Thomas B. Richards, MD Charles M. Croner, PhD Gerard Rushton, PhD Carol K. Brown, MS Littleton Fowler, DDS

Public Health Reports 1999;114:359-373.
Reproduced here with the permission of Oxford University Press.

LOOKING toward the 21st century, we anticipate that health planning, particularly at the community level, will be substantially improved by developments in informatics -- that is, through the application of information science and technology to public health practice and research. ( 1 ) Ideally, each community will have the capability to link together health information from a variety of different data sources and to recognize spatial data patterns that suggest where cost-effective public health interventions can be applied.

We believe that geographic information system (GIS) technology will be an important part of the toolkit to support this capability, but only if epidemiologic principles and methods provide the foundation for the data analyses to be displayed in GIS maps.( 2-6 ) Here, we provide our thoughts about current trends and future directions for GIS technology in public health and the challenges to institutionalizing this technology as a standard part of public health planning and practice by the Year 2010.

Current Trends

During the 1990s, local government use of GIS technology has grown substantially. ( 7 ) A national survey in 1997 of a sample of 200 cities and counties concluded that use of GIS technology in at least one department of local government had increased from 20% of jurisdictions in 1990 to a predicted 87% by the end of 1997. ( 7 ) Respondents named planning departments as the most frequent local government users (almost 80% of respondents in all jurisdictions).

Based on informal discussions with representatives of many local public health departments, the National Association of County and City Health Officials (NACCHO) reports that interest in GIS technology has increased during the 1990s, but many local public health departments still do not have the software, hardware, or trained staff that would enable them to apply GIS technology. ( 8 ) Local public health departments would like to see GIS components added to NACCHO's community planning tools such as the Assessment Protocol for Excellence in Public Health and its new iteration, Assessment and Planning Excellence through Community Partners for Health, scheduled for release in year 2000 in electronic format (for more details about these tools, see NACCHO's website). NACCHO is currently developing plans for a national survey to document local public health departments' GIS activities.

One of the initial steps in any GIS project is to geocode (georeference) each data record to the desired level of accuracy (for example, county, Census tract, Census block, US Postal Service zip code, or street address). (The smallest area of US Bureau of Census geography is the Census block. In urbanized areas, a Census block typically is a quadrangle bounded by four streets (a city block). In sparsely populated areas, a Census block has a population of about 70 people and is bounded by visible features such as roads, streams, or railroad tracks or by invisible boundaries such as city or county limits. In rural areas, a block may encompass many square miles.)

During the 1990s, geocoded public health data have been in relatively short supply, limited to states with initiatives to geocode vital statistics data or to individual investigators who could geocode their own data. In a 1997 survey of state Vital Statistics Project Directors, only 21 of 49 respondents reported that their states had some type of automated geocoding of vital statistics.( 9 )

The growing interest in GIS applications in public health is illustrated by the Third National Conference on GIS in Public Health, held in San Diego in August 1998. Participants addressed a wide variety of topics, ranging from the use of satellite images to measure ocean temperatures and forecast cholera epidemics in India to the use of global positioning system (GPS) technology to determine latitude-longitude coordinates for locations of billboards with cigarette advertisements in relation to school bus routes.

Epidemiology and GIS Technology

Epidemiologic principles and methods provide the foundation for public health and preventive medicine. To avoid drawing false conclusions from maps, users of GIS technology need to understand and apply these principles and methods in formulating study questions, testing hypotheses about cause-and-effect relationships, and critically evaluating how data quality, confounding factors, and bias may influence the interpretation of results. Conversely, epidemiologists need to be able to understand and critically evaluate maps prepared using modern GIS software, data, and spatial statistical methods.

