The Moving Target: A Geographic Index of Relative Wellbeing

P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 C 2004) Journal of Medical Systems, Vol. 28, No. 4, August 2004 ( The Moving Target: A Geographic Index of Relative Wellbeing N LY Jochen Albrecht1,3 and Laxmi Ramasubramanian2 FR E A D IN G O Geographic Information Systems (GIS) have been widely used by health planners and professionals to map and describe disease occurrence, spread, and exposure. Increasingly, GIS is being used to measure accessibility to health services in order to better manage scarce resources and to ensure equity and accountability. We argue that health planners can use readily available census data to understand the demands and needs of particular population by identifying key indicators that have a direct or indirect impact on individual health and community well-being. We present an Index of Relative Wellbeing, a weighted basket of 10 key variables from the Census that can be used to describe the health status of a particular census area. Health planners can use this index within a GIS to conduct spatial and temporal analyses. Our research demonstrates that the spatial distribution of health inequalities can be carefully documented and be directly used in the policymaking arena. INTRODUCTION O O KEY WORDS: census; community wellbeing; GIS surveillance; indicator. FO R PR Geographic Information Systems (GIS) is a set of tools and procedures that collectively facilitate the analysis and modeling of spatially referenced data.(1) Ever since Dr. John Snow used simple spatial analysis techniques to describe a correlation between location of sources of drinking water and high incidences of cholera in London,(2) spatial analysis has been invaluable in the investigation of health-related phenomena.(3) GIS and spatial analysis have traditionally supported the work of public health professionals by mapping the location and spread of disease occurrence especially in identifying the pathways of disease diffusion.(4–6) In recent years, public health professionals have used GIS extensively to measure other health-related variables that are not directly linked to disease. For example, researchers have used spatial analysis tools to investigate evidence of disparities in 1 Department of Geography, 1151 LeFrak, University of Maryland at College Park, College Park, Mary land 20742. 2 Department of Urban Planning and Policy, University of Illinois at Chicago, Chicago, Illinois. 3 To whom correspondence should be addressed at e-mail: [email protected]. 369 C 2004 Plenum Publishing Corporation 0148-5598/04/0800-0369/0  P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 370 Style file version June 5th, 2002 Albrecht and Ramasubramanian access to health services(7,8) and inequities in public health spending and its impacts on overall health of the population(9,10) and to highlight how factors such as race, age, nutritional status, and socioeconomic status affect overall health and well-being. Investigating health problems at a variety of spatial scales is essential to develop appropriate health policy.(11,12) Health is a state of complete physical, mental, and social well-being and not merely the absence of disease or infirmity.(13) Furthermore, citizens, state and local governments, and international organizations like the World Health Organization and the United Nations now acknowledge that health is socially embedded. This shift in thinking brings into sharp focus the relationship between poverty and health, and particularly the role that governments can play in improving the health status of vulnerable populations. Although most health researchers and professionals acknowledge that socioeconomic status and poverty in particular, are critical determinants of health and help to identify disparities in health behaviors and outcomes,(14) there is considerable disagreement about the inclusion of specific poverty measures as part of a health indicator set. While health professionals often recommend inclusion of measures related to personal health behavior, access to services, social factors, environmental factors, and occupational issues to develop a comprehensive assessment of health status,(15) there is a recognition that documenting these measures is an expensive process that is difficult to conduct and almost impossible to repeat on a regular basis. In these situations the periodicity and timeliness of data availability are often compromised limiting their use for meaningful health decision making on an on-going basis. SCOPE OF PAPER In this context, the authors propose the use of a set of standard indicators (based on the national U.S. Census) that can provide a proxy measure of health and its relationship to socio-economic status. These indicators are organized into an index, the Index of Relative Wellbeing (hereafter referred to as the Wellbeing Index), which is a unique number from 0 to 9. The Wellbeing Index provides a reliable and easily recognizable measure of relative socio-economic status. The construction of the Wellbeing Index (which is influenced by, and based on the New Zealand Index of Relative Deprivation) is described in Section Three. This description identifies the census variables included in the composition of