U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) Committee on Guidance for Designing a National Healthcare Disparities Report; Swift EK, editor. Guidance for the National Healthcare Disparities Report. Washington (DC): National Academies Press (US); 2002.

Cover of Guidance for the National Healthcare Disparities Report

Guidance for the National Healthcare Disparities Report.

Show details


Thomas C.Ricketts, III

This paper examines how health status, access to health care, and health outcomes vary by geographic location. It also examines which aspects of location appear most to affect health care access, services, and utilization. There are clear geographic differences in health status that vary according to the level of aggregation. At the national level, overall mortality rates are much higher in the Southeast, the Appalachians, and parts of the Intermountain West (Pickle et al., 1996). That pattern changes for Black males to include very high rates in the urban East and Midwest. For White females higher rates cluster in the Midwest and Mississippi Valley. There are likewise differences among states that mirror regional patterns. Within states, differences are associated with areas with lower incomes, higher numbers of minority populations, and cultural and historical risk factor patterns that contribute to higher rates of morbidity and mortality. The same gradients can be seen within cities and counties where neighborhoods and census tracts reflect similar patterns of health disparities. These differences are both apparent and persistent when subjected to statistical controls and comparisons (Geronimus et al., 1999).


Geography is often thought of as the generation and interpretation of maps that describe the physical world. Geography is far more than that, but the physical description of boundaries has a great deal to do with how we view communities and how we construct society (Giddens, 1984). The physical aspects of a community are usually defined by boundaries that may have been developed for a specific public purpose, but that often create gradients that separate one population group from another. This can be apparent in zoning rules or in the creation of jurisdictions that have different systems of social support. Areas can also become different through social and economic processes that create regions or communities whose boundaries are essentially invisible.

When speaking of health, the domain of medical geography is most often invoked. Medical geography, however, is more aligned with the study of disease and disease diffusion without explicit consideration of other aspects of human interaction. The structure of health services and how people use health services in ways that reflect and create disparities are factors that span the human and the medical parts of geography. The discourse of the geographer involved in describing health care delivery and health status has become controversial within the discipline itself. While space and place in health care delivery are important, their structure and interpretation are, to some, irrelevant to practical decision making because they are the result of overwhelming social forces and power relationships. To others, a point of view that includes spatial and landscape perspectives can be useful for local purposes and for broader policy development (Mohan, 1998).

Nevertheless, the power of geographic comparisons and boundary setting is real in the policy world, and the application of policy is very sensitive to location and scale. As one leading researcher has observed:

There is no agreement about how to best define a geographical area in terms of socioeconomic position or about which area-based measures of socioeconomic position are most informative, especially across multiple kinds of health outcomes (Krieger, 2002).

This paper does not contradict that conclusion, but does recognize that there are options for understanding the geography of health disparities as well as for implementing solutions. For example, regions, states, and localities are different in several ways. Regional differences show that history, environment, culture, and politics have combined to create disparities that cross state boundaries. Those regional differences—apparent in the Southeast, Appalachia, portions of the Intermountain West, and selected parts of the Southwest-point to the need for interstate collaboration or federal coordination and sharing of resources. Urban-rural comparisons do not reveal consistent patterns of disparity, but rural and inner city conditions tend to magnify differences associated with other social, economic, and health system characteristics. There are variations in rates of illness and access to appropriate care that reveal themselves in comparisons across states. These consistent variations imply that there are state-level policy levers that can be used to reduce disparities. Town, city, and county boundaries may describe communities that can develop solutions using local government or social, religious, or external systems.

The small area geographies used most often to depict health status are appropriate for identifying and verifying health status disparities. They include units of census geography such as tracts and block groups, counties, ZIP code areas, and clusters of ZIP codes. These can be used to construct service or market areas that contrast health outcomes and utilization for primary care, general hospital care, and tertiary care. But the level of intervention appropriate to specific patterns of disparity is not always coterminous with those boundaries. While we may identify disparities in rates using ZIP code areas or census tracts, it is not easy to mobilize an intervention based on those boundary sets. People do not feel a sense of “membership” or citizenship to such areas, and neither government nor the health care system is structured to act at those levels unless the boundaries identify real neighborhoods or communities.

There is no consensus on a fundamental unit of geography to use in measuring health and health care in the United States or elsewhere. There are many reasons for this, including the problems of relating individual events to population rates. However, the most important reason lies in the way in which health data are reported (Meade and Earickson, 2000). Data are compiled according to the political and administrative organization of governments and, to a lesser extent, society. Denominators in rates are most often expressed as the population of some political unit such as a state. It would be more clinically useful to express rates in terms of gender, age, or even occupation. Those relate more directly to health care delivery, to health status, and to outcomes for individuals.


Interstate geography includes several different units commonly used to analyze health care delivery and service and to formulate health care policy. The following is a brief overview of two: regions and rural-urban areas.


Regional systems and structures have been developed to cope with health problems across state borders. They include the health care system development of the Tennessee Valley Authority (TVA) and the Appalachian Regional Commission (ARC) as well as work in the lower Mississippi Delta. The ARC remains active in this field and supports work that illustrates disparities in health status and access through the University of Kentucky (www.mc.uky.edu/RuralHealth /ARC_AHPAC/ahpac.htm). There are regional initiatives in the Mississippi Delta through various organizations and governments. A regional study of asthma supported by the Trust for America's Health is illustrative (health-track.org/reports/ms0420/). The Health Resources and Services Administration (HRSA) announced a program to improve health care by supporting rural hospitals in the Delta region in late 2001. Similar cross-state efforts such as the U.S.-Mexico Border Health Commission are underway along the U.S.-Mexico border (www.borderhealth.gov/). These regional initiatives are supported through affiliations of state governments such as the Southern Governors' Association or the Southern Growth Policies Board or ad hoc groups of governors or state agency heads.

Rural-Urban Areas

One view of the geographic structure of the nation contrasts how the population is distributed between cities and rural areas. There are more than 60 million people classified by the U.S. Bureau of the Census as “rural” and 55 million living in “nonmetropolitan” counties in 2000. This is a population group comparable in size to the United Kingdom. Rural America would be among the top 20 nations in population. The structure of the Congress, which gives equal representation to states in the Senate, means that the rural issues that are important in sparsely populated western states such as Idaho, Wyoming, Montana, and North and South Dakota are given careful consideration in Congress. The political as well as physical geography of the U.S. makes rurality an important concept.