A recent special issue of the Journal of Public Health Management and Practice was devoted to GIS technology. (A second special issue is due out in July 1999.) In an editorial, Melnick and Fleming discuss the "promise and pitfalls" of GIS technology.( 5 ) They note that "perhaps the greatest potential of GIS lies in its ability to quickly, clearly, and convincingly show the results of a complex analysis.... The greatest strength of GIS is that its product is a picture." They go on to say, "Ironically, the power of the GIS tool may also be its biggest pitfall. The consequence [of integrating] complex data into a visually easy to understand picture ... is a setup for misunderstanding and misuse." Users, including policy makers, they point out, may be tempted to infer causation from correlation and to make inferences about individuals from population data (the ecologic fallacy).( 5 )

Another potential problem with drawing conclusions from maps is that, as Monmonier writes, "Not only is it easy to lie with maps, it's essential...." ( 10 ) The cartographer's paradox is that "to avoid hiding critical information in a fog of detail, the map must offer a selective, incomplete view of reality." ( 10 ) Public health practitioners need to be alert for "lies" that can range from "little white lies" (suppressing details selectively to help the user see what needs to be seen) to more serious distortions in which the visual image suggests conclusions that would not be supported by careful epidemiologic analysis. For example, when some geographic units of analysis have small denominators, disease rates computed for these areas may appear extremely high if any cases have occurred in these areas. When the rates for these geographic locations are displayed on a map, readers may incorrectly conclude that these are "hot spots," high priority locations for targeted interventions. More appropriately, these areas should be labeled to indicate that rates are statistically unstable due to small numbers and therefore not shown.

With advances in desk-top computing, it will become easier to produce multiple maps using different data sources and different methods, each offering a unique perspective on an issue. Again, the principles and methods of epidemiology need to be applied to evaluate the strengths and limitations of the data and the science "behind" the maps, in order to identify the map (or maps) that convey the most truthful messages.

The ability to invest in GIS hardware, software, and data may enable those with greater resources to be more influential in communicating their "selective, incomplete view of reality" to community decision makers.( 11 ) In the future, as advocacy groups, community organizations, hospitals, managed care organizations, and the news media increasingly use GIS technology with health-related data, the need for education about epidemiologic methods and principles becomes even more essential.

Definition

The phrase "geographic information systems" was first used in the 1960s to refer to a computerized system for asking questions of maps showing current and potential land use in Canada.( 4 ) Since that time, a number of definitions have been proposed, with variations that depend on the perspective of the author, the specific application, the software available at a given time, and the level of complexity appropriate for the intended audience. Some authors have begun to suggest that a different term, "geographic information science," might be advantageous in order to place greater emphasis on the underlying general principles and science and to be more independent of developments in software technology.( 4 )

From a community health planning perspective, the Federal Geographic Data Committee (FGDC) definition provides a useful starting point:

A computer system for the input, storage, maintenance, management, retrieval, analysis, synthesis, and output of geographic or location-based information. In the most restrictive usage, GIS refers only to hardware and software. In common usage, it includes hardware, software, and data. When organizations refer to their GIS, this latter usage is usually what they mean. For some, GIS also implies the people and procedures involved in GIS operation.( 12 )

The inclusion of "people" and "procedures" as part of the definition is essential for GIS applications in a public health context, given the need to link the science and methods of epidemiology to GIS maps. Without trained staff, one scenario is that the GIS software will not be used at all (given the time and staff constraints that exist in many public health agencies and organizations). Alternatively, without trained staff and standardized procedures, the technology may be used to develop maps that are invalid or misleading.

Advantages of GIS Technology

Several advantages of GIS technology for public health practice, planning, and research are as follows:

(a) GIS technology improves the ability of practitioners, planners, and researchers to organize and link datasets (for example, by using geocoded addresses or geographic boundaries). Geography provides a near-universal link for sorting and integrating records from multiple information sources into a more coherent whole. This ability to link datasets can help public health practitioners plan more cost-effective interventions. For example, suppose that a childhood lead poisoning prevention program could access residential databases maintained by the tax assessor's office and map the street addresses of houses built before 1950 (when lead-based paint was commonly used). Suppose that the prevention program could also access hospital and managed care plan electronic databases to identify street addresses for new births. Combining these datasets, the program could apply GIS technology to identify infants at high risk for exposure to lead-based paint and send a public health worker to follow up with specific households. By matching the addresses of these infants to a street map (from a "topologically integrated geographic encoding and referencing" [TIGER] file), using the "address-match" and "route-scheduling" functions of GIS software, the health worker can implement an efficient schedule of household visitations.

