The data in this book and its companion volume (Monitoring the Health Care Safety Net—Book 2. Data for States and Counties) are presented at various geographic levels. In this book, data are provided for 90 urban areas in 30 States at the following levels: (1) Metropolitan Statistical Areas (MSAs), (2) cities greater than 100,000 population within those MSAs, (3) counties within those MSAs, and (4) "county residuals" (the remaining portion of the county outside the boundaries of the city for which data are presented). In Book II, the data are presented at the county and State levels.
Data in this volume are always presented at the smallest geographic level for which they are available and for which they can be reasonably displayed. However, not all data are available at all levels, and some data must be aggregated up from smaller to larger levels, which can be problematic. For example, preventable hospitalization data come from hospital discharge abstracts that include the ZIP Code of residence for each patient. These data can be presented at all levels, but data at the city, county, and county residual levels are aggregated up from ZIP Codes to these levels. Because ZIP Code boundaries often overlap county or city boundaries, data at these aggregated levels may include small numbers of individuals from outside the jurisdiction or may omit some who actually should be included. Other data are available only at the MSA level; for example, data on managed care penetration and data from the National Health Interview Survey are available only for major metropolitan areas. Caution is necessary when interpreting these measures because large differences may exist between areas within an individual MSA. The MSA rate is an average across all counties and cities within the metropolitan area.
While the observation that "all politics is local" has become a well-accepted principle of American politics, it is also becoming increasingly recognized that "all safety nets are local." While Federal and State policies affect the demand on, support for, and structure of a health care safety net, in most areas the problems of meeting the health care needs of the Nation's uninsured, low-income, and other vulnerable populations are largely the responsibility of local providers, institutions, and governments. Although industrial development strategies, enterprise zones, and regional mass transportation systems have shown the need to cross political boundaries to solve many of the major problems facing the Nation's urban areas, relatively few initiatives attempt to cross political boundaries to address health care safety net issues.
In most States, the county is the primary political jurisdiction with responsibility for the health care needs of the indigent. In the mid-20th century, that often meant a county hospital or a county health department provided preventive care, especially for children. However, the number of county-owned and -operated hospitals has declined substantially, and health departments in many parts of the country have cut back on direct service operations. Counties have coped differently with these circumstances, and the level of financial support and the structure of health care systems differ widely, even within a metropolitan area. Accordingly, in these two data books, information is provided at the county level where available.
Political boundaries can also present complications. In many jurisdictions, many of the functions of local health departments are assumed by city authorities for major cities, which often have their own health departments or public facilities. An exception to this is in Michigan, where safety net care for Detroit residents and most of the health department functions are the responsibility of municipal authorities. Detroit has struggled to meet the needs of vulnerable populations, especially in the absence of a major public hospital. However, Detroit is located in Wayne County, more than half of whose residents live outside of Detroit. This means, in effect, that the "residual" of Wayne County constitutes a separate safety net. And while poverty levels in Detroit are generally more severe than those in the rest of Wayne County, almost 30 percent of all Wayne County low- income residents live outside of Detroit (in the "residual" of Wayne County). Furthermore, some areas within the perimeter of Detroit's political boundaries are not part of municipal Detroit (Go to Figure 2-1).
Figure 2-1: Political Boundaries and Population Income, Detroit Metropolitan Area
In Miami, FL, the circumstances are different. Again, a major municipality is located within a county (Dade County), with a substantial portion of area residents (more than 80 percent) living outside the boundaries of the city of Miami. Unlike in Detroit, however, the county has retained primary jurisdiction for safety net activities and health care for indigent populations, with a large public hospital system and a network of community-based clinics that are partially supported by a countywide 1/2 cent sales tax.
A third model exists in New Jersey, where local health departments have primary responsibility for safety net care. These health departments are organized at the municipal or township level as well as the county level. As illustrated in Figure 2-2 for northeastern New Jersey, this model creates a system with hundreds of entities, each of which has some safety net responsibilities. New Jersey residents are often unaware of their township residence, and health care utilization patterns cut across all these boundaries.