The two most common designations of rurality used in describing populations are those of the U.S. Bureau of the Census and the U.S. Office of Management and Budget (OMB). “Urbanized areas” are defined by the U.S. Bureau of the Census according to a complex set of characteristics that takes into consideration the economic nature of a place, transportation patterns, and the number of people living in a fixed area. That definition is undergoing revision and a final rule is expected to be published soon. For the 2000 census, rural areas are considered places outside urbanized areas. Urbanized areas are composed of “core census block groups or blocks that have a population density of at least 1,000 people per square mile and surrounding census blocks that have an overall density of at least 500 people per square mile” (www.census.gov/geo/www/ua/ua_2k.html). This delineation has not been used often to determine effects on health and health care. More often the OMB Metropolitan-Nonmetropolitan classification of counties is used for comparisons.

The OMB designation classifies counties as metropolitan or nonmetropolitan based on whether the county has a large city and a number of suburbs. It also takes into account a functional element that measures the extent to which peripheral counties are economically integrated with their surrounding metropolitan counties. A Metropolitan Area (MA) must contain either a place with a population of at least 50,000, or a census-defined urbanized area and a total MA population of at least 100,000, or reflect the economic activities of such a place. Various attempts to subclassify the counties within the metropolitan and nonmetropolitan categories exist, and they have been used to examine health care resource use and distribution and health status. In 2001 the National Center for Health Statistics (NCHS) included a rural-urban comparison in its Healthy People series. The NCHS report found that:

  • Residents of counties on the borders of large metropolitan areas generally are ranked highest on health indicators.
  • Indicators of health, health care use, and health care resources can differ by level of urbanization.
  • Regions do vary, which is reconfirmed by data.
  • Nationally, residents of the most rural counties have the highest death rates for children and young adults, the highest death rates for unintentional and motor vehicle traffic-related injuries, and the highest mortality for ischemic heart disease and suicide among men (Eberhardt et al., 2001).

These general comparisons are plagued by the problem of aggregation of widely divergent nonmetropolitan populations and communities into large, gross classifications that are meant to be consistent across the nation. There are regional patterns of rural disadvantage that are highly discernible. For example, there is higher infant mortality in the rural Southeast. Those conditions are clearly related to the income and educational differences between those rural regions and other parts of the nation. Geographic patterns of morbidity and mortality vary by race and ethnicity (Albrecht et al., 1998), and these differences are sometimes reinforced by rural location. Blacks and Whites living in nonmetropolitan counties have higher death rates from diabetes (Ricketts, 2001) and heart disease (Slifkin et al., 2000).

The ecological interaction of income and health has been widely reported (Kawachi et al., 1997). A clear and consistent relationship exists between the two: the lower the income of the place, the worse the health status. The same has been found for the relationship of health to income inequality, but with less convincing evidence (Mellor and Milyo, 2001). However, when examining income inequality and health at the state level, one study found an interesting stronger relationship between inequality and self-reported health for nonmetropolitan residents (Blakely et al., 2002). That finding suggests that the structure of income inequality differs for rural areas, but it also might be an artifact of the clustering of respondents in nonmetropolitan counties.

Access to Care in Rural Areas

Access to health care services in rural versus urban areas has been explored by health services researchers for decades. Rural residents are, on average, poorer, older, and, for those under age 65, less likely to be insured than persons living in urban areas (American College of Physicians, 1995; Hartley et al., 1994; Braden and Beauregard, 1994; Schur and Franco, 1999). Rural Americans also report more chronic conditions and describe themselves in poorer health than urban residents. Further, injury-related mortality and the number of days of restricted activity are higher in nonmetropolitan areas. The degree to which lower levels of access affect health outcomes and utilization for rural persons is at issue, however, given the conclusions drawn by MedPAC in its Report to Congress (MedPAC, 2001). It is easy to challenge its flat assertion that an access gap does not exist. The analysis did not always include controls for health status, and the risk adjustment for prior use may have made the analyses inaccurate. The access study also did not differentiate between underserved and adequately served communities and did not reveal whether there was an independent rural or travel effect for the measures of access. But most importantly, the sample was drawn with the assumption that rural places compose a homogenous sample stratum. While the wide variation in access in urban systems is accepted and comparisons within and between metropolitan areas are usual in national surveys, this is not feasible for rural places given the current construction of these surveys (Schur et al., 1998).

Race, Ethnicity, and Rurality

The interaction of race and ethnicity and rurality has been examined in a review of studies of six conditions highlighted by the U.S. Department of Health and Human Services (DHHS) in its disparities initiative. The conditions are infant mortality, cancer screening and management, cardiovascular disease, diabetes, HIV infection, and child and adult immunizations (Slifkin et al., 2000). The review found that rural minorities are further disadvantaged compared to their urban counterparts in cancer screening and management, cardiovascular disease, and diabetes. The gaps between Whites and minorities appear to be greater for these conditions in rural places, but the studies that made up the review did not carefully control for many variables that might describe problems with access to care. Likewise, comparisons did not include controls for regional effects. There are clear limitations to drawing inferences from geographical classifications at the county level.

In sum, there is credible evidence that being in a rural place has a strong and relatively consistent negative effect on one's economic chances. However, there is some difficulty in creating a strong claim that rurality has an independent and significant impact on people's health. The problem, it seems, is that the definitions of what are rural and nonmetropolitan are more closely tied to factors related to population and its density. These have a consistent economic effect, but an inconsistent health effect. Unfortunately, a definition of medical rurality is not at hand. Instead, various measures of medical underservice, health professional shortages, and vulnerability are available. While those measures are place specific and tend to be more rural, they are also applicable to highly urbanized areas. The search for a perfect measure of rurality that will capture its health effects may be a useful exercise, but will require a careful analysis of the effects of distance, culture, occupational context, and the spatial characteristics of technology and information diffusion. Such a metric will have to overcome the strong bias in favor of existing, well-documented, and relatively consistent systems of classifications of rurality. To do so, it will have to have a transparent application to populations and health care systems as well as a clear application to policy.

Distance as a Proxy for Rurality

Distance to health care is one of the most important geographic features that may affect health status and health outcomes and that may contribute to disparities. The effects of distance on access to health care services have been a subject of research for some time. For example, Weiss examined how distance to a hospital combined with social class determines patterns of use (Weiss and Greenlick, 1970). Conner and colleagues examined studies of distance to care to attempt to find standards for access (Conner et al., 1994). While they found evidence of distance decay in use and some indication that quality of care suffered when care was provided to people who lived at some remove from services, they were unable to develop clear guidance for what would be a fair standard for physical accessibility. Nor were they able to develop clear guidance on how to measure it. They were able to contrast units of analysis classifying areas as “town/community/ZIP”; county; “market-share defined”; and national. However, they made no recommendations concerning their ability to detect differences that might reflect disparity. There is evidence that underserved populations are located at a greater physical distance from services in rural communities. Low-income populations in urban areas are often adjacent to a high density of health care resources (Bohland and Know, 1989).