(b) GIS technology provides public health practitioners and researchers with several new types of data. For example, with GIS technology, local public health departments can use global positioning systems (GPS) to receive signals from satellites to determine latitude-longitude coordinates for point locations not found in TIGER files, such as rural residences, wells, and septic tanks. Public health practitioners can also use digital imagery from satellites or aerial photos to add details to (or improve the accuracy of) a mapping project (for examples of digital imagery, see www.terraserver.com, www.spaceimaging.com, www.ogeta.com, or www.usgs.org). If a sequence of digital images for a small area of interest is available, automated change detection can be used to observe changes over time, such as the addition of housing developments, roads, and landfills and other changes in land use and land cover. Public health practitioners can also begin to explore the utility of data collected by marketing firms about consumer spending patterns, retail expenditures, and lifestyle segmentation profiles (for examples, see www.natdecsys.com, www.demographics.com, or www.claritis.com). Businesses use these marketing data to identify likely customers--for example, to optimize targeting of cigarette advertising. Public health practitioners could also potentially use these data to identify the best "customers" for prevention interventions--for example, anti-smoking programs.( 13 )

(c) GIS technology encourages the formation of data partnerships and data sharing at the community level. For example, to develop a map of motor vehicle injuries and fatalities in a community, a local public health department could develop data partnerships with the Department of Transportation (for information about traffic flow and accidents), local ambulance services (for information about injuries requiring transportation by ambulance to hospital emergency rooms), and the Medical Examiner's office (for information about fatalities). Some GIS projects may be feasible only if all parts of local government join together and contribute (for example, developing a regional data warehouse or obtaining digital aerial photos or satellite images for an entire region).

(d) As new GIS methods are developed, they can be added to the "toolkits" of epidemiology and health services research. For example, for an exploratory data analysis in response to community concern about an apparent excess of cancer cases, tests for spatial randomness (such as the spatial scan statistic described by Kulldorff ( 14 ) could be used to evaluate if a cluster is present using Monte Carlo simulations. The National Cancer Institute has developed software known as SaTScan to perform these calculations. Copies of this software are available free of charge.( 14 ) Maps of "smoothed" rates could be developed to reveal neighborhoods at maximum risk using the DMAP Program (University of Iowa) to compute grid values (see Figure).( 15 ) GIS technology also could be used to link data for an individual with contextual information aggregated at a variety of geographic levels (for example, Census block group, Census tract, county, or state). This capability enables preparation of a multi-level spatial model to better evaluate and distinguish biologic, contextual, and ecologic effects. While conclusions based on an analysis at the aggregate level are likely to be limited by the ecologic fallacy and by aggregation bias (failing to identify the true nature of cause-effect relationships at the level of the individual), conclusions based on analysis at the individual level may be limited by the atomistic fallacy (failing to consider the context in which individual behavior occurs). ( 16-18 ) For each factor considered, a multi-level model can include both individual predictors (data for each individual) and ecologic predictors (average or aggregate measures).( 18 )

(e) Compared with tables and charts, maps developed using GIS technology can be an extremely effective tool to help community decision makers visualize and understand a public health problem.( 19 ) In addition, action is more likely when the decision maker can see on a map that a problem is occurring in his or her "backyard." GIS technology enables detailed maps to be generated with relative speed and ease. In turn, this provides public health practitioners with the ability to provide quick responses to questions or concerns raised in a community meeting, for example, by preparing supplemental maps or by displaying more information about a point on the map during the course of the meeting.

