaucoup plus longtemps), ce qui renforce le besoin d'un outil comme le CAN-Marg.

Remerciements: L'elaboration de l'indice de marginalisation canadien a beneficie du financement du Conseil de recherches en sciences humaines du Canada (subvention ordinaire de recherche no 410-2009-1894), des Instituts de recherche en sante du Canada, du Reseau de recherche sur l'amelioration de la sante des populations (RRASP--www.rrasp-phirn.ca) et de la chaire de recherche sur les quartiers urbains, le logement communautaire et la sante de l'Universite McMaster (CRUNCH--www.crunch.mcmaster.ca). J. Dunn est titulaire d'une bourse des Instituts de recherche en sante du Canada et de l'Agence de la sante publique du Canada (chaire en sante publique appliquee). L'etude est appuyee par le Centre de recherche Keenan a l'Institut des connaissances Li Ka Shing de l'hopital St. Michael de Toronto et par le ministere de la Sante et des Soins de longue duree de l'Ontario. Les opinions, resultats et conclusions du present article sont ceux des auteurs et ne dependent ni du financement recu, ni des organismes de soutien. Ils ne representent aucunement une approbation expresse ou implicite du Ministere ni des organismes de soutien.

Conflit d'interets: Aucun a declarer.

BIBLIOGRAPHIE

(1.) Schuurman N, Bell N, Dunn J, Oliver L. Deprivation indices, population health and geography: An evaluation of the spatial effectiveness of indices at multiple scales. J Urban Health 2007;84(4):591-603.

(2.) Townsend PDN. Inequalities in Health: The Black Report. Harmondsworth (Royaume-Uni), Penguin, 1982.

(3.) Krieger N, Williams DR, Moss NE. Measuring social class in US public health research: Concepts, methodologies, and guidelines. Annu Rev Public Health 1997;18(1):341-78.

(4.) Broadway MJ, Jesty G. Are Canadian inner cities becoming more dissimilar? An analysis of urban deprivation indicators. Urban Stud 1998;35(9)1423-38.

(5.) Frohlich N, Mustard C. A regional comparison of socioeconomic and health indices in a Canadian province. Soc Sci Med 1996;42(9):1273-81.

(6.) Carstairs V, Morris R. Deprivation and health. BMJ 1989;299(6713):1462.

(7.) Jarman B. Underprivileged areas: Validation and distribution of scores. BMJ1984;289(6458):1587-92.

(8.) Morris R, Carstairs V. Which deprivation? A comparison of selected deprivation indexes. J Public Health Med 1991;13(4):318-26.

(9.) Crampton PSC, Sutton F. NZDep91 Index of Deprivation Instruction Book. Wellington (Nouvelle-Zelande), Health Services Research Centre, 1997.

(10.) MacLeod CM, Eisenberg A. The normative dimensions of equality. Dans Green DA, Kesselman JR (ed.), Dimensions of Inequality in Canada. Vancouver (Colombie-Britannique), UBC Press, 2006:33-64.

(11.) Kim JO, Mueller CW. Factor Analysis: Statistical Methods and Practical Issues. Beverly Hills (Californie), Sage Publications, 1978.

(12.) Kim JO, Mueller CW. Introduction to Factor Analysis: What It Is and How to Do It. Beverly Hills, Sage Publications, 1978.

(13.) Matheson FI, White HL, Moineddin R, Dunn JR, Glazier RH. Drinking in context: The influence of gender and neighbourhood deprivation on alcohol consumption. J Epidemiol Community Health 2012;66(6):e4.

(14.) Urquia ML, Frank JW, Glazier RH, Moineddin R, Matheson FI, Gagnon AJ. Neighborhood context and infant birthweight among recent immigrant mothers: A multilevel analysis. Am J Public Health 2009;99(2):285-93.

(15.) Matheson FI, Moineddin R, Dunn JR, Creatore MI, Gozdyra P, Glazier RH. Urban neighborhoods, chronic stress, gender and depression. Soc Sci Med 2006;63(10):2604-16.

(16.) McLoone P, Boddy FA. Deprivation and mortality in Scotland, 1981 and 1991. BMJ1994;309(6967):1465-70.

(17.) Townsend P, Phillimore P, Beattie A. Health and Deprivation: Inequality and the North. New York, Croom Helm, 1988.