the index and the methodologies for weighting and computing the index. In Section Four, we describe the procedural steps involved in the construction of the index using Census 2000 data for the city of Los Angeles. This section also summarizes the analysis and discussion of the data generated. In Section Five, spatial analyses and maps generated using the Wellbeing Index are compared to maps created a more traditional measure of socioeconomic status (number of persons living below the poverty level in the same area) to discuss the advantages and limitations in using the Wellbeing Index. Finally, Section Six discusses the advantages of using GIS methodologies and small area statistics to conduct health surveillance, specifically using American Community Survey data and provides some directions for future research. P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 The Moving Target: A Geographic Index of Relative Wellbeing 371 THE INDEX OF RELATIVE WELLBEING Conceptual Basis Evidence from the United States and other OECD countries suggests that here is a strong correlation between socio-economic status and health status of a community.(16,17) Supported by the literature,(18) we propose that socioeconomic status is established through a complex process, where individuals are able to access material gains (income) because they have access to certain forms of cultural capital bestowed by education, familial and social networks, individual creativity and access to opportunity. The socioeconomic characteristics of a neighborhood or community play a significant role in establishing an individual’s status. For example, low-income individuals living in a relatively affluent neighborhood may find opportunities they would not otherwise have. They are likely to have access to material and cultural resources and are relatively better off than low-income individuals who live in a resourcepoor neighborhood. From a social capital perspective,(19–21) these opportunities are “bridging capital” (connections that occur between individuals with dissimilar backgrounds). Other researchers(22) have argued that even informal or “weak ties” among diverse individuals provide many opportunities for social networking and consequent access to new ideas, relationships, and opportunities. While internal social cohesion or “bonding capital” among members of a group that have shared norms and values is positive and central to creating a sense of community identity, the sometimes exclusionary nature of social cohesion can create marginalization and disempowerment. This is particularly true in neighborhoods and communities, where there is a great reliance on a traditional geographic or spatial sense of community. Therefore, we propose that the aggregated characteristics of the individuals in a neighborhood or community (area-based measures) become important in understanding how disparities in socioeconomic status of an area influence and impact individual health behaviors and outcomes. We propose that the use of a set of indicators gathered from data that is collected regularly and reliably for other purposes (such as the Census) can be used to create an index that integrates a range of socioeconomic variables to develop a proxy measure of public health status of a population. Simple mapping and analysis of demographic variables such as race, ethnicity, or age and correlations with a single measure of poverty (number of persons living below a particular income threshold) results in an area-based measure that provides only minimal understanding of socioeconomic need. However, the introduction of race and ethnicity often results in simplistic maps and spatial analyses that show that people of color are worse off than their white counterparts. Health advocates and indeed the individuals who live in “low-income” communities are intimately familiar with this observation. They hardly need a GIS-generated map to tell the story. Since many of our larger cities are highly segregated, race/ethnicity based analyses mask and overwhelm other variables that could potentially demonstrate the heterogeneity within segregated neighborhoods. Another limit of race/ethnicity based analyses is the propensity of inadvertent stigmatization through the assignment of P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 372 Albrecht and Ramasubramanian causal relationships between variables used in the analysis. Finally, race/ethnicity based analyses, while extremely useful for disease prevention and targeted health care provision, are not useful to describe and ameliorate the day to day sociospatial opportunities and constraints experienced by individuals. We believe that these opportunities or constraints have a direct or indirect bearing on their health status. Index Design and Selection of Variables In designing a comprehensive census-based measure of socioeconomic status, we excluded demographic variables (age, sex, race, and ethnicity) and focused on variables that are likely to provide an insight into material and social status of the population. Whereas demographic variables are usually immutable, variables associated with well-being are changeable, providing intervention opportunities. Our choices of variables are pertinent to determining individual and group socioeconomic status: income/poverty (two variables), disability, home ownership, overcrowding, educational attainment, parental status, transportation, and communication. A detailed description of the variables is provided in Table I. The variables provide an understanding of both material and social well being. Three of these variables (poverty, educational attainment, and disability) are considered to be indicators of health determinants and outcomes according to the Institute of Medicine’s Leading Health Indicators for Healthy People 2010 Report.