Figure. A smoothed map of infant mortality rates in southeast
Des Moines, Iowa, 1989-1992

(click on map for an enlarged view)

The Figure shows a "smoothed" map of infant mortality rates in southeast Des Moines, Iowa, that was prepared for a public meeting. Births and infant deaths in this area for 1989 - 1992 were geocoded by matching addresses in vital statistics records to the US Census data TIGER files of streets for the area. The contour lines show the variations in infant mortality rates; the highest rates were found near the McKinley School and the drive-in theater. (The feature names are from the digital data files of the US Geological Survey's 1:24,000 Quadrangle Series, available for most of the US.)

The circle shows the size of the spatial filter used to "smooth" the map. (Small filters show more geographic detail but risk showing patterns of random variability; high-rate areas on such maps are often due to chance variability in incidence of the disease or condition, as discussed above. Consequently, the study of such maps should involve comparing results of varying sizes of spatial filters and statistical analyses of the apparent patterns to determine significant clusters.)

At the meeting, members of the public could comment on specific locations of interest. With each click of the mouse on the map, a circle appeared around the point selected and a box appeared on the screen showing the numbers of births and infant deaths within the area of the circle. This dynamic feature of geographic information technology (the ability to display information linked to the map) can be very useful to public health practitioners and the public in identifying problems and in searching for solutions. By quickly showing the information on which differences in rates are based, this feature helps avoid erroneous interpretations of rate differences that are based on small numbers.

Current Limitations of GIS Technology

Some of the current limitations of GIS technology from a public health perspective are as follows:

(a) Community health planning and other public health applications remain a relatively underdeveloped marketplace niche for GIS technology. The Department of Housing and Urban Development's Community 2020 software represents an important first effort by a federal agency to integrate GIS technology into community health planning.( 20 ) Considerable room remains, however, for improvements and developments in GIS software for public health applications. For example, GIS technology could be linked with community planning tools such as NACCHO's Assessment and Planning Excellence through Community Partners for Health, and specialized GIS software products, including data entry forms and automated procedures, could be designed to help public health practitioners map and plan interventions at the community level.

(b) Current, accurate, low-cost base street maps are essential for epidemiologic uses. Without an up-to-date base street map, for example, a public health practitioner investigating a disease outbreak may have to spend considerable extra time and effort to digitize the locations of cases or may not be able to map all case reports. Current and accurate base street maps are especially needed for urban areas with high growth and for those rural areas where residents only have post office box addresses.

(c) Practitioners, planners, and researchers, and especially state and local public health department staff, need training and user support in GIS technology, data, and epidemiologic methods in order to use GIS technology appropriately and effectively. The cost of training programs offered by commercial GIS vendors can be a financial burden for a small local public health agency or individual practitioner ($500 to $1000 for a two- to three-day course is common). GIS training programs specifically custom-designed for public health professionals are still relatively limited or in the early stages of development. At least two groups have started to "break the ice" in this area. Gerard Rushton, PhD, at the University of Iowa, has developed a CD-ROM on "Improving Public Health Through Geographic Information Systems,"( 15 ) and the Agency for Toxic Substances and Disease Registry (ATSDR) has developed course handouts and a syllabus for a two-day introductory course on GIS technology for ATSDR personnel (for a copy, contact Dr. Virginia Lee). The time required for training can be a severe challenge for organizations in which demands on personnel are already high. Another drawback is that public health professional specialties currently do not recognize continuing education credit for individuals who participate in GIS software training.

(d) Statistical and epidemiological methods need to be developed to protect individual and household confidentiality.( 21 ) Even if a single database may appear to have effective confidentiality safeguards, when several databases are linked within a geographic information system, the "sum" may be less well protected than the "parts." A false identification may be just as damaging to an individual as a correct identification that is not kept confidential.

(e) GIS software continues to evolve rapidly; typically, a new iteration (or upgrade) is released about every 18 months.( 22 ) Every software package has its strengths and weaknesses. Current prices for some GIS products (in particular, for Web-enabled GIS applications and for neighborhood lifestyle segmentation datasets) remain a potential barrier (running as much as $10,000 or more). In addition, costs for maintenance and upgrades can be substantial.