There are several geographic units that are often used to analyze health care delivery and services and to develop health care policy. The following briefly examines states, communities, local health department jurisdictions, census and postal geography, and market areas.


In the U.S., states are the fundamental polities for the support and regulation of most local health care delivery. When the federal government chooses to provide support for nationwide public health programs, each of its three major options involves the states:

  • Grants-in-aid to states based on their populations, or so-called block grants;
  • Formula grants that take into consideration some factors of need, with the Medicaid program an example of such a system; and
  • Program or project grants that involve states either as umbrella applicants or as passive reviewers, with community health centers an example.

State public health systems and their vital and health statistics reporting systems provide much of the data on health care disparities. States have the primary responsibility for the protection of public health. As part of that responsibility, states have developed a coordinated system of data collection and reporting. They have also developed programmatic interventions that vary. State governments vary in the degree of support for public health and health care delivery, and there are differences in the structures of their health care delivery systems that are due to their respective populations, cultures, and histories. The states vary as markedly in investment in health as they do in health outcomes. Figure 5–1 describes a potentially close relationship between per capita health spending and years of potential life lost (YPLL).



Variations among States in Life Years Lost and Per Capita Spending for Health

Key to the identification of a substantial difference in health status or access between geographically defined populations or population segments is the degree to which the boundaries separate or include the population that is negatively affected or the degree to which the nature of the area itself affects health and health care. Maps of the United States at the state level show strong and important differences in mortality, morbidity, and access to care measures. There are different ranking and ratings systems that reveal health disparities at the state level, including those distributed by the UnitedHealth Group (UnitedHealth Group, 2000), Morgan Quitno (Morgan and Morgan, 2001), the National Conference of State Legislatures (Siegel, 1998), AARP (Lamphere et al., 1999), and the Urban Institute (Liska, et al., 1998). The National Center for Vital and Health Statistics of the Centers for Disease Control and Prevention (CDC) does not explicitly rank states, but data it distributes can easily be ranked and grouped. Those rating systems are criticized for their inaccuracy and the inclusion of subjective judgments of what constitutes a summary measure of health (Gerzoff and Williamson, 2001).

There are other compilations of state-level data that allow for comparisons, but that do not specifically rank or rate states. These include the Kaiser Family Foundation “50 State Comparisons” web site (www.statehealthfacts.kff.org), state-level data that are compiled by the Maternal and Child Health Bureau in the Health Resources and Services Administration (HRSA) to track Title V progress (www.mchdata.net/), and a series of health profiles for states compiled by HRSA that covers a wide range of indicators (stateprofiles.hrsa.gov/StateProfilesIndex.html). State agencies and the public pay close attention to these rankings systems, and they are sometimes used to guide policy decisions. The UnitedHealth Group rankings are circulated widely and commented upon regularly. The indicators used in that ranking system have been modified slightly for use as a performance measuring system for the state of Nebraska.

States have attempted to lead in the implementation of comprehensive programs to improve health status and the coordination of services either through overt political reform or through administrative emphasis on health (Nelson, 1994). The degree of variation in state efforts to improve population health is illustrated by the variation of their policies. For example, the Robert Wood Johnson Foundation's State Coverage Initiative and its tracking of insurance coverage by states illustrates the range of coverage decisions and the potential for state-level policy to influence how health care is paid for (www.statecoverage.net/matrix.htm).

The comparison of state-level data is important and reveals differences in health status and in overall measures of access and use of services. Comparisons of the use of certain therapeutic strategies for Medicare beneficiaries revealed patterns at the state level that could be interpreted generally across regions. Jencks et al. found that Medicare beneficiaries in less populous states and those in the Northeast were more likely to receive appropriate care as measured by 24 process indicators than those in more populous and southeastern states (Jencks et al., 2000). These patterns are illustrated in the work included in the Dartmouth Atlas of Health Care and its companion publications (Dartmouth Medical School, 1998).


Current policy emphasizes targeting “communities” for interventions to improve health and reduce disparities (Dorch et al., 1997). The Dictionary of Human Geography defines a community as “[a] social network of interacting individuals, usually concentrated into a defined territory. The term is widely used in a range of both academic and vernacular contexts generating a large number of separate (often implicit) definitions” (Johnston et al., 2000, p. 101). The Robert Wood Johnson Foundation recently commissioned papers to explore the appropriate geographic definition of a community that would allow the optimization of programs to improve population health. According to the conclusions reached by its contractors,

Community is a difficult concept to work with empirically and it has many, often overlapping, sometimes competing, definitions. Little consensus exists about boundaries or membership either in a general sense or in the context of measuring capacity for improving population health, or measuring a community's performance with regard to specific health status indicators. Race, income, sexual orientation, geography, and service areas, inter alia, are all viewed as valid parameters for defining a community (O'Keeffe et al., 2001, p. 23).

The relationship between socioeconomic characteristics and health in small areas has been described and validated in multiple studies at the census block group, census tract, and ZIP code levels (Krieger et al., 1997; Krieger, 1992; Kwok and Yankaskas, 2001). The field of “small area analysis” has amply demonstrated that variations can be found, but the determination of what are unacceptable variations remains open especially for the investigation of health services and access to health care (Stano, 1991; Diehr et al., 1990; Diehr et al., 1992).

Natural Communities and Social Networks

“Natural” communities or natural areas are described by the activities of people living in a named place or neighborhood. There are empirical techniques for identifying and summarizing natural areas in geography and sociology. The geographic relationship between the health care-seeking behavior of people and the spaces they use for work, shopping, and leisure have been described using maps that show areas of higher potential and actual use (Gesler and Meade, 1988). Natural communities might emerge from secondary analysis of rates that show contrasts. These could be developed and compared using techniques of geographical and sociological analysis. The development of a “landscape of disparities” may be more of a visualization exercise than an empirical problem, but there is some movement toward using Geographical Information Systems (GIS) to relate problem locations to populations and population activity to suggestions for solutions (Rushton et al., 2000).

Epling, Vandale, and Steuart describe extended family networks as perhaps the most appropriate denominator for epidemiological characterization of populations because this would allow for “more efficient units of diagnosis and therapy” (Epling et al., 1975, p. 87). In this case the denominators and numerators used to determine disparities in health would be developed on the basis of kinship and connection. They suggest that the validity of the construction of household networks can be determined by testing the hypothesis that there is greater similarity of health and disease episodes and behaviors within distinct social networks than between them.