(18.) Eames M, Ben-Shlomo Y, Marmot MG. Social deprivation and premature mortality: Regional comparison across England. BMJ1993;307(6912):1097-102.

(19.) Pampalon R, Raymond G. Un indice de defavorisation pour la planification de la sante et du bien-etre au Quebec. Maladies chroniques au Canada 2000;21(3):199-213.

(20.) Grabb EG, Guppy LN (ed.). Social Inequality in Canada: Patterns, Problems, and Policies. 5e ed., Scarborough (Ontario), Prentice-Hall Canada, 2009.

(21.) Braithwaite RL, Treadwell HM, Arriola KR. Health disparities and incarcerated women: A population ignored. Am J Public Health 2005;95(10): 1679-81.

(22.) Matheson FI, Dunn JR, Smith KL, Moineddin R, Glazier RH. Canadian Marginalization Index User Guide. Hamilton (Ontario), McMaster University, 2011 (consulte le 13 aout 2012). Internet: http://www.crunch.mcmaster.ca/documents/CAN-Marg_user_guide_1.0_FINAL_MAY2012.pdf.

(23.) Beland Y. Enquete sur la sante dans les collectivites canadiennes: apercu de la methodologie. Rapports sur la sante 2002;13(3):9-15.

(24.) Wilkins R, Khan S. FCCP+ Version 5H Guide de l'utilisateur. Logiciel de codage geographique base sur les Fichiers de conversion des codes postaux de Statistique Canada mis a jour en octobre 2010. Ottawa, Division des statistiques sur la sante, Statistique Canada, no 82F0086-XDB au catalogue, 2011.

(25.) Hox JJ. Applied Multilevel Analysis. Amsterdam (Pays-Bas), TT-Publikaties, 1995.

(26.) Raudenbush SW, Sampson R. Assessing direct and indirect effects in multilevel designs with latent variables. Sociol Methods Res 1999;28(2):123-53.

(27.) Matheson FI, Moineddin R, Glazier RH. The weight of place: A multilevel analysis of gender, neighborhood material deprivation, and body mass index among Canadian adults. Soc Sci Med 2008;66(3):675-90.

(28.) Matheson FI, White HL, Moineddin R, Dunn JR, Glazier RH. Neighbourhood chronic stress and gender inequalities in hypertension among Canadian adults: A multilevel analysis. J Epidemiol Community Health 2009;64(8):705-13.

(29.) Matheson FI, LaFreniere MC, White HL, Moineddin R, Dunn JR, Glazier RH. Influence of neighborhood deprivation, gender and ethno-racial origin on smoking behavior of Canadian youth. Prev Med 2011;52(5):376-80.

(30.) White HL, Matheson FI, Moineddin R, Dunn JR, Glazier RH. Neighbourhood deprivation and regional inequalities in self-reported health among Canadians: Are we equally at risk? Health & Place 2011;17(1):361-9.

Flora I. Matheson, Ph.D. [1,3,5], James R. Dunn, Ph.D. [1-3,5], Katherine L.W. Smith, M.Sc.S. [1], Rahim Moineddin, Ph.D. [3,4], Richard H. Glazier, M.D., PhD [1,3-5]

Affiliations des auteurs

[1.] Centre for Research on Inner City Health, Centre de recherche Keenan, Institut des connaissances Li Ka Shing de l'hopital St. Michael, Toronto (Ontario)

[2.] Departement de la sante, du vieillissement et de la societe, Universite McMaster, Hamilton (Ontario)

[3.] Institut de recherche en services de sante, Toronto (Ontario)

[4.] Departement de medecine familiale et communautaire, Universite de Toronto, Toronto (Ontario)

[5.] Ecole de sante publique Dalla Lana, Universite de Toronto, Toronto (Ontario)