(f) The technology to prepare and display maps on the Web is still in the very early stages of development.( 23 ) Models and methods for Web-enabled GIS technology need to be developed for public health applications and field tested.( 23 ) As Foresman points out, full GIS capability on the Web is a considerable technical challenge because GIS software has only recently started to be developed using Web-accessible programming languages, and the size of GIS map images and data files can be large and significantly slow access and display functions over the Web.( 23 ) Spatial statistical software programs will also need to be developed for use with these Web-enabled GIS applications; as Foresman writes, "The next logical step is to actually provide the Web tools to perform spatial statistical analysis over the Web." ( 23 )

Anticipated Developments in the Near Future

The next two to three years will see exciting developments in public health applications of GIS technology:

First, we anticipate that an increasing number of state public health departments will use automated geocoding for some or all of their vital statistics data. The geographic information included in vital records can play a critical role in the distribution of state and federal funds for infrastructure, community development, public education, and initiatives such as Healthy Start.( 9 ) The expanded use of automated geocoding would be consistent with the draft Healthy People 2010 objective of "Increas[ing] the use of geocoding in all major national and State health data systems to promote the development of [GIS] capability at national, State, and local levels."

Second, we anticipate that public health practitioners and researchers will apply GIS technology to analyze data from disease registries, to track patterns and trends in, for example, disease incidence, stage at diagnosis for many cancers, survival, mortality, and use of medical services. Public health agencies, managed care organizations, and other providers will study these geographic patterns to develop and locate services that address the needs of areas which are identified as disproportionately sharing the burden of particular diseases. The Surveillance Implementation Group of the National Cancer Institute, for example, recently recommended that future studies "explore the feasibility and utility of employing GIS for geocoding surveillance data and reporting geographic relationships among screening measures, risk factors (including environmental exposures), and improved cancer outcomes...."( 24 )

GIS-related statistical and epidemiological methods will be improved, and new methods will be developed. Promising areas include the use of spatial scan statistics to identify disease clusters; the use of smoothed mapping techniques to display and distinguish differences in rates at the neighborhood level; and multi-level spatial models to better evaluate and distinguish biologic, contextual, and ecologic effects.( 14-18 )

Over the next two to three years, we also anticipate that new software products and training materials will be developed to facilitate wider use of GIS technology by state and local public health practitioners. Training materials will need to be in a variety of formats to facilitate learning at a distance (for example, CD-ROMs; self-instruction training courses on the Web; and national educational broadcasts via satellite). Low-cost, public domain software solutions are needed, especially for small local public health agencies and programs with limited resources. As noted above, NACCHO is currently working to identify support and partnerships to integrate GIS technology, data, and methods with its community planning tools. In addition, the Centers for Disease Control and Prevention (CDC) is currently developing Epi Info 2000, a Windows NT/95/98 version of the public domain Epi Info and Epi Map computer software tools.( 25 ) Epi Info 2000 is being designed to allow public health practitioners to import, utilize, and display map boundary files and data from a GIS system. This additional capability will potentially enable Epi Info 2000 to be used to expand public health use of GIS technology. For example, a state or federal public health agency could make Epi Info 2000 available to smaller agencies, together with map boundaries, data files, and training. Epi Info 2000 will be available free of charge on the Internet.

Over the next few years, an increasing number of state public health departments and other public health agencies are likely to explore development of websites that employ GIS software, data, and methods. In some states, access may be restricted; selected staff members of local public health departments may use passwords to download confidential datasets geocoded by the state public health department. Some states may also begin to explore Web-enabled GIS applications; local health departments could then add their own data and custom-design their own GIS maps. (For early prototypes of interactive Internet mapping, see www.ciesin.org and www.epa.gov/enviro.)