(14) Except for disability, all other variables are included in the more established New Zealand Index of Relative Deprivation (NZDep) created in New Zealand as early as 1991.(23) Data Sources In the United States census, the data required to compute the Index is available in the Census Long Form (STF3), a sample data set based on questionnaires distributed to one in six households across the nation. Although it is a sample, it represents the most comprehensive data available to the public and can be analyzed at a variety of spatial scales. We elected to analyze the data at the scale of census block group since this type of analysis ought to be conducted at the neighborhood level and a block Table I. Census (SF3) Variables Used to Build the Wellbeing Index In Order of Their Weight Rank Variable SF3 # Weight (%) 1 People with any form of income support, incl. (a) social security, (b) supplemental security, and (c) public assistance income People with any form of disability People not living in own home People living in homes with too few bedrooms People over 18 without high school qualification People in single parent households People in households below poverty level People in households without car People unemployed (official 1999 figure) People in households without phone P62 P63 P64 P41 H15 H20 P37 P17 P89 H44 P43 H43 17.37 2 3 4 5 6 7 8 9 10 16.04 15.58 13.39 13.20 7.81 7.37 6.36 1.52 1.36 P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 The Moving Target: A Geographic Index of Relative Wellbeing Style file version June 5th, 2002 373 group is the smallest areal unit for which data for our variables are publicly available. The population size of a census block group varies in our case fairly evenly between 0 and 999 people. Choice of Variables While the variable definitions are provided in Table I, the rationale for using these variables is discussed in this section. Each variable is computed as a percentage based on data from the summary tape file 3 (STF3) data file.(24) Income Support Individuals receiving some kind of income support (social security, supplemental security income, public assistance) from a governmental agency are vulnerable. Typically, their incomes are stretched to the point where any sudden changes in the surrounding social and physical environment (however miniscule) can potentially have an adverse impact on their well being. Disability Approximately 20% of the American population (approximately 30% of our sample study area, Los Angeles) is affected by some of form of physical, cognitive/developmental, or emotional disability that limits their opportunities for employment. Health care provision needs and other service costs (transport, nutrition, job training, counseling) increase when there is a need to serve disabled populations. Homeownership Homeownership is an indicator of individual financial and economic stability. In addition, homeownership creates opportunities for individuals to build social support networks through participation in a variety of community development activities. Homeownership has a positive correlation with overall wellbeing.(25) Overcrowding From a public health perspective, overcrowding facilitates the spread of infectious diseases and creates unhealthy and untenable living conditions. High housing costs, accessibility factors (related to accessibility to services, educational institutions, care providers), and economic constraints can cause overcrowding.(26,27) Educational Attainment In a knowledge driven society, socioeconomic status is positively correlated with higher education. The absence of a high school diploma effectively shuts out many individuals from holding a meaningful job. With the decline of the manufacturing sector, traditional working class jobs that did not rely on education are fast disappearing, to be replaced by service sector jobs that often require specialized training and certifications (e.g., health care, personal care) beyond a high school diploma.(28) P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 374 Style file version June 5th, 2002 Albrecht and Ramasubramanian Parental Status Single parent status often imposes economic hardships. Children living in single parent households may be more vulnerable. Retraining single parents to join the workforce requires job training and access to other social services such as child care. Social stigma attached to single parent status and the demands of parenting often limit opportunities for informal social networking and active involvement in community-building activities. Poverty Status The relationship between poverty and ill-health is well-documented. Individuals living in households with incomes that are below the federal poverty level are disadvantaged because their nutritional, educational, and emotional needs are often adversely affected. Material deprivation contributes to social deprivation, collectively resulting in poor health outcomes.