Correspondance: Flora I. Matheson, Centre for Research on Inner City Health, St. Michael's Hospital, 30 Bond St., Toronto (Ontario) M5B 1W8, tel.: 416-864-6060, poste 77482, telec.: 416-864-5485, courriel: mathesonf@smh.ca Tableau 1. Saturations des facteurs (SF) et correlations (R) pour les quatre dimensions du CAN-Marg Secteurs de recensement 2001 2006 SF R SF R Instabilite residentielle Proportion de personnes 96 0,94 96 0,95 vivant seules Proportion de jeunes de 95 0,88 79 0,66 5 a 15 ans * Entassement: nombre moyen 93 0,90 92 0,89 de personnes par logement * Proportion de complexes 80 0,87 81 0,88 d'habitation Proportion de personnes 75 0,89 71 0,87 mariees/en union libre * Proportion de logements 71 0,87 69 0,87 dont l'occupant est proprietaire * Proportion de la mobilite 61 0,60 56 0,54 residentielle (meme maison qu'il y a 5 ans) Defavorisation materielle Proportion de personnes de 90 0,77 89 0,74 25 ans et plus sans certificat ni diplome Proportion de familles 82 0,85 82 0,84 monoparentales Proportion de paiements de 77 0,87 52 0,49 transfert gouvernementaux Proportion de personnes de 72 0,78 61 0,72 15 ans et plus au chomage Proportion de personnes 65 0,86 49 0,78 sous le seuil de faible revenu Proportion de maisons 56 0,54 68 0,68 necessitant des reparations majeures Dependance economique Proportion de personnes 88 0,90 89 0,91 agees (65 ans et plus) Rapport de dependance 91 0,86 92 0,89 (0-14 + 65+)/(15-64) Taux de participation a 78 0,83 80 0,84 la population active (15 ans et plus) * Concentration ethnique Proportion d'immigrants 95 0,93 93 0,92 recents (depuis moins de 5 ans) Proportion de minorites 97 0,93 96 0,90 visibles Aires de diffusion 2001 2006 SF R SF R Instabilite residentielle Proportion de personnes 89 0,89 90 0,90 vivant seules Proportion de jeunes de 74 0,71 70 0,65 5 a 15 ans * Entassement: nombre moyen 87 0,83 86 0,82 de personnes par logement * Proportion de complexes 85 0,87 85 0,87 d'habitation Proportion de personnes 80 0,87 78 0,85 mariees/en union libre * Proportion de logements 80 0,86 78 0,85 dont l'occupant est proprietaire * Proportion de la mobilite 68 0,63 60 0,55 residentielle (meme maison qu'il y a 5 ans) Defavorisation materielle Proportion de personnes de 76 0,72 76 0,72 25 ans et plus sans certificat ni diplome Proportion de familles 52 0,59 51 0,58 monoparentales Proportion de paiements de 70 0,80 70 0,79 transfert gouvernementaux Proportion de personnes de 70 0,69 64 0,63 15 ans et plus au chomage Proportion de personnes 51 0,71 39 0,60 sous le seuil de faible revenu Proportion de maisons 57 0,49 57 0,52 necessitant des reparations majeures Dependance economique Proportion de personnes 88 0,91 89 0,92 agees (65 ans