GIS/Atlas Cancer Research

Overview

Geographic Information Systems (GIS)
Overview of GIS and Cancer Since the first use of a geographic system to assess the location of childhood leukemia [Openshaw et al, 1987], GIS applications have been developed, utilized and continuously refined for the mapping and spatial analysis of cancer incidence, morbidity and mortality data. Four general approaches for utilizing GIS in cancer research have been taken: 1) disease mapping; 2) geographic correlation studies; 3) risk assessment in relation to a pre-specified point or line source; and 4) cluster detection and disease clustering [Wakefield & Elliott, 1999].

The National Cancer Institute (NCI) provides funding and maintenance support for GIS-related cancer research. A major initiative of NCI has been the development of a prototype GIS, the Geographic Information System for Health (GIS-H), as part of the Long Island Breast Cancer Study Project [National Cancer Institute, 2003]. GIS-H is a web-based resource and provides a broad and growing body of data to public and individual researchers, in addition to statistical and GIS tools to facilitate research. The GIS-H can be used to study various types of cancer and conditions. In a second major initiative, the NCI awarded 12 grants in Fiscal Years 2001 and 2002 on Geographic-based Research in Cancer Control and Epidemiology. These projects include investigations of the determinants of geographic patterns of cancer uncovered by the Atlas of Cancer Mortality in the United States, 1950-1994; geographic information systems; and methodological research to advance the use of GIS in understanding cancer etiology, including geo-visualization and spatial analysis of cancer data study by the submitting investigators [MacEachren et al, 2003].

Recently, the North American Association of Central Cancer Registries (NAACCR) has been developing a GIS handbook for central cancer registries. The handbook is expected to be released in July 2003 and will serve as a guide for registries as they choose to employ GIS.

What is GIS?
GIS.com has a nice, brief overview of GIS: http://www.gis.com/whatisgis/index.html.  US Geological Survey provides a more detailed explanation: http://erg.usgs.gov/isb/pubs/gis_poster/.

Why use GIS?
GIS.com has short article with examples: http://www.gis.com/whatisgis/whyusegis.html.

Cancer Atlas - Background
Previously developed atlases of geo-referenced health statistics have, in the past, led to insights concerning various health-environment-behavior interactions [Mason et al, 1975; Pickle et al, 1987; Pickle et al, 1990; Pickle et al, 1997; Devesa et al, 1999aMason, 1995]. Spatial associations identified have prompted hypotheses about the causal relations, some of which have been verified [Devesa et al, 1999]. Examples specifically related to cancer include identification of "hot spots" of esophageal cancer in China and oral cancer in the U.S. state of North Carolina [Winn et al, 1981].

Devesa and colleagues [Devesa et al, 1999a], in one of the most recent printed cancer atlases, cite a wide range of findings about cancer derived by correlation and field studies that were stimulated by the publication of previous atlases. They note, for example, that maps lead to studies indicating that "some evidence that farming and agricultural exposures may contribute to the geographic variation in prostate cancer, including high rates among whites in north central and northwestern areas, and among blacks in parts of the Southeast" and that "high colon cancer death rates in eastern Nebraska are linked to persons of Czechoslovakian background, and nutritional factors appear to contribute to the elevated risk".

NCI has produced six cancer mortality atlases since 1975. More recently, NCI has begun to disseminate maps, graphs, and other cancer statistics over the web with an online version of the most recent cancer atlas (http://www3.cancer.gov/atlasplus/). NCI recently launched Cancer Control PLANET (http://cancercontrolplanet.cancer.gov/) as a portal to the five steps in Comprehensive Cancer Control planning for state health departments. However, the initial launch of PLANET limited its use of GIS/Atlas methods. Drs. B. Sue Bell and Linda Pickle, members of our advisory committee, have been instrumental in NCI's various efforts to analyze, interpret, and present geotemporal cancer data.

Our Model GIS Atlas
The power of GIS for comprehensive cancer control, however, comes from the flexibility and extensibility of the digital environment. A digital, GIS-based Atlas has the potential to combine the ability of paper maps to prompt insight about spatial distributions and relationships with the ability of the digital environment to support exploratory analysis, hypothesis generation, statistical and computational testing of hypotheses, policy decision making, and dissemination of information in a variety of forms (many of which retain the dynamic characteristics of the underlying GIS).

The product we propose will integrate new data continuously and produce new outputs to meet particular needs. It will also provide a framework for extending the GIS functionality over time. We propose a series of research, development, and implementation tasks within this aim that will allow the development of an integrated, dynamic, extensible GIS-based Atlas. These tasks address aspects of: data compilation and transformation, map design and dynamic visualization, interface use and usability, developing spatial analysis methods, and extending spatial analysis capabilities of GIS. Finally, we propose to test the products and methods through both dissemination and to complete two case studies in hypothesis-generated epidemiological research on colorectal and prostate cancer.