(29) Vehicle Ownership The spatial settlement patterns of most American cities and communities suggest that there is a spatial mismatch between where the jobs are located and where the people who seek those jobs actually live. There are several research studies that show that low-income workers living in and around central cities are unable to access the service sector jobs that are available in suburban communities. While employers, the government, and nongovernmental organizations are working to address this problem, in practical terms, individuals who have access to a car are more likely to find steady employment than those without.(30) Unemployment The official unemployment figure records individuals who are in the work force but have been unsuccessful in finding employment. Unemployment has practical and emotional health-related consequences. Unemployed individuals are more likely to neglect their own immediate health care needs and the needs of their family created pressing and urgent problems later on. In addition, unemployed individuals are likely to have physiological and emotional problems. They may adopt risky health behaviors that are sometimes triggered by stress. Telephone at Home Telephone penetration in the United States in very high, almost 98% in most regions of the country. However, access to a telephone at home sharply declines in low-income communities. Telephones enhance communication networks. Individuals accessing job leads, government services, or merely seeking to gather public information have to rely on the telephone. In future years, access to the Internet at home and cell phone usage may become more meaningful indicators to measure connectivity.(31) P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 The Moving Target: A Geographic Index of Relative Wellbeing 375 Computing the Index The Wellbeing Index is a relative measure of material and social characteristics that collectively confer socioeconomic status and opportunity. We started with computing percentage counts of each variable across for all 2560 census block groups that comprise the city of Los Angeles. We then used Principal Components Analysis (a data reduction technique for multivariate data) to establish linear correlations between successive combinations of variables to progressively account for variations in the data. The Relative Index of Wellbeing is the first Principal Component applied to each of the ten variables that results in an absolute score of wellbeing. All scores are then grouped into deciles, allowing for a relative measure that allows comparisons across the study area. The result is an ordinal scale or score from 1 to 10. A score of 1 represents the score for areas of greatest sense of wellbeing (least deprivation) and a score of 10 represents the score for areas with the lowest sense of wellbeing (highest deprivation). Ten percent of all census tracts in a particular region (study area) fall within one of the decile groups. For example, if a census tract receives a score of 1, it means that it is in the most privileged 10% of areas in the region. Advantages There are three main advantages in using the Wellbeing Index: First, the Index is simple to use. The scales from 1 to 10 mask the complexity of the underlying construction. The data can be represented graphically and can convey the relative social and economic opportunity structure of a particular region. Second, since race and ethnicity variables are not included in the construction of the Index, additional analyses comparing relationships between socio-econcmic wellbeing and race/ethnicity can be computed for different racial and ethnic categories. Finally, the data is available directly or can be computed from the Census questionnaires. Although this is sample data, it provides information at a small, stable spatial unit of analysis. We argue that the Wellbeing Index can be used for research funding allocation strategies, and for targeted health service provision at the census tract level. As a relative measure, the Wellbeing Index will provide researchers, public health professionals and decision makers a quick way to focus in on a problem area or a “hot spot” to conduct additional survey research. Researchers and practitioners could also focus in on a region with high wellbeing scores to further investigate the significance of correlation between socioeconomic status and positive health outcomes. The principal components analysis method allows researchers to identify the variables most likely to make a difference in the creation of social and economic opportunity, thereby streamlining ancillary service programs in education and social services. The last column in Table I lists the weight of each of the ten variables in its contribution to the wellbeing index for Los Angeles. It is worth pointing out that classic measures such as poverty status or unemployment rank low, while home ownership and disability rank surprisingly high (the data reflects relatively low levels of unemployment during the economic