et plus) Rapport de dependance 79 0,69 84 0,77 (0-14 + 65+)/(15-64) Taux de participation a 78 0,79 76 0,78 la population active (15 ans et plus) * Concentration ethnique Proportion d'immigrants 88 0,84 85 0,83 recents (depuis moins de 5 ans) Proportion de minorites 96 0,89 94 0,87 visibles * Codes de facon inverse Tableau 2. Valeurs propres et variances expliquees pour les quatre dimensions du CAN-Marg Dimension de la Secteurs de Aires de marginalisation recensement diffusion 2001 2006 2001 2006 Instabilite residentielle Valeur propre 7,92 7,03 6,31 6,02 Proportion de variance expliquee 0,44 0,39 0,35 0,33 Defavorisation materielle Valeur propre 2,77 2,75 2,96 1,95 Proportion de variance expliquee 0,15 0,15 0,16 0,11 Dependance economique Valeur propre 2,31 1,60 2,09 2,99 Proportion de variance expliquee 0,13 0,09 0,12 0,17 Concentration ethnique Valeur propre 1,55 2,20 1,48 1,54 Proportion de variance expliquee 0,09 0,12 0,08 0,09 Tableau 3. Resultats mixtes des modeles de regression a partir des versions fusionnees du CAN-Marg et des cycles ESCC 3.1 et 2007-2008 combines, Canada Resultat Quintile Rapports de cotes (intervalles de confiance) Instabilite Defavorisation residentielle materielle Comportements lies a la sante Exces d'alcool Q2 1,07 (1,02,1,11) 0,99 (0,95,1,03) (>5 verres Q3 1,06 (1,02,1,11) 1,01 (0,97,1,05) [superieur ou Q4 1,06 (1,02,1,11) 1,01 (0,97,1,05) egal a] une Q5 1,09 (1,05,1,14) 1,06 (1,02,1,11) fois/mois) Surpoids Q2 1,15 (1,11,1,18) 1,12 (1,09,1,15) (IMC * Q3 1,13 (1,10,1,17) 1,19 (1,15,1,23) [superieur ou Q4 1,09 (1,06,1,13) 1,24 (1,20,1,27) egal a] Q5 0,93 (0,91,0,96) 1,35 (1,31,1,39) 25 kg/[m.sup.2]) Vaccin antigrippal Q2 1,09 (1,05,1,14) 1,01 (0,97,1,05) au cours de la Q3 1,10 (1,06,1,14) 0,98 (0,94,1,01) derniere annee Q4 1,17 (1,13,1,21) 0,97 (0,94,1,01) Q5 1,17 (1,13,1,21) 0,89 (0,86,0,92) Fumeur actuel Q2 1,24 (1,19,1,30) 1,35 (1,29,1,41) Q3 1,40 (1,34,1,46) 1,61 (1,54,1,68) Q4 1,57 (1,51,1,64) 1,89 (1,81,1,97) Q5 1,74 (1,67,1,82) 2,44 (2,35,2,54) Inactivite Q2 1,09 (1,05,1,13) 1,16 (1,12,1,19) physique Q3 1,13 (1,10,1,17) 1,29 (1,25,1,33) Q4 1,18 (1,14,1,22) 1,44 (1,40,1,49) Q5 1,32 (1,27,1,36) 1,62 (1,57,1,67) Invalidite/ Q2 1,21 (1,17,1,25) 1,16 (1,12,1,20) limitation Q3 1,31 (1,26,1,35) 1,25 (1,21,1,30) d'activite Q4 1,41 (1,36,1,46) 1,32 (1,28,1,37) (parfois/souvent) Q5 1,49 (1,44,1,54) 1,47 (1,42,1,52) Resultats sanitaires Asthme Q2 1,03 (0,98,1,08) 1,04 (0,99,1,10) Q3 1,08 (1,03,1,14) 1,05 (1,00,1,10) Q4 1,12 (1,07,1,18) 1,14 (1,08,1,19) Q5 1,20 (1,14,1,26) 1,23 (1,17,1,28) Hypertension Q2 1,25 (1,20,1,31) 1,14 (1,10,1,19) arterielle Q3 1,37 (1,31,1,42) 1,28 (1,23,1,33) Q4 1,37 (1,32,1,43) 1,37 (1,32,1,42) Q5 1,39 (1,33,1,44) 1,47 (1,41,1,52) Diabete Q2 1,27 (1,19,1,35) 1,36 (1,28,1,45) Q3 1,38 (1,30,1,47) 1,58 (1,49,1,68) Q4 1,42 (1,34,1,51) 1,67 (1,57,1,77) Q5 1,52 (1,44,1,61) 1,97 (1,86,2,08) Cardiopathie Q2 1,30 (1,22,1,39) 1,19 (1,12,1,27) Q3 1,49 (1,40,1,59) 1,36 (1,28,1,45) Q4 1,55 (1,45,1,64) 1,48 (1,39,1,57) Q5 1,77 (1,66,1,88) 1,65 (1,55,1,74) MPOC * (30 ans Q2 1,23 (1,08,1,42) 1,20 (1,05,1,37) et plus) Q3 1,30 (1,14,1,49) 1,25 (1,10,1,42) Q4 1,34 (1,17,1,53) 1,35 (1,18,1,53) Q5 1,65 (1,45,1,89) 1,39 (1,22,1,57) Emphyseme (30 Q2 1,29 (1,11,1,52) 1,35 (1,15,1,60) ans et plus) Q3 1,53 (1,32,1,78) 1,74 (1,50,2,04) Q4 1,66 (1,43,1,93) 2,08 (1,79,2,41) Q5 2,17 (1,87,2,51) 2,40 (2,07,2,77) Bronchite Q2 1,35 (1,22,1,48) 1,20 (1,09,1,33) chronique Q3 1,55 (1,41,1,70) 1,59 (1,45,1,74) Q4 1,69 (1,54,1,85) 1,77 (1,61,1,94) Q5 1,99 (1,82,2,18) 2,28 (2,09,2,48) Trouble de Q2 1,12 (1,07,1,18) 1,13 (1,08,1,19) l'humeur/anxiete Q3 1,21 (1,16,1,28) 1,17 (1,11,1,23) Q4 1,35 (1,28,1,42) 1,23 (1,17,1,29) Q5 1,73 (1,64,1,81) 1,45 (1,38,1,52) Etat de sante autodeclare/ autopercu Stress autopercu Q2 0,91 (0,88,0,94) 0,98 (0,95,1,02) (important/ Q3 0,91 (0,87,0,94) 0,98 (0,94,1,01) extreme) Q4 0,94 (0,90,0,97) 0,97 (0,94,1,01) Q5 1,04 (1,00,1,07) 0,93 (0,90,0,96) Sante autoevaluee Q2 1,26 (1,20,1,32) 1,31 (1,25,1,37) (mauvaise/ Q3 1,44 (1,37,1,51) 1,57 (1,50,1,64) moyenne) Q4 1,59 (1,52,1,67) 1,85 (1,77,1,93) Q5 1,87 (1,79,1,96) 2,35 (2,25,2,45) Sante mentale Q2 1,16 (1,08,1,24) 1,20 (1,12,1,29) autoevaluee Q3 1,21 (1,13,1,29) 1,32 (1,23,1,41) (mauvaise/ Q4 1,45 (1,36,1,55) 1,52 (1,43,1,63) moyenne) Q5 1,76 (1,65,1,88) 1,87 (1,76,2,00) Sentiment Q2 1,00 (0,96,1,04) 0,98 (0,94,1,02) d'appartenance Q3 1,00 (0,96,1,04) 0,98 (0,94,1,01) communautaire Q4 0,91 (0,88,0,95) 0,94 (0,90,0,97) (fort/assez Q5 0,67 (0,64,0,69) 0,97 (0,93,1,00) fort) Resultat Quintile Rapports de cotes (intervalles de confiance) Dependance Concentration ethnique Comportements lies a la sante Exces d'alcool Q2 0,86 (0,83,0,90) 1,05 (1,01,1,09) (>5 verres Q3 0,82 (0,79,0,86) 1,06 (1,03,1,11) [superieur ou Q4 0,80 (0,77,0,84) 1,06 (1,02,1,10) egal a] une Q5 0,70 (0,67,0,73) 0,79 (0,75,0,82) fois/mois) Surpoids Q2 1,04 (1,01,1,07) 0,92 (0,89,0,94) (IMC * Q3 1,12 (1,09,1,15) 0,84 (0,81,0,86) [superieur ou Q4 1,22 (1,18,1,25) 0,78 (0,76,0,80) egal a] Q5 1,29 (1,25,1,33) 0,64 (0,62,0,66) 25 kg/[m.sup.2]) Vaccin antigrippal Q2 1,22 (1,18,1,27) 0,99 (0,96,1,02) au cours de la Q3 1,34 (1,30,1,39) 0,95 (0,92,0,99) derniere annee Q4 1,44 (1,39,1,49) 0,87 (0,84,0,90) Q5 1,81 (1,75,1,87) 0,86 (0,83,0,90) Fumeur actuel Q2 0,93 (0,89,0,97) 0,98 (0,94,1,02) Q3 0,97 (0,93,1,01) 1,00 (0,96,1,04) Q4 0,98 (0,94,1,02) 0,99 (0,96,1,03) Q5 0,93 (0,89,0,97) 0,83 (0,79,0,86) Inactivite Q2 1,01 (0,98,1,05) 0,94 (0,91,0,97) physique Q3 1,06 (1,03,1,10) 0,91 (0,88,0,94) Q4 1,11 (1,08,1,15) 0,89 (0,86,0,92) Q5 1,23 (1,19,1,27) 1,02 (0,99,1,06) Invalidite/ Q2 1,12 (1,09,1,16) 0,95 (0,92,0,98) limitation Q3 1,24 (1,20,1,28) 0,90 (0,87,0,93) d'activite Q4 1,37 (1,33,1,42) 0,81 (0,78,0,84) (parfois/souvent) Q5 1,67 (1,62,1,72) 0,71 (0,69,0,74) Resultats sanitaires Asthme Q2 0,96 (0,91,1,00) 1,06 (1,02,1,11) Q3 0,97 (0,93,1,02) 1,06 (1,02,1,11) Q4 0,98 (0,93,1,02) 1,09 (1,05,1,14) Q5 0,96 (0,92,1,01) 0,97 (0,92,1,02) Hypertension Q2 1,38 (1,33,1,44) 0,90 (0,87,0,93) arterielle Q3 1,64 (1,57,1,70) 0,78 (0,75,0,81) Q4 1,93 (1,86,2,00) 0,65 (0,63,0,68) Q5 2,51 (2,42,2,60) 0,60 (0,58,0,62) Diabete Q2 1,30 (1,22,1,39) 0,93 (0,89,0,98) Q3 1,51 (1,42,1,61) 0,85 (0,81,0,89) Q4 1,85 (1,74,1,96) 0,72 (0,68,0,76) Q5 2,17 (2,05,2,29) 0,73 (0,69,0,77) Cardiopathie Q2 1,47 (1,38,1,58) 0,89 (0,85,0,94) Q3 1,87 (1,75,2,00) 0,80 (0,76,0,84) Q4 2,27 (2,13,2,42) 0,62 (0,59,0,66) Q5 3,12 (2,94,3,31) 0,55 (0,51,0,58) MPOC * (30 ans Q2 1,24 (1,08,1,43) 1,02 (0,92,1,13) et plus) Q3 1,35 (1,17,1,55) 1,02 (0,92,1,14) Q4 1,45 (1,27,1,66) 0,88 (0,78,0,99) Q5 1,71 (1,51,1,94) 0,72 (0,63,0,83) Emphyseme (30 Q2 1,34 (1,14,1,57) 0,88 (0,78,0,98) ans et plus) Q3 1,48 (1,26,1,73) 0,90 (0,80,1,01) Q4 1,76 (1,52,2,05) 0,79 (0,70,0,90) Q5 2,20 (1,91,2,53) 0,64 (0,55,0,75) Bronchite Q2 1,10 (1,01,1,21) 0,95 (0,88,1,02) chronique Q3 1,28 (1,17,1,40) 0,92 (0,85,0,99) Q4 1,42 (1,30,1,54) 0,83 (0,77,0,90) Q5 1,61 (1,49,1,75) 0,80 (0,74,0,88) Trouble de Q2 0,92 (0,88,0,97) 1,08 (1,03,1,13) l'humeur/anxiete Q3 0,95 (0,90,1,00) 1,15 (1,10,1,20) Q4 0,96 (0,92,1,01) 1,18 (1,12,1,23) Q5 0,97 (0,93,1,01) 1,09 (1,04,1,14) Etat de sante autodeclare/ autopercu Stress autopercu Q2 0,94 (0,91,0,97) 1,07 (1,03,1,10) (important/ Q3 0,89 (0,86,0,92) 1,18 (1,14,1,21) extreme) Q4 0,81 (0,79,0,84) 1,25 (1,21,1,29) Q5 0,73 (0,71,0,76) 1,31 (1,27,1,36) Sante autoevaluee Q2 1,17 (1,12,1,23) 0,92 (0,88,0,95) (mauvaise/ Q3 1,37 (1,31,1,44) 0,85 (0,81,0,88) moyenne) Q4 1,54 (1,47,1,61) 0,78 (0,75,0,82) Q5 1,95 (1,87,2,03) 0,83 (0,80,0,87) Sante mentale Q2 0,95 (0,89,1,01) 0,98 (0,93,1,04) autoevaluee Q3 0,99 (0,93,1,06) 1,01 (0,95,1,07) (mauvaise/ Q4 1,01 (0,95,1,07) 1,07 (1,01,1,14) moyenne) Q5 1,06 (0,99,1,12) 1,23 (1,15,1,30) Sentiment Q2 1,14 (1,10,1,19) 0,87 (0,84,0,90) d'appartenance Q3 1,21 (1,17,1,26) 0,80 (0,78,0,83) communautaire Q4 1,32 (1,27,1,37) 0,67 (0,64,0,69) (fort/assez Q5 1,48 (1,43,1,54) 0,60 (0,57,0,62) fort) N.B.: Le Q1 est la categorie de reference pour tous les resultats. * IMC = indice de masse corporelle; MPOC = maladie pulmonaire obstructive chronique.