Using GIS and gemographics to characterize communities at risk: A model from ATSDR. (Special Report).

Introduction

The demographic characterization of a population living near a hazardous waste site provides the Agency for Toxic Substances and Disease Registry (ATSDR) with information essential for assessing and addressing public-health issues in local communities. ATSDR uses this information to (a) estimate the size and demographic characteristics of the population that potentially has been exposed to hazardous substances, (b) identify sensitive populations in the area (e.g., children, childbearing-aged women, and the elderly), and (c) provide insight about the social context of the area.

The ability to link population and housing data to the geographic location (i.e., a point defined by longitude and latitude) of a hazardous waste site through geographic information system (GIS) technology improves ATSDR's ability to estimate current population numbers. This results in a more accurate characterization of site demographics in a shorter time. It also helps the agency to develop communication and health education efforts for the local community and to identify potential environmental-justice issues at a site.

GIS and U.S. Census Data

ATSDR relies principally on data from the U.S. Department of Commerce, Bureau of the Census, to characterize community demographics. Decennial census data are the most complete source of residential-population information for the United States and its territories (e.g., Puerto Rico and the U.S. Virgin Islands). Demographic data are currently available from the 1990 census, and data from Census 2000 are expected to be released in spring 2001. The Bureau of the Census estimated that 248.7 million persons resided in the United States in 1990. By 2000, this estimate was expected to be more than 275 million.

Understanding the population distribution across the nation's landscape and using this information in a GIS requires knowledge of census geography (i.e., the political and statistical areas the Bureau of the Census uses to tabulate and report demographic data). Most U.S. residents know the country's geographic/political areas (e.g., states, counties, cities, and Native American reservations), but many are less familiar with the geographic/statistical areas that have been developed by the Bureau of the Census and that form the basis ATSDR uses to analyze and report population and housing data for small geographic areas (e.g., census tracts and census blocks) (Table 1) (Bureau of the Census, 1990, 1992).

U.S. census data describing residential population and housing provide the most complete picture of the nation and its sub-areas. Basic population and housing data (e.g., age, race/ethnicity, and sex) are reported for all census geographic levels (e.g., state, city, census tract, and census block). Additional demographic data, including socioeconomic information, are available at the block group level and at the larger geographic levels (Bureau of the Census, 1998).

ATSDR incorporates U.S. census population and housing data into a GIS application, which generates demographic maps for the areas near active hazardous waste sites (e.g., those listed by the U.S. Environmental Protection Agency [U.S. EPA] on the National Priorities List). A map of each site contains the location of the site and its surrounding area (within a one-mile buffer). Using an area-proportion spatial analysis technique in a GIS, demographic statistics from the 1990 census are calculated and the results listed in a site map. ATSDR also records these data in its HazDat database, which contains detailed site-specific information regarding hazardous waste sites.

Through GATHER, a site map can be generated that contains these demographic statistics for any hazardous waste site in the comprehensive HazDat database.

GIS maps that overlay data on population, environmental contaminants, and health outcomes are produced routinely for hazardous waste sites. When ATSDR staff request additional current and historical demographic data, such as housing age, length of residence, and socioeconomic status, those data are easily incorporated into the GIS system. ATSDR also provides GIS-generated demographic statistics in response to requests from other public-health agencies, including the Centers for Disease Control and Prevention (CDC) and the 28 state health departments that have cooperative agreements with ATSDR.

Demographic Data and Segmentation Systems

An issue in using census data to examine area demographics is that the data reflect the population and housing status of U.S. residents on April 1 of a census year (e.g., 1990, 2000), but area demographics can change over time, particularly in smaller geographic areas. Several companies use the most recently published census data to project current-year demographic estimates for the period between official censuses. These projections are then incorporated into a system that segments small geographic areas into categories reflecting the demographic characteristics, lifestyle, and consumer behavior patterns of residents.

To meet the agency's need for updated data statistics, ATSDR licenses with Claritas, Inc., for estimates of current-year demographic data at the block-group level. These data include age, race/ethnicity, sex, income, education, and housing. ATSDR uses the PRIZM segmentation system developed by Claritas, Inc., to obtain additional demographic and lifestyle information about populations living near hazardous waste sites (Claritas, Inc., 1994). The concept behind the PRIZM segmentation system is that individuals are more likely to live near others who share similar demographic and behavioral characteristics. The PRIZM segmentation system divides the U.S. population into 62 neighborhood types, or "clusters," on the basis of residential location (i.e., geodemographics).

PRIZM has copyrighted names for each of the 62 neighborhood types (e.g., "Kids and Cul-de-Sacs" and "Southside City") that are assigned to census block groups, census tracts/block numbering areas, and zip codes. These neighborhood types can be used to analyze and display demographic data in a GIS. The cluster names are meant to catch interest and convey a general sense of the character of an area; however, they should not be interpreted literally. PRIZM contains demographic and market research data that define and differentiate each cluster. The market research data include information on lifestyles (e.g., smoking, gardening, and exercise), media habits (e.g., television, newspaper, and radio), financial product usage (e.g., health insurance information), and general product use (e.g., foods and alcohol).

Used successfully as a marketing tool for many years by the business industry, segmentation systems are increasingly used by public-health agencies to (a) explore the relationship between community and disease, (b) target health interventions, and (c) develop communication strategies. For example, the Georgia Division of Public Health used cluster information to target mammography programs in factory towns designated as "Mines and Mills" because women in those communities had higher rates of breast cancer (Weiss, 2000). The CDC Office of Communication (OC) collaborates with the centers that constitute CDC on projects that integrate PRIZM data with epidemiologic and other data for communication planning (Pollard & Kirby, 1999). One example of this collaborative work concerns prevention efforts for hantavirus, Using zip code information, OC and the National Center for Infectious Diseases (NCID) identified the PRIZM clusters in which 164 cases of hantavirus had been diagnosed in the United States during 1993 thr ough 1997. Residents of these clusters then could be targeted as the priority audience which prevention messages about hantavirus should be addressed (Pollard & Kirby, 1999).