boom of the late 1990s). P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 376 Style file version June 5th, 2002 Albrecht and Ramasubramanian Limits and Caveats The indicator is a proxy measure to describe a complex set of phenomena. Because it is a simple measure, there may be a tendency to confuse the index with the phenomena it attempts to describe. Thus a community with a low rating score based on the Wellbeing Index should not be described as a deprived or marginalized community. The Wellbeing Index is constructed using data gathered at the individual level aggregated over an area (census block group). It would be an ecological fallacy to use the Index, in the way we have designed it, to rank individuals. The Scale of the Index (from 1 to 10) is an ordinal scale and not an interval scale. Thus, a Wellbeing score of 8 is not twice as good as a Wellbeing score of 4. At any given time, 10% of the population will live in an extremely deprived area (Wellbeing score of 10). Users of this Index will do well to remember that the Wellbeing Index is a relative index of wellbeing and therefore, some group of the population will always be significantly less well-off (materially and socially) than the rest of the population. COMPUTING THE WELLBEING INDEX FOR THE CITY OF LOS ANGELES Step 1: Identify Data Source We identified Census 2000 STF3 as the primary data source. Downloadable data is available from the official Census website at www.census.gov. Step 2: Determine the spatial unit of analysis We identified block groups as the elementary unit of analysis, over which a principal component analysis is then performed for all block groups within a census tract. Step 3: Calculate percentages The absolute numbers of the US Census tables are converted to percentages for each of the ten variables with respect to the population applicable. This is usually the number of people living in a census block group. For example, for the variable Parental Status, the calculated percentage is the number of people living in singleparent households expressed as a percentage of all people living in the census block group. Step 4: Standardization of percentages The percentage values for each of the 2484 block groups are then grouped into deciles, which allow an areal comparison of variables across categorical boundaries. Figure 1 shows thematic maps of the area east of Beverly Hills (wider downtown Los Angeles) of all ten variables. Step 5: Principal Component Analysis For each of the 864 census tracts that make up the city of Los Angeles, a principal component analysis is applied to the ten percentage variables in all block groups. The loading of the first principal component is then applied to the percentage values and the sum of the weights forms the absolute score of wellbeing in each block group. P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 The Moving Target: A Geographic Index of Relative Wellbeing Fig. 1. Mapping of the decile percentage ranges of 10 indicator variables for wellbeing. 377 P1: JQX Journal of Medical Systems [joms] 378 pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 Albrecht and Ramasubramanian A1 Fig. 2. Detailed (block group-level) map of the index of relative wellbeing. The no-data areas are either outside the very irregularly shaped boundary of the city of Los Angeles or represent downtown areas with resident population. Step 6: Standardization of PCA scores Similar to the procedure in step 4, all PCA scores are grouped into deciles, resulting in a relative score for each of the census block groups. An example for the resulting thematic map is Fig. 2. This map (using the same spatial extend as the ten small maps in Fig. 1) is a highly detailed account (high spatial resolution) of the relative wellbeing in one of the most notorious areas of Los Angeles. The high level of detail makes them very suitable for neighborhood-level analysis but overwhelming to communicate more general issues. Step 7: Spatial aggregation From a pure research perspective, aggregations should be avoided as they perturb the results. However, for overview purposes, the Wellbeing Index can be aggregated if each block group score is weighted by its population percentage within the census tract that it belongs to. Figure 3 shows such an aggregation of 2484 block groups into 864 census tracks. COMPARING THE RELATIVE WELLBEING INDEX WITH TRADITIONAL INDICATORS In the previous section, we described the procedural steps involved in the construction of the index using Census 2000 data for the city of Los Angeles. In the P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 The Moving Target: A Geographic Index of Relative Wellbeing 379 Fig. 3. Tract-level overview map of relative wellbeing in Los Angeles. following, spatial analyses and maps generated using the Wellbeing Index are compared to maps created a more traditional measure of socioeconomic status (number of persons living below the poverty level in the same area) to discuss the advantages and limitations in using the Wellbeing Index. All maps in this article use the same symbology, i.e., the darkest tones represent highest levels of overall wellbeing with the scale being uniform throughout the study area. This means that with respect to income support or disability