Social networks and social support are understood to be important in determining health status (Weissbourd, 2000). But the only tractable way to understand these ties that bind seems to be through anthropological and ethnographical study that involves primary data collection. There may be proxy indicators for family and community cohesion that are reflected in church attendance and membership or participation in family-focused activities through employment, schools, or recreation. These proxy measures then become community indicators rather than measures of individual family unit cohesion and are reflected in the extant measure of social capital. It might be, however, that there are strong ties within families, but weak connections to other families. This might fit the characterization of a “clan” structure in the southern Appalachians or strong ethnic divisions in an urban neighborhood.

Identifying the “Healthy Community”

The characteristics of healthy communities have been described by organizations like the Healthy Communities movement associated with the Civic League. Healthy communities, according to Norris and Pittman, exhibit seven patterns that unite mind, spirit, and body (Norris and Pittman, 2000). A community that is healthy shapes its future; cultivates leadership everywhere; creates a sense of community; connects people and resources; knows itself; practices ongoing dialogue; and embraces diversity. These characteristics would appear to reduce disparities. However, they raise the question of whether communities that have these characteristics as well as differences in health outcomes by race or other population groups should be considered to have the same degree of disparity.

In the recent past, the idea that “social capital” contributes to the capacity of a community to improve health has been proposed. As an example of a social capital index, Joshua Galper describes an empirical approach to clustering and ranking counties on the basis of their social or civic capital (Galper, 1998). His indicators include the structure of the local economy as indicated by, for example, the number of large firms; the payroll of membership organizations; the number of museums, gardens, and zoos; crime and unemployment rates; educational levels; the age distribution; and newspaper readership, among other variables. This grouping and ranking system is similar to that used in the article, “How To Build Strong Home Towns” (Irwin et al., 1997). The Pew Charitable Trusts, the Population Association of America, and community and government agencies in Canada and Australia have also created approaches to measure social capital or community capacity (Pew Charitable Trusts, 1997; Teachman et al., 1997). These assessments of social capital have many of the same limiting characteristics that are encountered in community indicators of health care needs. They depend on fixed and often irrelevant units of analysis or denominators. These assessments are composed of indicators whose original purpose was to characterize some other element of the society or discrete activity. In addition, they are not very predictive of “outcomes,” whether they are measured in terms of health status or economic performance.

Local Health Department Jurisdictions

One likely focus for the implementation of health-enhancing and disparity-reducing policies on a geographic basis is the local health department. However, only half of the states currently have local health departments that are controlled by local government (Turnock, 2001). Fifteen states have centralized systems with control over local health units exercised by a state health agency, and remaining states have some form of mixed or shared control. The population coverage for local health departments may be very small and local: one quarter of health departments are responsible for 14,000 people or less. Health department districts or units represent the local presence of public health, and these units have a responsibility for monitoring health status. It is less clear that these districts are responsible for measuring their capacity for affecting health. However, there are currently energetic efforts on the part of the CDC to promote the evaluation and assessment of the performance of local health departments (Halverson et al., 1998; Halverson, 2000; Mays and Halverson, 2000). These assessment measures for public health may provide some input for “actionability” since the health department is often a key element in identifying local health priorities and developing programs.

Primary Care Service Areas

In the delivery of health services, there is a prevailing belief that the fundamental unit for constructing a rational health care delivery system is the primary care practice. In the Community-oriented Primary Care (COPC) paradigm, these areas often become coterminous with public health target areas. Primary care practices staffed by a generalist physician or other primary care practitioner are, under this regionalized scheme, appropriate caregivers for the small village or community of 1,000 or so people.

Primary care service areas have been developed in several states including Arizona, California, Maine, North Carolina, and Tennessee. They are used for the analysis of access to care or to create subcounty areas for designation for federal programs. These are clusters of ZIP codes (NC) and sub-county census geography (AZ, ME, CA). The system used in California is perhaps the oldest continuously used system and may present a template for other states to consider in developing a set of communities of solution for health services at a geographic level that is appropriate to local action (Smeloff and Kelzer, 1981). Whether these areas represent communities of solution for health improvement has not been addressed. But their use in California and North Carolina in the examination of preventable hospitalizations points to a broader set of causal factors for health beyond health care (Ricketts et al., 2001; Bindman et al., 1995).

Hospital Service Areas

The determination of medical service areas became an important part of health policy considerations in the 1980s due to the attention paid to legal and economic issues surrounding competition (Morrisey et al., 1988; Morrisey, 1993). Geographic methods for health care service area construction were the subject of a comprehensive review in the context of geography (Simpson et al., 1994).

There are three major types of methods for creating service areas: geographic distance, geopolitical areas, and patient origins. A distance approach would create radii or ellipses that surround a central place or limited numbers of nodes that represent core activities. This method is appropriate where a legislature or a regulating agency wishes to set a general standard for access. For example, “all enrollees must have a primary care clinic or office within 30 miles of their home.” These systems usually create a “crow-fly” or straight line standard, but occasionally travel time is used. Geopolitical boundaries are most commonly used to define health care service areas. This is largely due to their close links to policy-making bodies such as local and state governments that often operate health services or have public health responsibilities. The use of public funds is most often restricted to benefit-specific, pre-existing jurisdictions. Crossing those boundaries runs counter to the mutually exclusive nature of local government and its operations.

The use of patient origins to create service boundaries usually aggregates smaller geographic units such as ZIP codes or census tracts into areas using an inclusion rule based on proportion of total hospital admissions or hospitalizations from the small area. For the Dartmouth Atlas of Health Care, the Dartmouth Medical School team led by John Wennberg created an algorithm for the development of hospital service areas for the entire United States with the assistance of professional geographers (Dartmouth Medical School, 1998). There were 3,436 hospital service areas for the 4,900 general hospitals in the nation in the final service area map constructed for the Atlas. The Atlas and its derivative products are used for benchmarking many rates of treatment and resource allocation. The Atlas also provides data for the determination of comparative needs and points to important disparities in the health care system. The service areas that the Atlas uses may have the potential to serve as “communities of solution.” It should be noted that the authors of the Atlas have not made this proposal. Nonetheless, the ubiquity of the Atlas may create a perception that these areas can be used for these kinds of analyses as more and more policy makers refer to it and its structure.

Census and Postal Geography

Key to the collection of denominator statistics for local health measurement are the census geographies used to organize the extensive data collected regularly by the U.S. Bureau of the Census. Table 5–1 lists political, census, postal, and special geographies, all of which are used for statistical reporting.

TABLE 5–1. United States Political and Statistical Jurisdictions.


United States Political and Statistical Jurisdictions.

