Urbanisation and spatial inequalities in health in Brazil and India Tarani Chandola University of Manchester Sergio Bassanesi UFRGS - Universidade Federal Sitamma Mikkilineni Souvik Bandyopadhyay Anil Chandran Indian Institute of Public Health, Hyderabad Health is related to income differences within rich societies but not to those between them Between (rich) societies Within societies 80 Life expectancy (years) 79 78 77 76 75 74 73 72 as td ep riv e d 71 70 Le Most deprived Electoral wards in England & Wales ranked by deprivation score Source: Wilkinson & Pickett, The Spirit Level (2009) www.equalitytrust.org.uk Life expectancy and income inequality: Brazil, 2000 Plot showing the odds ratios (ORs) and 95% confidence interval (CI) for one-standard deviation change in Gini coefficient for the risk of being underweight, pre-overweight, overweight and obese. Subramanian S V et al. J Epidemiol Community Health 2007;61:802-809 ©2007 by BMJ Publishing Group Ltd Increasing income inequality in Brazil and India Increasing spatial inequality in poverty and income - urbanisation and concentration of economic activity - spatial concentration of affluence reproduces privileges of the rich - spatial concentration of poverty results in segregation, involuntary clustering in ghettos Effects on Individual and Population Health? “Triple health jeopardy: being poor in a poor neighbourhood that is spatially isolated from life-enhancing opportunities…” Nancy A Ross EVENNESS ISOLATION EXPOSURE CLUSTERING Dimensions of spatial segregation Sean F. Reardon & David O'Sullivan. “Measures of Spatial Segregation” p. 121-162, 2004 Sociological Methodology. V. 34, n.1, EXPOSURE/ISOLATION DIMENSION SPATIAL EXPOSURE INDEX Average proportion of group n in the localities of each member of group m SPATIAL ISOLATION INDEX Average proportion of group m in the local environments of each member of group m (spatial exposure of group m to itself) EVENNESS/ CLUSTERING DIMENSION SPATIAL NEIGHBOURHOOD SORTING INDEX Proportion of the variance between the different localities that contributes to the total variance of the variable X in the city GENERALIZED SPATIAL DISSIMILARITY INDEX Average difference of the population composition of the localities from the population composition of the urban area as a whole Key hypotheses: Districts, cities and states with less spatial socioeconomic inequalities have better population health than areas with greater spatial socioeconomic inequalities For a given level of income/socioeconomic position, people living in areas with less spatial socioeconomic inequalities have better health than those living in more segregated areas. Methods: Brazil Data (for the 25 largest cities): Demographic and Socioeconomic data: 2000 Census (census tract level) Mortality data: SIM Mortality Information System (district level data) India Data: Demographic and Socioeconomic data: 2001 census (sub-district Tehsil level) Mortality data: District Level Household and Facilities Survey 2002-04 and 2007-08 (Individual and district level) EVENNESS EXPOSURE ISOLATION CLUSTERING Dimensions of spatial segregation Spatial CLUSTERING Moran Scatter Plot Spatially lagged variable SLOPE OF THE REGRESSION LINE Variable to be lagged, standardized INDEX Moran Cluster Map Spatial CLUSTERING Within each district, the Spatial Clustering Index is the proportion of census tracts that are low income tracts and are surrounded by other low income tracts. INDEX EVENNESS ISOLATION EXPOSURE CLUSTERING Dimensions of spatial segregation Spatial Isolation Index Income >20 ms GLOBAL Ŏ>20=0.228 p<0.01 BW:400m LOCAL Local >20 ms 10-20 ms 2-5 ms 5-10 ms Spatial Isolation Indexes Income Groups BW:400m