for instance, there are very few high ranking areas in central Los Angeles, whereas with respect to homeownership or access to phone, large parts of central Los Angeles rank among the highest ten percent of the whole city. Some variables (such as educational attainment) show relative spatial homogeneity, while others (such as single parent families) have a high degree of negative spatial autocorrelation. Another illuminating form of analysis is to look at individual census block groups and to observe how they fare over the ten variables. Some neighborhoods (especially the hillside communities in the NW of each map in Fig. 1) are very consistent across the range of indicators, while others tend to change their ranking repeatedly. P1: JQX Journal of Medical Systems [joms] 380 pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 Albrecht and Ramasubramanian Observing the constantly high values for some of the communities surrounding Beverly Hills, it is at first sight surprising to see that they slip slightly in the overall wellbeing index (Fig. 2), while some other block groups in central Los Angeles move into the highest ranks. This is due to the ever changing weights that each variable has in its contribution to the overall index. The ranking of variables given in Table I is the overall average for the city of Los Angeles but the principal component loading varies sometimes extremely from one neighborhood to another. Figure 2 is a good example for the sometimes overwhelming character of such highly detailed maps. Although it is lost in the comparison of the small scale maps in Figure 1 with the large scale map in Fig. 2, the Wellness Index has actually a spatial smoothing effect. In spite of this, the high degree of variation in Fig. 2 makes it difficult to come to more general conclusions about wellbeing inequalities in the city of Los Angeles. Figure 3 is a population-weighted aggregation resulting in a map that matches the subjective perspective of scientists well acquainted with the social geography of Los Angeles. CONCLUSIONS AND DIRECTIONS FOR FUTURE RESEARCH While it is widely acknowledged that a single (census) variable such as poverty level is insufficient for determining the need for health services, there is no widely agreed upon standard as to what would be good indicators of health and wellbeing. Our research identifies one set of variables that help assess material and social factors that affect health and wellbeing. Although the authors consider crime to be an important determinant of wellbeing, it has not been included in these analyses. This is partly because of methodological difficulties in linking crime directly with the ten area-based measures of material and social deprivation. Similarly, our rendition of the overcrowding variable is a very simplistic one. The Canadian national occupancy standard(32) would be a preferable measure but it requires access to individual-level records that was not available to the authors. Our research uses statistical analysis techniques and the power of GIS technologies to demonstrate that the spatial distribution of health inequalities can be carefully documented and be directly used in the policymaking arena at the neighborhood/community level of analysis. Likewise, we believe that this research is a good example of the usefulness of easily accessible census data. The spatial resolution is more than sufficient for most health researchers; however the temporal grain leaves lots to be desired. Los Angeles is a prime example for a rapidly changing geography that is not well captured by the decennial census. The current endeavors of the US Census Bureau to popularize the American Community Survey with its adaptive survey periods promise rapid access to census data. The methods developed in this paper and the resulting Relative Index of Wellbeing can be computed by individuals or organizations with access to public information. The transparency and accessibility of our analyses strengthens other indicators-based research initiatives. P1: JQX Journal of Medical Systems [joms] pp1231-joms-488003 April 23, 2004 The Moving Target: A Geographic Index of Relative Wellbeing 19:39 Style file version June 5th, 2002 381 REFERENCES 1. Longley, P., Goodchild, M., Maguire, D., and Rhind, D., Geographic Information Systems (Vol. 1 and 2), Wiley, London, 1999. 2. Snow, J., On the mode of communication of cholera, 2nd edn, John Churchill, London, 1855. Available online from UCLA’s John Snow site at www.ph.ucla.edu/epi/snow/snowbook.html. Retrieved on November 7, 2003. 3. Croner, C., Public health, GIS, and the Internet. Annu. Rev. Public Health 24: 57–82, 2003. 4. Cliff, A. D., and Haggett, P., Disease Diffusion: The spread of epidemics as a spatial process. In Pacione, M. (ed.), Medical Geography: Progress and Prospect, Croom Helm, London, 1986. 5. Devesa, S. S., Grauman, D. J., Blot, W. J., Pennello, G., Hoover, R. N., and Fraumeni, J. F., Jr., Atlas of cancer mortality in the United States, 1950–94, Government Printing Office [NIH Publication No. NIH 99–4564], Washington, DC, 1999. Available online from National Cancer Institute’s website at: www3.cancer.gov/atlasplus/new.html. Retrieved on November 7, 2003. 