One common geographic unit is the Zone Improvement Plan Code, or ZIP code. ZIP codes are not always bounded areas. They are, by definition, a collection of postal addresses aggregated to improve mail delivery. A ZIP code may be assigned to a single building, a post office, or an institution. ZIP codes that cover a defined area may be interlaced as one delivery route passes and even crosses another, although that is rarely the case. ZIP code boundaries and route aggregations change continuously and do not require clearance at a central national level. They are reported in the publication ZIP ALERT, which is issued quarterly by the United States Postal Service1 (www.ribbs.usps.gov/files/zipalert/).

Market Areas

Markets are both observed, empirically derived assessments of human commercial behavior and conceptualizations of an intended consumption or activity pattern. While markets are most often associated with the buying and selling of goods and services in a commercial sense, markets can also be applied to activity spaces that describe general behavior. In the health sector, a hospital's market area may reflect where its patients come from, but also the people it reaches in information dissemination and prevention programs acting through intermediary agents.

Markets are defined at varying levels of geography:

  • for local goods and services. For example, these can take the form of a neighborhood bounded by streets or roads, collections of ZIP codes, or a city and its surrounding area.
  • for regional markets. These are usually described in terms of a set of counties or a region of a state or states. Examples include central Missouri and the Delmarva Peninsula.
  • for national and global markets.

There are theories or generalizations about markets and market areas that may apply to the questions at hand. Health as a function of lifestyle, diet, and exercise may be considered exclusively within an individual's control, but the ability to exercise and the diet choices available to a person are tied to his or her lived space. The forces that shape those choices are, in turn, influenced by national trends and policies. They are also influenced by the structure of health care delivery systems related to a higher order market befitting a complex, technology-associated service industry. Health promoting or shaping goods and services are usually “produced” in central places where local economies can support the people and systems necessary to produce those services and goods.

Even the development of data that might identify local disparities depends on geographically large market areas. Epidemiological and statistical analysis and interpretation is efficiently done for markets that are centered on the larger state health departments and research universities. Nationally, a market might be made up of perhaps 100 centers that “sell” or provide these services. The idea of devolving this process of statistical abstraction to localities may not adequately consider the realities of these market structures.

There are a number of potential general market-derived geographies that are candidates for assessment of disparities in health. These include Labor Market Areas (LMAs)2 and ZIP code clusters. Currently, there are 394 multi-county LMAs in the U.S. that are constructed from 741 multi-county “commuting zones,” which are defined using census data. Labor Market Areas are generally considered too large for meaningful local or community interventions.


The complex considerations associated with formulating and applying geographic units to health care policy necessarily involve important technical issues. The following presents an overview of data and localities, technical problems with community indicators, and GIS.

Data and Localities

In describing localities, data are often drawn from systems that use the county as the denominator for a population rate or the state as the sampling frame for a survey. The problems of applying data from multiple levels of aggregation to analyze conceptually coherent neighborhoods or communities in the U.S. have been described in several places (Diez-Roux, 1998; Duncan et al., 1998). The analytical difficulties inherent in this type of statistical work can create an “ecological fallacy,” which attributes collective characteristics to very dissimilar individuals. They can also reflect a lack of agreement on the power and specificity of multi-level modeling.

The geographic unit of analysis is often key to the ability of a measure to be sensitive to the underlying construct or local characteristic that is being measured. In a review of studies of geographic access to health care in rural areas, Connor and colleagues described studies that used “town/community/ZIP code areas,” counties, “market share defined areas,” and “other areas,” which were usually aggregates of ZIP codes or clusters of counties (Conner et al., 1994). They were seeking guidance on the appropriate unit of analysis for assessments of the adequacy of access and guidance for allocating resources. The review did not support the idea of access as a unifying concept that would lead to a consensus definition of an appropriate geographic unit. The general geographic size of places where access was most effectively measured was at the local level, usually consisting of small counties or clusters of ZIP areas. It was closely associated with the system that was meant to affect or provide access to primary care. In these areas, the fit between a measurable disparity in access closely approximated the area in which a solution could be achieved either through the enhancement of availability (for example, creating a clinic) or modifying some factor that reduced access (for example, developing a subsidy for care). However, many of the studies they reviewed made note of, but seldom measured, important effects and influences on the programs and projects from adjacent areas or state systems.

Technical Problems with Community Indicators

The determination of small area rates and indices describing the health status and health care resources available to populations is subject to varying degrees of error. In creating these rates and indicators, analysts rely on a largely dispersed and cooperative system of reporting that is based on local and state rules and laws, although the standards and guidelines are centrally agreed upon. Mortality rates, overall, are generally considered accurate, but there is evidence that cause of death is often miscoded on death certificates that are the source of mortality data (Kircher, 1985; Goodman and Berkelman, 1987). The accuracy of health care resource data is not often called into question, but for secondary data analysis there are problems with national data sources that may skew a picture of a county or community. The American Medical Association (AMA) Masterfile is the most frequently used source for national estimates of physician supply down to the county level, but it has been shown to have a degree of error due to reporting lags and the high mobility of physicians (Cherkin and Lawrence, 1977; Grumbach et al., 1995; Williams et al., 1996). For rural areas, the difference between the number of physicians reported in the Masterfile and the actual, locally verified number is striking in many places (Konrad et al., 2000; Ricketts et al., 2000). At the state level, license and survey data indicate that the Masterfile may overestimate primary care physician supply by as much as 20 percent. Data for nurses, pharmacists, and other health professionals are far less accurate when drawn from national sources because of the lack of a national inventory system (Kresiberg et al., 1976; Osterweis et al., 1996).

Geographic Information Systems as Savior?

GIS has been proposed by some as an all-purpose answer to problems of community characterization. It is touted as capable of solving resource allocation problems as well as of being an essential part of the field epidemiologist's armamentarium. The widespread use of GIS in public health came relatively late in the development of computer-assisted cartography and geographic analysis largely due to the lack of useful data to attach to geographic coordinates (Rushton et al., 2000).3 Healthy People 2010 includes the goal of increasing “the proportion of all major national, State, and local health data systems that use geocoding to promote nationwide use of geographic information systems (GIS) at all levels” from a baseline of 45 percent to 90 percent (Office of Disease Prevention and Health Promotion, 2001, pp. B23–4).

Geographic information systems carry the strong promise of a new, liberating technology and are often advertised to have the capacity to allow complex information to be displayed clearly and transparently, making both problems and solutions apparent. However, GIS is not really a new technology, but an expansion and intensification of older technologies. The expansion of the use and capacity of computers has facilitated collection of data by using remote sensing or by tapping into administrative, statistical, or clinical datasets. However, the massive amount of data that is now available has not immediately led to marked improvements in health care, the identification of health problems, or the formulation of health solutions because the volume of data has outpaced our ability to understand it.