Figure 2-2: City/Town and County Boundaries, Northeastern New Jersey
Sorting out these local boundaries and the relative responsibilities of the various political entities is beyond the scope of the data books, which provide data at as many levels as possible to facilitate their use by policymakers, analysts, and planners. However, an important first step in examining the needs of a local safety net is defining the geographic area of concern and determining the specific responsibilities of the various political entities in the area.
Another potentially important aspect of "place" is the distinction between cities and suburbs. While the simple model of a large central city surrounded by residential suburbs of commuters is no longer accurate given intrasuburban traffic and employment patterns, important differences often exist between central city safety nets and those of suburban areas. The most obvious differences may be among the most important: demographics and population density.
Almost by definition, central cities are typically more densely populated than are suburban areas. With respect to the safety net, the implications of density relate to the potential geographic concentration of population subgroups that may be dependent on the safety net, including low-income populations, recent immigrants, and others. The analysis in Chapters 3 and 6 also demonstrates that central cities typically have higher percentages of low-income, uninsured, and racial/ethnic minority populations.
This combination of large and concentrated numbers of individuals likely to be dependent on the safety net in central cities generates two potentially contradictory concerns about the safety net. First, the magnitude of the "problem" is likely to be greater in central city areas, creating special pressures and problems for safety nets. Second, the dispersion of populations in need throughout suburban areas can make meeting their needs more difficult, preventing the placement of appropriate providers near the populations they serve.
Accordingly, in the data presented in Chapters 3 through 7 and the tables in Part II of this book, large differences between the circumstances and health outcomes for central city and suburban populations are documented and highlighted to help observers begin to explore these issues, with tables breaking out central city and suburban data. These data do not provide a single answer to these competing concerns, but further illustrate the differences among safety nets. In some communities, outcomes and performance (after adjusting for differences in population mix) are actually worse in the suburbs than in the central city. In examining a safety net, it is important to assess the differences between the needs and problems of safety nets in central city and suburban areas and to consider solutions or interventions that address these differences.
As noted above, data are presented at various levels of aggregation (MSA, county, city, and county "residual"), including the smallest area for which the data are available. This organization reflects the view that greater geographic resolution allows for a more precise understanding of the safety net and its problems, and can help target interventions most effectively.
A fairly typical example involves hospital discharge data and examining rates of preventable hospitalization for ambulatory care sensitive (ACS) conditions-those for which timely and effective ambulatory care can help prevent or avoid hospital admissions. Data at the county level can be valuable, especially in less populated areas where small population sizes can result in an instability in rates due to random variation. For example, in Georgia, an analysis of county-level preventable hospitalization rates for adults ages 40-64 reveals relatively high rates for some counties in the east-central and south-central portions of the State (Go to Figure 2-3).
Figure 2-3: Preventable Hospitalizations, County Level, Georgia, 1999
However, for more densely populated areas, county-level data can mask important differences at the subcounty level, where some areas within the county have much higher rates than others, but the overall county rate is within the average range of other counties. For example, in the Atlanta metropolitan area, preventable hospitalization rates for residents ages 40-64 are relatively low across all counties, as illustrated in Figure 2-4. However, examining these data at the ZIP Code level reveals large differences within the metropolitan area, with some areas having very high rates. Many, but not all, of these areas are within the Atlanta city boundaries (Go to Figure 2-5). Space limitations, lack of data, and, in some cases, confidentiality concerns make the presentation of ZIP Code-level data impractical for this book, but the importance of examining the smallest geographic area possible cannot be overemphasized.
Figure 2-4: Preventable Hospitalizations, County Level, Atlanta Metropolitan Area, 1999
The "smallest geographic area possible" in any given case is dependent both on the data source and the relative frequency of the events being measured. For low population density areas or for measuring relatively rare events (such as infant mortality), county-level data may be the smallest practicable level. But in densely populated urban areas, even the ZIP Code level may provide insufficient geographic resolution, and smaller levels of aggregation, such as the census tract, if the data source permits geocoding to that level, may be preferable.