We also anticipate that, over the next few years, there will be increasing discussion of ways to best protect individual and household confidentiality in a GIS environment. These discussions are likely to result in the development of new statistical and epidemiologic methods to assure data confidentiality, such as methods that mask the geographic location of data but still permit meaningful analysis( 21 ); the creation of secure environments with limited access (perhaps maintained by federal agencies such as the National Center for Health Statistics), in which public health researchers can be carefully monitored to ensure protection of individual and household confidentiality( 1 ); and the development and strict enforcement of policies, laws, and regulations to protect individual and household confidentiality.( 1 )

In the near future, public health practitioners will increasingly have access to aerial photos and satellite digital images, to data from global positioning systems (GPS), and to lifestyle segmentation marketing data. For example, Georgia's Geographic Information Systems Coordinating Committee has recently begun working to develop low-cost, high resolution images (Digital Orthophoto Quarter Quadrangles, or DOQQs) for the entire state. In 1998, the state entered into an agreement with the US Geological Survey to acquire updated aerial photographs through the National Aerial Photography Program. The total cost will be about $1.5 million for more than 4000 DOQQs covering the entire state.

Similarly, the Washington State Department of Health began a project in 1998 in which GPS technology is being used to develop a 10-county regional database on water systems, including information on well depth, ownership, test results, inspection dates, and number of connections.

With regard to lifestyle segmentation marketing data, we anticipate increased research in the near future on how to use GIS technology to improve the development, production, and delivery of health promotion and education information for national, state, and local campaigns. Private sector marketing firms have developed ways to categorize consumer behavior patterns at the neighborhood level and to select the best media channels to advertise a product to a specific market segment. Although public health agencies may decide to lease lifestyle segmentation marketing information databases from commercial vendors, the cost is relatively expensive (as much as $40,000 for a national database). With the help of GIS technology, public health researchers should be able to develop a low cost, public domain analog. These public health lifestyle segmentation profiles could potentially be linked to national surveys such as the CDC Behavioral Risk Factor Surveillance System and then used to develop behavioral risk factor projections at the community level. Along these lines, the Health Care Financing Administration is starting to explore the use of segmentation and cluster profiling of populations eligible for services under Medicare, Medicaid, the Children's Health Insurance Program, and the Health Insurance Portability and Accountability Care Act (approximately 70 million people) to help the agency "in its outreach, health promotions, educational campaigns, and in identifying the best channels for reaching these beneficiaries."( 26 )

Finally, we anticipate that community knowledge and sophistication about the application and interpretation of GIS technology will increase and that a variety of interested groups will develop GIS maps in addition to the maps developed by government public health agencies. The Department of Housing and Urban Development (HUD) provides training so that community representatives can use HUD Community 2020 software at the neighborhood level. GIS training is now being included in science programs in some schools as early as the 9th grade; the National Geographic Society and Environmental Systems Research Institute, Inc. [ESRI], currently co-sponsor an annual competition for GIS projects by high school students. In addition, the US Geological Survey, a number of other federal agencies, and ESRI have been involved in developing annual Summer Faculty Geographic Information System Workshops for the Historically Black Colleges and Universities.

Anticipated Developments in 5 - 10 Years

Over the next 5 to 10 years, we expect to see growth in local data partnerships and regional GIS consortia with shared data and automated systems. Three examples of early prototypes along these lines are the San Diego Regional Urban Information System in California; the Louisville/ Jefferson County Information Consortium in Kentucky; and the vision for the Clackamas County Community Health Mapping Engine in Oregon described by Melnick et al.( 27 )

Within 10 years, it is likely that public health practitioners will be able to perform spatial analyses on their computers through the Web in a cost-effective fashion. We anticipate the development of GIS software tools specifically custom-designed for use by local public health departments and other local agencies and organizations with limited staff and resources. To some extent, GIS technology even may become embedded in public health practice to the extent that the technology is so deeply "buried" that it is invisible to the worker. For example, with "embedded" GIS technology, a public health nurse would enter the laboratory data for a childhood lead poisoning prevention program into a computer program, and the computer would automatically generate a GIS map displaying the locations of cases needing follow-up.