GIS, however, has renewed interest in the use of spatial data as well as of statistical data of all types to explore questions, and to conduct surveillance of health systems and communities. The power of a map or data displayed in reference to space cannot be underestimated. The ability to quickly depict data in maps and graphs using GIS has made many problems seem more tractable because they can be understood in a context that is shared by analysts, policy makers, and stakeholders. At the same time, the classical errors of the mapmaker are repeated, and the ability to “lie with maps” is increasingly recognized as a threat to the validity of analysis on the order of more standard statistical misapplications (Monmonier, 1991).


Geographers who examine the relationship between place and health believe that it is formed less by the intrinsic nature of fixed places than by how people interact across space to make a particular place more or less healthy. The relationship between HIV infection and interstate highway locations represents a perfect example of a health consequence that is literally in motion and dependent upon place only to facilitate transmission. The consequences are felt at a distance. Injury prevalence is dependent on risks that are tied to geography: higher rates of trauma in rural areas are due to factors related to exposure and behavior (snowmobile use, chainsaws, tractors, higher highway speeds, lower seatbelt use) that reflect the interaction between human activity and space and places. These are disparities in risks are related to geography. Paradoxically, urban places tend to be a bit safer in terms of trauma. There are more guns in rural places, and firearm injury rates are higher. Also, the urban-rural differential in drug and substance abuse is no longer so great as to create clear contrasts in the net health effects of crime. There are obvious structural and physical differences between the decaying inner city of Scranton, Pennsylvania and of the “cotton trail” area of South Carolina. However, the health disparities in access, services, and quality are fundamentally the same and described in the same terms. Across geographies there is a convergence of human health status and of how we deal with it.

While geographic location is associated with wide variations in access, health care use, and health status, two core geographic elements and their relationship to health disparities are not well understood. They are distance (time and topography fit under this heading as well) and weather. Measuring distance often involves the use of rough estimations that mask actual geographic patterns of use. In many studies of the effects of distance, populations are described by some geographic entity such as a ZIP code or county, and the “average” distance to some location of care using the center of the geographic unit is calculated. This means that the variation or disparity due to differences in distance that exist within this geographic unit is lost to the analysis. The option is then to examine the relationship between an individual's distance to care and health status or outcomes. The latter analytical approach is feasible, but the former is far less expensive. Much of what we know about the effects of distance on health is based on the former type of studies. The degree to which true effects of distance are missed by this ecological approach is not well understood. Similarly, analysts and researchers often ignore differences in weather and environment and their effects on access, especially in the United States. We are constrained by our boundaries in such a way that we may not be able to completely understand how geography does affect disparities.

The relationship of neighborhood residence to health may be considered a form of pure geographic effect since neighborhoods are a combination of topography and social interaction. However, a reliable definition of neighborhood is elusive, and bringing some form of consistency to its measurement may be antithetical to a concept that strives to reflect the variety of human interaction. Measuring true geographic disparity has been difficult, and summary approaches that compare populations often mask evidence of disparity. We may have to begin to think of geography in the study of health disparities as more of an individual characteristic as opposed to a way to organize population analysis.