6. Theophilides, C. N., Ahearn, S. C., Grady, S., and Merlino, M., Identifying West Nile virus risk areas: The Dynamic Continuous-Area Space–Time system. Am. J. Epidemiol. 157(9):843–854, 2003. 7. Krieger, N., Chen, J., Waterman, P., Rehkopf, D., and Subramanian, S. V., Race/ethnicity, gender, and monitoring socioeconomic gradients in health: A comparison of area-based socioeconomic measures—The Public Health Disparities Geocoding Project. Am. J. Public Health 93(10): pp. 1655–1671, 2003. 8. Rushton, G., Public health, GIS, and spatial analytic tools. Annu. Rev. Public Health. 24:43–56, 2003. 9. Mendez, C., Wallace, S., Yu, H., Meng, Y., Chia, J., and Brown, E. R., California’s new Assembly and Senate Districts: Geographic Disparities in Health Insurance Coverage, Policy Brief from UCLA’s Center for Health Policy Research, May 2003 (PB2003-3:1–7), 2003. Available online at: http://www.healthpolicy.ucla.edu/pubs/pubList.asp. Retrieved on November 7, 2003. 10. Juarez, P., Robinson, P., and Matthews-Juarez, P., 100% access, zero health disparities, and GIS: An improved methodology for designating health professions shortage areas. J. Health Soc. Policy 16(1–2):155–167, 2002. 11. Petit, K., Kingsley, T., Coulton, C., and Cigna, J., Neighborhoods and Health: Building Evidence of Local Policy, A report submitted to the Ofice of the Assistant Secretary of Planning and Evaluation, U.S. Department of Health and Human Services. Available online at: http://aspe.hhs.gov/hsp/neighborhoods-health03 Retrieved on November 7, 2003. 12. Sturm, R., Ringel, J., and Andreyeva, T., Geographic disparities in children’s mental health care Pediatrics. 112(4):e308–e308 2003. Available online from http://www.pediatrics.org 13. World Health Organization (WHO), Preamble to the Constitution of the World Health Organization As Adopted by the International Health Conference, New York, 19–22 June, 1946; signed on 22 July 1946 by the representatives of 61 States (Official Records of the World Health Organization, no. 2, p. 100) and entered into force on 7 April 1948. 14. Marmot, M., and Wilkinson, R. (eds.), Social Determinants of Health, Oxford University Press, London, 1999. 15. Chrvala, C., and Bulger, R. (eds.), Leading Health Indicators for Healthy People 2010: Final Report, Committee on Leading Health Indicators for Healthy People 2010, Division of Health Promotion and Disease Prevention, Institute of Medicine. National Academies Press, Washington, DC, 1999. Available online at: http://www.nap.edu/html/healthy3. Retrieved on November 7, 2003. 16. Kawachi, I., and Kennedy, B. P., The Health of Nations: Why Inequality is Harmful to Your Health. The New Press, New York, 2002. 17. Kawachi, I., and Berkman, L. F., Neighborhoods and Health, Oxford University Press, New York, 2003. 18. Bunker, J., Gomby, D., and Kehrer, B., Pathways to Health: The Role of Social Factors, The Henry Kaiser Family Foundation, Menlo Park, CA, 1989. 19. Putnam, R., Leonardi, R., and Nanetti, R., Making Democracy Work. Civic Traditions in Modern Italy, Princeton University Press, Princeton, NJ, 1993. 20. Woolcock, M., Social capital and economic development: Toward a theoretical synthesis and policy framework. Theory Soc. 27(2):151–208. 21. Putnam, R., Bowling Alone: The Collapse and Revival of American Community. Simon & Schuster, New York, 2000. 22. Granovetter, M., The strength of weak ties. Am. J. Sociol. 78(6): 1360–1380, 1973. 23. Crampton, P., Salmond, C., and Sutton, F., NZDep91: a new index of deprivation. Soc. Policy J. New Zealand. 9:186–193, 1997. P1: JQX Journal of Medical Systems [joms] 382 pp1231-joms-488003 April 23, 2004 19:39 Style file version June 5th, 2002 Albrecht and Ramasubramanian 24. U.S. Census Bureau., Summary Tape File 3. 2000 Census of Population and Housing. Technical Documentation, Electronic products development branch. U.S. Census Bureau, Washington, DC, 2003. 25. Rohe, W., Van Zandt, S., and McCarthy, G., The Social Benefits and Costs of Homeownership: A Critical Assessment of the Research, Low-income Homeownership Working Paper Series, Joint Center for Housing Studies. Harvard University, Cambridge, MA, 2001. 26. Baker, M., McNicholas, A., Garrett, N., Jones, N., Koberstein, V., and Lennon, D., Household crowding as a major risk factor for epidemic meningococcal disease in Auckland children. Pediatr. Infect Dis. J. 19(10):983–990, 2000. 27. Evans, G., and Kantrowitz, E., Socioeconomic status and health: The potential role of environmental risk exposure. Annu. Rev. Public Health 23:303–331, 2002. 28. Browning, Ch., Cagney, K., and Wen, M., Explaining variation in health status across space and time: Implications for racial and ethnic disparities in self-rated health. Soc. Sci. Med. 57:1221–1235, 2003. 29. Kawabata, M., Access to Jobs: Transportation Barriers Faced by Low-Skilled Autoless Workers in U.S. Metropolitan Areas, PhD thesis, Department of Urban Studies and Planning, Massachusetts Institute of Technology, Cambridge, MA, 2002. 30. Marmot, M., and Wilkinson, R. (ed.) The Social Determinants of Health, Oxford University Press, Oxford, UK, 1999. 31. O’Hare, W., Kids Count Snapshot, Annie E. Casey Foundation, Baltimore, MD, 2001. 32. Canada Mortgage and Housing Corporation (CMHC), Core Housing Need in Canada, Cat. No. NH 15-68/1991, Ottawa, Canada, 1991. View publication stats