As the number of public health practitioners applying GIS technology begins to increase, we also anticipate there will be increasing demand to share GIS maps. For example, suppose public health department X and neighboring public health department Y are addressing a common infectious disease problem and they would like to join their independently developed geographic information systems maps into a common map for both jurisdictions. Doing so requires consensus on issues such as the software to be used, the baseline street map for geocoding cases, the scale, the projection (i.e., the method for representing the curved surface of the earth on a flat surface), case definitions, sources for case reports, the time period for the study, and so forth.

Given the frequent need for sharing and, in some cases, combining maps, we anticipate more efforts at the local, state, and federal levels to develop standards for a national public health spatial data infrastructure.( 12 )

Conclusions

A rise and fall in enthusiasm is a typical part of the life cycle of any technologic innovation: Early adopters demonstrate initial success. The majority then begin to "jump on the band wagon," and some experience problems or incorrectly apply the technology. Critics then denounce limitations and abuses, and the popularity of the technology may decline. Ultimately, a middle ground is recognized between enthusiasts and critics, where the technology can be demonstrated to be most useful. In terms of this framework, public health applications of GIS technology in 1999 are still in the early stages.

Many challenges remain that will need to be addressed before the full potential of GIS technology can be realized for public health practice, planning, and research. Longer-term solutions are likely to require a series of small successes, carefully built upon in incremental fashion over time. One of the greatest challenges for public health applications will be to incorporate epidemiologic principles and methods into the analysis to be mapped. Another major challenge will be to develop methods and procedures to assure the confidentiality of individuals and individual households. A continuing local/state/national dialogue, interagency and private-public partnerships, and uniform local/state/federal standards will be needed to address these challenges.

The authors thank Aniruddha Banerjee, MA, Department of Geography, University of Iowa, Iowa City, for preparing the Figure.

Dr. Richards is a Medical Officer with the Public Health Practice Program Office at the Centers for Disease Control and Prevention (CDC), Atlanta. Dr. Croner is the Editor of Public Health GIS News and Information and a Geographer/Statistician with the Office of Research and Methodology, National Center for Health Statistics, CDC, Hyattsville, MD. Dr. Rushton is a Professor of Geography and Adjunct Professor of Preventive Medicine, University of Iowa, and a member of the editorial boards of Location Science, Geographical Information Systems, and Applied Geographical Studies. Ms. Brown is the Director of Research and Development, National Association of County and City Health Officials, Washington DC. Dr. Fowler is the President of the Association of State and Territorial Local Health Liaison Officials and Programs Administrator, Cleveland County Health Department, Norman, OK.