Reference List

  • Albrecht, S.L., L.L.Clarke, and M.K.Miller. 1998. Community, family, and race/ethnic differences in health status in rural areas. Rural Sociology 63 (2):235–52.
  • American College of Physicians. 1995. Rural primary care. Annals of Internal Medicine 122 (5):380–90. [PubMed: 7847653]
  • Bindman, A.B., K.Grumbach, D.Osmond, M.Komaromy, K.Vranzian, N. Lurie, J.Billings, and A.Stewart. 1995. Preventable hospitalizations and access to health care. JAMA 264 (4):305–11. [PubMed: 7609259]
  • Blakely, T.A., K.Lochner, and I.Kawachi. 2002. Metropolitan area income inequality and self-rated health—a multi-level study. Social Science & Medicine 54 (1):65–77. [PubMed: 11820682]
  • Bohland, J., and P.L.Know. 1989. Growth of proprietary hospitals in the United States: a historical geographic perspective. In Health Services Privatization in Industrial Societies , ed. J.L.Scarpaci, editor. . New Brunswick, NJ: Rutgers University Press.
  • Braden, J.J., and K.Beauregard. 1994. Health Status and Access to Care of Rural and Urban Populations. Rockville, MD: Agency for Health Care Policy and Research.
  • Brasure, M., S.C.Stearns, B.C.Norton, and T.C.Ricketts. 1999. Competitive behavior in local physician markets. Medical Care Research and Review 56 (4):395–414. [PubMed: 10589201]
  • Cherkin, D., and D.Lawrence. 1977. An Evaluation of the American Medical Association's Physician Masterfile as a Data Source—One State's Experience. Medical Care 15 (9):767–69. [PubMed: 578284]
  • Conner, R.A., J.E.Kralewski, and S.D.Hillson. 1994. Measuring geographic access to health care in rural areas. Medical Care Review 51 (3):337–77. [PubMed: 10138051]
  • Conover, C.J. 1998. “Do States Matter?” Online. Available at http://www​.hpolicy.duke​.edu/cyberexchange​/StateGov/slides/st1–2d.html [accessed May 30, 2002].
  • Dartmouth Medical School. 1998. The Dartmouth Atlas of Health Care 1998 . Ed. J.E.Wennberg, editor. . Chicago: American Hospital Publishing, Inc.
  • Diehr, P., K.C.Cain, W.Kreuter, and S.Rosenkranz. 1992. Can small-area analysis detect variation in surgery rates? The power of small-area variation analysis. Medical Care 30 (6):484–502. [PubMed: 1593915]
  • Diehr, P., K.Cain, F.Connell, and E.Volinn. 1990. What is too much variation? The null hypothesis in small-area analysis. Health Services Research 24 (6):741–71. [PMC free article: PMC1065599] [PubMed: 2312306]
  • Diez-Roux, A.V. 1998. Bringing context back into epidemiology: variables and fallacies in multi-level analysis. American Journal of Public Health 88 (2):216–22. [PMC free article: PMC1508189] [PubMed: 9491010]
  • Dorch, J.S., editor; , L.A.Bailey, editor; , and M.Stoto, editor. , eds. 1997. Improving Health in the Community. A Role for Performance Monitoring . Washington, DC: National Academy Press. [PubMed: 25121202]
  • Duncan, C., K.Jones, and G.Moon. 1998. Context, composition, and heterogeneity: using multi-level models in health research. Soc Sci and Medicine 46 (1):97–117. [PubMed: 9464672]
  • Eberhardt, M.S., D.D.Imgram, D.M.Macuc, E.R.Pamuk, V.M.Freid, S.B. Harper, C.A.Schoenorn, and H.Xia. 2001. Urban and Rural Chartbook . Health United States, 2001. Washington DC: U.S. Government Printing Office.
  • Epling, P.J., S.E.Vandale, and G.W.Steuart. 1975. Beyond the individual for the practice of social medicine: household networks as etiologic-diagnostic units. In Topics and Utopias in Health. Policy Studies , eds. S.R.Ingman, editor; and A.E.Thomas, editor. . The Hague: Mouton.
  • Galper, J. 1998. An exploration of social capital, giving, and volunteering at the United States county level. Washington, DC: The Urban Institute.
  • Geronimus, A.T., J.Bound, and T.A.Waidmann. 1999. Poverty, time, and place: variation in excess mortality across selected US populations, 1980– 1990. Journal of Epidemiology and Community Health 53 (6):325–34. [PMC free article: PMC1756883] [PubMed: 10396478]
  • Gerzoff, R.B., and G.D.Williamson. 2001. Who's number one? The impact of variability on rankings based on public health indicators. Public Health Reports 116 (2):158–64. [PMC free article: PMC1497305] [PubMed: 11847301]
  • Gesler, W.M., and M.S.Meade. 1988. Locational and population factors inhealth care-seeking behavior in Savannah, Georgia. Health Services Research 23 (3):443–62. [PMC free article: PMC1065514] [PubMed: 3403277]
  • Giddens, A. 1984. The Constitution of Society . First paperback edition ed. Berkeley, CA: University of California Press.
  • Goodman, R.A. and R.L.Berkelman. 1987. Physicians, vital statistics, and disease reporting. JAMA 258 (3):379–81. [PubMed: 3599334]
  • Grumbach, K., S.H.Becker, E.H.Osborn, and A.B.Bindman. 1995. The challenge of defining and counting generalist physicians: an analysis of Physician Masterfile data. American Journal of Public Health 85 (10):1402–07. [PMC free article: PMC1615636] [PubMed: 7573625]
  • Halverson, P.K. 2000. Performance measurement and performance standards: old wine in new bottles. Journal of Public Health Management and Practice 6 (5):vi–x. [PubMed: 11067655]
  • Halverson, P.K., R.M.Nicola, and E.L.Baker. 1998. Performance measurement and accreditation of public health organizations: a call to action. Journal of Public Health Management and Practice 4 (4):5–7. [PubMed: 10186760]
  • Hartley, D.L., L.Quam, and N.Lurie. 1994. Urban and rural differences in health insurance and access to care. Jrnl Rur Health 7 (4):357–79.
  • Irwin, M., C.Tolbert, and T.Lyson. 1997. How to build strong home towns. American Demographics 19 (February):42–48.
  • Jencks, S.F., T.Cuerdon, D.R.Burwen, B.Fleming, P.M.Houck, A.E. Kussmaul, D.S.Nilasena, D.L.Ordin, and D.R.Arday. 2000. Quality of medical care delivered to Medicare beneficiaries: A profile at state and national levels. JAMA 284 (13):1670–76. [PubMed: 11015797]
  • Johnston, R.J., editor; , D.Gregory, editor; , G.Pratt, editor; , and M.Watts, editor. , ed. 2000. The Dictionary of Human Geography . 4th ed. Oxford: Blackwell Publishers Ltd.
  • Kawachi, I, B.P.Kennedy, K.Lochner, and D.Prothrow-Stith. 1997. Social capital, income inequality, and mortality. American Journal of Public Health 87 (9):1491–98. [PMC free article: PMC1380975] [PubMed: 9314802]
  • Kircher, T., J.Nelson, and H.Burdo. 1985. The autopsy as a measure ofaccuracy of the death certificate. N Engl J Med 313 (20):1263–69. [PubMed: 4058507]
  • Konrad, T.R., R.Slifkin, C.Stevens, and J.Miller. 2000. Using the AMA Physician Masterfile to measure physician supply in small towns. Journal of Rural Health 16 (2):162–67. [PubMed: 10981368]
  • Kresiberg, H.M., J.Wu, E.Hollander, and J.Bow. 1976. Methodological approaches for determining health manpower supply and requirements. Volume I. Analytical Perspective. In Health Planning Methods and Technology Series . Rockville, MD: National Health Planning Information Center.
  • Krieger, N. 1992. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. American Journal of Public Health 82 (5):703–10. [PMC free article: PMC1694121] [PubMed: 1566949]
  • Krieger, N. 2002. Researcher tries to accurately measure socioeconomic differences and health disparities . Harvard School of Public Health, January 12, 2001. Available at www​.hsph.harvard.edu/ats/Jan12/jan12_03​.html [accessed February 21 2002].
  • Krieger, N., D.R.Williams, and N.E.Moss. 1997. Measuring social class in U.S. public health research: concepts, methodologies and guidelines. Annual Review of Public Health 18:342–78. [PubMed: 9143723]
  • Kwok, R.K., and B.C.Yankaskas. 2001. The use of census data fordetermining race and education as SES indicators: a validation study. Annals of Epidemiology 11 (3):171–77. [PubMed: 11293403]
  • Lamphere, J., N.Brangan, S.Bee, and K.Griffin. 1999. Reforming the Health Care System: State Profiles 1999 . Washington, DC: AARP Public Policy Institute.
  • Liska, D.W., N.J.Brennan, and B.K.Bruen. 1998. State-Level Databook on Health Care Access and Financing . Washington, DC: The Urban Institute Press.
  • Mays, G P., and P.K.Halverson. 2000. Conceptual and methodological issues in public health performance measurement: results from a computer-assisted expert panel process. Journal of Public Health Management and Practice 6 (5):59–65. [PubMed: 11067662]
  • Meade, M.S., and R.J.Earickson. 2000. Medical Geography . 2nd ed. New York: The Guilford Press.
  • MedPAC. 2001. Report to the Congress: Medicare in Rural America. Washington, DC: Medicare Payment Advisory Commission.
  • Mellor, J.M., and J.Milyo. 2001. Re-examining the evidence of an ecological association between income inequality and health. Journal of Health Politics, Policy and Law 26 (3):487–522. [PubMed: 11430248]
  • Mohan, J.F. 1998. Explaining geographies of health care: a critique. Health & Place 4 (2):113–24. [PubMed: 10671016]
  • Monmonier, M. 1991. How to Lie with Maps . Chicago: University of Chicago Press.
  • Morgan, K.O'L., editor; , and S.Morgan, editor. , eds. 2001. Health Care State Rankings, 2001 . Lawrence, KS: Morgan Quitno Press.
  • Morrisey, M.A. 1993. On defining small areas. Medical Care 31 (5 Suppl):YS89–95. [PubMed: 8492594]
  • Morrisey, M.A., F.A.Sloan, and J.Valvona. 1988. Defining geographic markets for hospital care. Law and Contemporary Problems 51 (2):165– 94. [PubMed: 10295962]
  • Nelson, H.. 1994. The states that could not wait: lessons for health reform from Florida, Hawaii, Minnesota, Oregon, and Vermont. In Five States That Could Not Wait , eds. D.M.Fox, editor; and J.K.Iglehart, editor. . Cambridge, MA: Milbank Memorial Fund.
  • Norris, T., and M.Pittman. 2000. The healthy communities movement and the coalition for healthier cities and communities. Public Health Rep 115(2–3) :118–24. [PMC free article: PMC1308699] [PubMed: 10968742]
  • O'Keeffe, J., K.N.Lohr, and E.R.Brody. 2001. Community Capacity for Improving Population Health: Feasibility and Design Study. Research Triangle Park, NC: Research Triangle Institute.
  • Office of Disease Prevention and Health Promotion. 2002. Healthy People 2010 . US Department of Health and Human Services 2001. Available at http://www.health.gov/healthypeople/document/HTML/Volume2/23PHI. htm [accessed February 21 2002].
  • Osterweis, M., C.J.McLaughlin, H.R.Manasse, and C.L.Hopper. 1996. The U.S. Health Workforce: Power, Politics and Policy . Washington, DC: Association of Academic Health Centers.
  • Pew Charitable Trusts. 1997. Trust and Citizen Engagement in Metropolitan Philadelphia: A Case Study. Philadelphia: Pew Research Center for the People & The Press.
  • Pickle, L.Williams, M.Mungiole, G.K.Jones, and A.A.White. 1996. Atlas of United States Mortality . Hyattsville, MD: U.S. Department of Health and Human Services.
  • Ricketts, T.C. 2002. Toward a new designation of underservice: final report of proposed method, analysis of effects and commentary on related areas. Chapel Hill, NC: Cecil G.Sheps Center for Health Services Research, University of North Carolina at Chapel Hill.
  • ——. 2001. Community Capacity to Improve Population Health: Defining Community. Research Triangle Park: Research Triangle Institute.
  • Ricketts, T.C., L.G.Hart, and M.Pirani. 2000. How many rural physicians are there? Journal of Rural Health 16 (3):198–207. [PubMed: 11131758]
  • Rushton, G., G.Elmes, and R.McMaster. 2000. Considerations for improving geographic information research in public health. URISA Journal 12 (2):31–49.
  • Schur, C.L., and S.J.Franco. 1999. Access to health care. In Rural Health in the United States , ed. T.C.Ricketts, editor. . New York: Oxford University Press.
  • Schur, C.L., C.D.Good, and M.L.Berk. 1998. Barriers to Using National Surveys for Understanding Rural Health Policy Issues. Bethesda, MD: Project HOPE Walsh Center for Rural Health Analysis.
  • Siegel, S. 1998. Access to Primary Health Care: Tracking the States . Washington, DC: National Conference of State Legislatures.
  • Simpson, K.N., S.DesHarnais, A.Jacobs, and A.Menapace. 1994. Methods for defining medical service areas. In Geographic Methods for Health Services Research , eds. T.C.Ricketts, editor; , L.A.Savitz, editor; , W.Gesler, editor; and D. Osborne, editor. . Lanham, MD: University Press of America.
  • Slifkin, R.T., L.J.Goldsmith, and T.C.Ricketts. 2000. Race and Place: Urban-Rural Differences in Health for Racial and Ethnic Minorities. Chapel Hill, NC: Rural Health Research and Policy Analysis Program, University of North Carolina at Chapel Hill.
  • Smeloff, E., and B.W.Kelzer. 1981. A geographic framework for coordination of needs assessment for primary medical care in California. Public Health Reports 96 (4):310–14. [PMC free article: PMC1424228] [PubMed: 6894796]
  • Stano, M. 1991. Further issues in small area variations analysis. Journal of Health Politics, Policy and Law 16 (3):573–88. [PubMed: 1761828]
  • Teachman, J.D., K.Carver, and K.Paasch. 1997. Social capital and the generation of human capital. Department of Human Development, Washington State University.
  • Turnock, B.J. 2001. Public Health: What It Is and How It Works . 2nd ed. Gaithersberg, MD: Aspen Publishers, Inc.
  • U.S. Bureau of the Census. 2002. Geographic Areas Reference Manual [Accessed March 20, 2002]. Available from http://www​.census.gov:80/geo/www/garm​.html.
  • UnitedHealth Group. 2000. The UnitedHealth Group State Health Rankings, 2000 Edition . St. Paul, MN: UnitedHealth Group.
  • Weiss, J.E., and M.B.Greenlick. 1970. Determinants of medical care utilization: the effects of social class and distance on contacts with the medical care system. Medical Care 8 (6):456–62. [PubMed: 5489515]
  • Weissbourd, B. 2000. Supportive communities for children and familes. Public Health Reports 115 (2):167–73. [PMC free article: PMC1308707] [PubMed: 10968750]
  • Williams, P.T., M.Whitcomb, and J.Kessler. 1996. Quality of the family physician component of AMA Masterfile. Journal of Am Board of Fam Pract 9 (2):94–99. [PubMed: 8659271]