ms: minimum salaries <2ms INCOME Moran I Index: 0.65 ( ρ< 0.0001) Distribution of income of the head of the household by district, Porto Alegre, 2000. Source: IBGE AGE AND SEX ADJUSTED MORTALITY RATE Moran I Index: 0.34 ( ρ< 0.0001) Relative Index of Inequality: 1.8 Slope Index of Inequality: - 4.6 18 16 14 12 SALUD 10.0 10 8 6 5.4 4 2 0 0.0 0.2 0.4 0.6 RIDIT 0.8 1.0 Distribution of age and sex adjusted mortality rate by district, Porto Alegre, 2000. Source: DATASUS-SIM CARDIOVASCULAR DISEASES MORTALITY 45-64 YEARS CVD Deaths by 100,000 Moran I Index: 0.52 ( ρ< 0.0001) Distribution of age specific cardiovascular diseases mortality coefficient* , adjusted for age and sex, by district. Porto Alegre, 2000-2004. Sources: IBGE and SIM * results after smoothing Isolation indexes Simple Linear Regression Independent variables Dependent variables Standardized B coefficients and (R2) Total mortality Premature CV mortality External causes mortality Pulmonary tuberculosis incidence Without income 0.28* (0.08) 0.26 * (0.07) 0.35* (0.12) 0.45** (0.20) With income to < 2 ms 0.36 * (0.13) 0.37* (0.11) 0.42 ** (0.17) 0.52** (0.27) 2 to < 5 ms 0.19 (0.04) 0.18 (0.03) 0.22 (0.05) 0.30* (0.09) 5 to < 10 ms - 0.16 (0.03) - 0.19 (0.04) - 0.21 (0.04) - 0.13 (0.02) 10 to < 20 ms - 0.41** (0.17) - 0.44** (0.19) - 0.46** (0.21) - 0.37* (0.13) 20 or more ms - 0.53** (0.28) - 0.52** (0.27) - 0.53** (0.28) - 0.47** (0.22) Income groups Isolation indexes * Significant p<0.05 ** Significant p<0.001 ms: minimum salaries/month Band Width: 400 m Exposure indexes Simple Linear Regression Dependent variables Standardized B coefficients and (R2) Independent variables Income groups Exposure indexes Total mortality Premature CV mortality External causes mortality Pulmonary tuberculosis incidence >0 to <2 ms No income 0.31* (0.09) 0.29* (0.08) 0.38* (0.15) 0.49** (0.24) 2 to <5 ms < 2 ms 0.28* (0.08) 0.26* (0.07) 0.33* (0.11) 0.43** (0.19) 10 to <20 ms ≥ 20 ms - 0.52** (0.27) - 0.53** (0.28) - 0.54** (0.29) - 0.46** (0.21) 5 to <10 ms ≥ 10 ms - 0.41** (0.17) -0.44** (0.19) - 0.45** (0.21) - 0.36* (0.13) * Significant p≤0.05 ** Significant p ≤ 0.001 Band Width: 400 m Average proportion of group n in the localities of each member of group m Spearman Correlation Coefficient 2 to <5 ms < 2 ms -0.488** Tuberculosis Tuberculosis Spatial Exposure Index >0 to <2 ms No income -0.634** Tuberculosis 0.679** Tuberculosis 0.698** 10 to <20 MS ≥ 20 ms 5 to <10 ms ≥ 10 ms CLUSTERING INDEX Simple Linear Regression Independent variable Spatial CLUSTERING INDEX Standardized B R2 Dependent variables Total mortality Premature CV mortality External causes mortality Pulmonary tuberculosis incidence 0.65** 0.63** 0.64** 0.68** 0.42 0.39 0.41 0.46 Scattergram Clustering Index ** Significant p ≤ 0.01 Clustering Index Clustering Index Clustering Index Linear Regression Dependent variables Standardized B coefficients and R2 Independent variables Mean Income Clustering Index R2 Mean Income Isolation Index 10 or more ms R2 Mean Income Exposition Index 5 to <10 ms R2 * Significant p<0.05 ** Significant p<0.01 ≥ 10 ms Total mortality Premature CV mortality External causes mortality Pulmonary tuberculosis incidence - 0.40** - 0.30* - 0.31* - 0.33* 0.33* 0.39* 0.41** 0.42** 47.7 42.6 45.0 49,8 - 0.54** - 0.46** - 0.47** - 0.59** - 0.21 - 0.26* - 0.27* - 0.12 46.5 41.5 43.8 44.1 - 0.59** - 0.52** - 0.53** - 0.60** - 0.22* - 0.27* - 0.28** - 0.17 48.0 43.3 46.0 45,6 ms: minimum salaries/month Next steps: Brazil: Obtain and analyse data for other Brazilian cities India: Analyse DLHS-3 data in a multilevel and spatial context Workshops on Spatial and Multilevel Analysis: Brazil: May 18-20 2010 India: June 2-4 2010