---

References

1. Yasnoff WA, Sondik EJ. Geographic information systems (GIS) in public health practice in the new millennium. Journal of Public Health Management Practice 1999;5(4):ix-xii.
2. Richards TB, Croner CM, Novick LF. Atlas of state and local geographic information systems (GIS) maps to improve community health. Journal of Public Health Management Practice 1999;5(2):2-8.
3. Richards TB, Croner CM, Novick LF. Getting started with geographic information systems (GIS), part 1. Journal of Public Health Management Practice 1999;5(2):73-6.
4. Longley PA, Goodchild MF, Maguire DJ, Rhind DW. Introduction. In: Longley PA, Goodchild MF, Maguire DJ, Rhind DW, editors. Geographical information systems, principles and technical issues. Vol. 1. 2nd ed. New York: John Wiley & Sons;1999:1-27.
5. Melnick AL, Fleming DW. Modern geographic information system -- promise and pitfalls. Journal of Public Health Management Practice 1999;5(2):viii-x.
6. Goodman DC, Wennberg JE. Maps and health: the challenges of interpretation. Journal of Public Health Management Practice 1999;5(4):xiii-xvii.
7. Warnecke L, Beattie J, Kollin C, Lyday W. Geographic information technology in cities and counties: a nationwide assessment. Washington, DC: American Forests; 1998.
8. Bouton PB, Fraser M. Local health departments and GIS: the perspective of the National Association of County and City Health Officials. Journal of Public Health Management Practice 1999;5(4):33-41.
9. MacDorman MF, Gay GA. State initiatives in geocoding vital statistics data. Journal of Public Health Management Practice 1999;5(2):91-3.
10. Monmonier M. How to lie with maps. 2nd ed. Chicago: University of Chicago Press; 1996.
11. Pickles J. Ground truth: the social implications of geographic information systems. New York, NY: Guilford Press. 1995.
12. Federal Geographic Data Committee (US). Framework introduction and guide. Washington: The Committee; 1997.
13. Rogers MY. Using marketing information to focus smoking cessation programs in specific census block groups along the Buford Highway corridor, DeKalb County, Georgia, 1996. Journal of Public Health Management Practice 1999;5(2):55-57.
14. Kulldorff M. Geographic information systems (GIS) and community health: some statistical issues. Journal of Public Health Management Practice 1999;5(2):100-6.
15. Rushton G. Improving public health through geographical information systems: an instructional guide to major concepts and their implementation [CD-ROM]. Version 2.0. Iowa City: University of Iowa, Department of Geography; 1997 Dec. Available at: URL: http://www.uiowa.edu/~geog/
16. Wrigley N, Holt T, Steel D, Tranmer M. Analyzing, modeling, and resolving the ecological fallacy. In: Longley P, Batty M, editors. Spatial analysis: modeling in a GIS environment. Cambridge [UK]: GeoInformation International; 1996. p. 23-40.
17. Jones K, Duncan C. People and places: the multilevel model as a general framework for the quantitative analysis of geographical data. In: Longley P, Batty M, editors. Spatial analysis: modeling in a GIS environment. Cambridge[UK]: GeoInformation International; 1996. p. 79-104.
18. Morgenstern H. Ecologic studies. In: Rothman KJ, Greenland S. Modern epidemiology. 2nd edition. Philadelphia, PA: Lippincott-Raven Publishers, 1998: 459-480.
19. Exeter DJ. An evaluation of cartographic visualization techniques for epidemiology [master's thesis]. Auckland [NZ]: University of Auckland Department of Geography; 1998. Available at: URL: ftp://ftp.geog.auckland.ac.nz/pub/outgoing/dxthesis.zip
20. Dean L. Revitalizing communities with geographic information systems (GIS): HUD's Community 2020 software. Journal of Public Health Management Practice 1999. 5(4):47-53.
21. Armstrong MP, Rushton G, Zimmerman DL. Geographically masking health data to preserve confidentiality. Stat Med 1999;18:497-525.
22. Thrall SE. Geographic information system (GIS) hardware and software. Journal of Public Health Management Practice 1999;5(2): 82-90.
23. Foresman TW. Spatial analysis and mapping on the Internet. Journal of Public Health Management Practice 1999;5(4):57-64.
24. National Cancer Institute (US), Division of Cancer Control and Population Sciences, Surveillance Implementation Group. Cancer surveillance implementation plan. Rockville (MD): National Cancer Institute, 1999 Mar.
25. Dean AG. Epi Info and Epi Map: current status and plans for Epi Info 2000. Journal of Public Health Management Practice 1999;5(4):54-7.
26. Health Care Financing Administration (US). Geomapping, outreach and health promotions support --- DHHS. PSA # 2334. Commerce Business Daily 1999 Apr 28.
27. Melnick A, Seigal N, Hildner J, Troxel T. Clackamas County Department of Human Services Community Health Mapping Engine (CHiME) geographic information systems project. Journal of Public Health Management Practice 1999; 5(2):64-69.
28. Rogers MY. Getting started with geographic information systems (GIS): a local health department perspective. Journal of Public Health Management Practice 1999; 5(4):22-23.
29. Baker EL, Melton RJ, Stange PV, Fields ML, Koplan JP, Guerra FA, Satcher D. Health reform and the health of the public. JAMA 1994;272(16):1276-1282.
30. Thrall GI. The future of GIS in public health management and practice. Journal of Public Health Management Practice 1999;5(4):75-82.
31. Davis RM. Measuring the performance of public health agencies. BMJ 1999; 318:889-90.