The Census Bureau maintains a Master Area Block Level Equivalency file (MABLE) that crosslists ZIP codes with the census boundary files. Using that crosswalk, the census reports data at the ZIP code level on “Summary Tape File-3” (STF-3). The ZIP codes included on that file are modified in that they are the ZIP codes that have some boundary characteristics. They include within those boundaries the ZIP codes that are assigned, for example, to a post office and its related boxes or to a “point” ZIP that is a building or institution.


LMAs are formally described using county-level data and are based on a clustering algorithm that makes use of county-to-county commuting flows that are part of the census data collection process. The basic clusters of counties that are used to develop labor market areas are called commuting zones (CZs). In 1990, 741 commuting zones were delineated for all U.S. counties and county equivalents. These commuting zones are intended to represent more local labor markets. They are then aggregated into 394 Labor Market Areas (LMAs) by the Bureau of the Census, which uses a population threshold of 100,000 for the LMA designation. In health care policy LMAs are used for the calculation of certain inputs to payment systems for the Centers for Medicare and Medicaid Services (CMS) and have been used in the analysis of the ability of rural areas to recruit physicians (Brasure et al., 1999).


However, some of the first applications of automated cartography were used to address health services problems.

Copyright 2002 by the National Academy of Sciences. All rights reserved.
Bookshelf ID: NBK221045


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.9M)

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...