Community Research

In support of ATSDR programs and initiatives (e.g., the Minority Health Program and the Child Health Initiative) being undertaken to define the issues concerning hazardous substances and special populations, agency staff also use GIS technology and techniques to conduct demographic research on populations living near hazardous waste sites. This research helps ATSDR to understand the health status of these segments of the population, For example, if racial/ethnic minority groups and/or residents of areas designated as having lower socioeconomic status are more adversely affected by close proximity to hazardous waste sites, ATSDR must determine the implications for public-health activities in these areas.

In 1995, ATSDR published an article in the International Journal of Occupational Medicine and Toxicology about a GIS approach used to analyze the 1990 racial and ethnic composition of the population near 1,200 National Priorities List (NPL) sites (Heitgerd, Burg, & Strickland, 1995). Approximately 11 million persons were found to live within a one-mile buffer of these sites. ATSDR staff are updating this analysis using current-year demographic estimates for race/ethnicity and socioeconomic status and reviewing data for 1980-1999 to better understand the types of demographic changes that occur in areas near hazardous waste sites. Using the PRIZM segmentation system database, ATSDR is also studying the types of neighborhoods in which NPL sites are found. An initial analysis of the data suggests that a concentration of these sites exists in neighborhoods described as "Middle America," "Family Scramble," and "Military Quarters."

Creating Community Profiles

ATSDR is using a wide range of small-area demographic data to create community profiles as a logical first step in its public-health activities for communities near hazardous waste sites. Data products from Census 2000 will provide up-to-date, small-area population and housing data. The 2000 TIGER/Line File, the digital database of U.S. geographic features developed by the Bureau of the Census, includes Census 2000 statistical-area boundaries. Significant changes are not expected, except possibly in the reporting of race/ethnicity Census 2000--for the first time--instructed individuals to check one or more categories on the form's race/ethnicity question. ATSDR will continue to refine its GIS-based applications for population estimation to produce more timely and accurate demographic data for communities located near hazardous waste sites.

(Adapted from Hazardous Substances and Public Health, Vol. 10, No. 2, Summer/Fall 2000, a publication of the Agency for Toxic Substances and Disease Registry)

TABLE 1

Census Statistical Areas

Geographic Area       Definition

Census tract/block    Census tracts are small, locally
numbering area (BNA)  defined statistical areas in
                      metropolitan areas and densely
                      populated counties. In
                      nonmetropolitan areas, BNAs are
                      equivalent to census tracts. In
                      Census 2000, the term "Census
                      tract" replaces "BNA." Initially
                      delineated to be homogeneous with
                      respect to population
                      characteristics, economic status,
                      and living conditions, census
                      tracts generally have stable
                      boundaries to facilitate
                      longitudinal population
                      comparisons. Total population size
                      ranges from 2,500 to 8,000 persons,
                      with an average of 4,000. Area
                      size is determined by settlement
                      density. Census tract boundaries
                      are linked to census data in a GIS
                      with a six-digit Federal
                      Information Processing Standards
                      (FIPS) code.

Block group (BG)      BGs are groupings of census blocks
                      within census tracts or BNAs. The
                      first number assigned to each block
                      identifies the block group of which
                      it is a part. A block group may
                      contain up to 97 blocks, but
                      usually the number is substantially
                      less. BGs generally contain between
                      250 and 550 housing units, with an
                      average of 340.

Block                 Blocks are the smallest census
                      geographic area used by the Bureau
                      of the Census for data tabulation
                      and presentation. Block boundaries
                      are usually streets and other
                      prominent physical features.
                      Political divisions can also serve
                      as block boundaries. Blocks may be
                      as small as a typical city block
                      bounded by four streets or as large
                      as several square miles in rural
                      reaas. The 1990 census is the first
                      in which the entire country is
                      block-numbered. In the past, blocks
                      had a three-digit FIPS code (some
                      contained a letter at the end of
                      the code). In Census 2000, blocks
                      have a four-digit numeric code.

Geographic Area       Number in 1990

Census tract/block        62,276
numbering area (BNA)























Block group (BG)         229,192










Block                   7,017,425

REFERENCES

Bureau of the Census. (1990). Census '90 basics (Publication No. CPH-I-8). Washington, DC: U.S. Department of Commerce.

Bureau of the Census. (1992). 1990 census of population and housing: Summary tape file 3 technical documentation. Washington, DC: U.S. Department of Commerce.

Bureau of the Census. (1998). United States census 2000: Uses for questions on the census 2000 forms: Appendix B. Washington, DC: U.S. Department of Commerce, <http://www.census.gov/dmd/www.content.htm>.

Claritas, Inc. (1994). PRIZM methodology. Arlington, VA: Author.

Heitgerd, J.L., Burg, J.R., & Strickland, H.G. (1995). A geographic information systems approach to estimating and assessing National Priorities List site demographics: Racial and Hispanic origin composition. International Journal of Occupational Medicine and Toxicology, 4(3), 343-363.

Pollard, WE., & Kirby, S.D. (1999, August 2-4). Geographic information systems (GIS), public health data, and syndicated market research data bases in health communication. Paper presented at the National Conference on Health Statistics, Washington, DC.

Weiss, M.J. (2000). The clustered world. Boston, MA: Little, Brown and Company.