—from Epidemiological Bulletin, Vol. 22 No. 3, September 2001—

SIGEpi: Geographic Information System in Epidemiology and Public Health

Introduction
Since 1995, in response to the health services needs of the countries of the Americas, the Special Program for Health Analysis (SHA) of the Pan American Health Organization (PAHO) developed a technical cooperation project, the purpose of which is the dissemination and use of Geographic Information Systems (GIS) as a tool for analysis and problem-solving in epidemiology and public health (1) (SIG-SP for its Spanish name).

Generating a series of activities, this project promotes development of low-cost computer systems, among them the software package SIGEpi. The SIGEpi package offers simplified tools and interfaces to efficiently carry out biostatistical and geographical analysis to support decision-making in public health.

Background information is presented here on the development of SIGEpi, its characteristics and general functions, as well as an example of how its analytical tools can be used. In this article, SIGEpi is applied to identify populations exposed to environmental risks in Mexico.

Background
Resulting from meetings, seminars, consultation workshops and requests made directly to the SIG-SP Project, some of the most common problems in the use of GIS in public health were defined as: high costs of commercial GIS software packages, making them inaccessible to the majority of users; insufficiency of epidemiological and public health analysis tools in GIS; and lack of integration between statistical and epidemiological programs and GIS.

To address such limitations, development components (a) from commercial programs were taken advantage of, particularly those handling cartographic data which allow the user to create products that can be distributed at a low cost and respond to the specifications and requirements proposed in the Project. With this consideration, SIGEpi was built based on ESRI’s (2) MapObjects (TM).

SIGEpi’s Beta version is currently used as an analytical tool for the surveillance and control of malaria in Brazil, and in a project to prevent the reintroduction of DDT for malaria control in Mexico and Central America.

Principal characteristics of SIGEpi
Developed for personal computers (PC) on the Windows platform, SIGEpi was designed following the conceptual elements and general systemic framework of the PAHO GIS (3) proposal.

SIGEpi’s graphic interface permits the management of multiple types of programs (from this point on they will be referred to as documents), each in independent windows. Typical documents consist of: Projects, Maps, Tables, Graphs, Results, and Presentations; each with its own functions, menus, buttons, and tools (see Figure 1).

Project controls all other documents and forms of data presentation, and guarantees that the current work session can be recovered in future sessions in the same state as was saved.

Data presentation and visualization are rendered through maps, tables and graphs. A dynamic link is maintained between them, allowing for simultaneous queries among the table, map, and graph registries. The Map document is the central document in a GIS, allowing the incorporation, manipulation, classification and visualization of cartographic data. The Tables document enables the presentation and handling of the databases’ cartographic layers and attributes, while the Graph document shows alternate representations of the map layers attribute data.

The “Results” section, visualizes in HTML format the results from processing and statistical analysis of data, producing data that can be managed in a word processor program or published on the Internet. The “Presentations” section prepares documents for high quality printing.

Functionality of SIGEpi
From a data management standpoint, SIGEpi follows an open approach and does not require the establishment of an a priori structure. This approach offers greater versatility in a framework where the user decides on the type of application and data necessary for its development, allowing the user to take advantage of data existing in other information systems.

Management of the digital maps is based on the vector model. It has the capacity to read and process files in Shapefile and ArcInfo coverage formats from ESRI; other formats include Vector Product Format (VPF) (.pft, .lat, .aft, .tft); CAD (.dwg, .dxf) and EpiMap (.bnd).

The SIGEpi system also can integrate different image formats and display them as a background image for a map. In addition, SIGEpi’s database management system handles MS Access’97 (.mdb) databases as the native format, and allows for the importing/exporting of data tables from other popular formats such as Excel, Dbase, Btrieve, EpiInfo, ASCII delimited text, etc. Database tables can be linked through an index to cartographic bases and overlaid on a map. Other operations and calculations also can be carried out for epidemiological analysis.

The design and selection of analytical procedures in SIGEpi are the product of a systematic and shared effort with the project’s collaborating groups, and other professionals and experts in public health. Following are the principal processing and analytical functions, according to their areas of application, offered by SIGEpi.

a. Basic functions and spatial data processing (geo-processing)
Data management and processing functions are: integration of attributes from data tables with digital cartographic bases (layers of spatial data), for visualization on a map through the superposition of multiple data layers; selection and querying of spatial data to generate new layers based on attributes and spatial operations between layers; geo-referencing or plotting points on a map from data tables with x, y coordinates; geo-processing operations such as the creation of catchment areas (“buffers’’) to delineate areas of impact or influence, and production of radial schemes (spider-diagrams) to measure linear distances between origin and destination.

Another essential function of SIGEpi is the creation of thematic maps, such as unique value or dot-density maps, bar and pie graphs, and intervals or ranges calculated with different classification methods.

b. Quantitative methods in Epidemiology
The functions include measures for quantitative analysis in epidemiology, which are particularly useful in exploratory data analysis. Among them are: descriptive statistics to calculate the set of measures of central tendency and scatter and prepare frequency distributions; correlation analysis; and both simple and multiple linear regression. Some functions for the calculation of rates, ratios, and proportions are included, as well as adjustment using direct and indirect methods (4) and spatial smoothing (5).

c. Useful methods for Public Health Practice
Some methods useful in analysis and decision-making processes in Public Health have been incorporated in SIGEpi. These include: identification of critical and priority areas; construction of a composite health index —such as basic unmet health needs or poverty - or identification and detection of spatial and time-space clusters6; measurement of the association between environmental exposure factors and health events for case-control or cohort epidemiological studies7; and evaluation methods for access to health services (based on the radial schemes technique), such as a simple measure of accessibility using linear origin-destination distances. An example of the use of SIGEpi in the area of environmental health is presented below.

Notes:
(a) Component: a block of programs that brings together a set of discrete funcions, operations, logic and user interface that can be used in the development of other programs.

References
(1) PAHO. PAHO Technical Cooperation in Geographic Information Systems Applied to Epidemiology (GIS-Epi) in the Americas. Epidemiological Bulletin. 1996; 17(2):8-10.
(2) ESRI. http://www.esri.com.
(3) PAHO/Special Program for Health Analysis. Geographic Information System in Health. Basic Concepts. Washington, DC.: PAHO, 2000: 92.
(4) Fleiss JL. Statistical Methods for Rates and Proportions. 2nd ed, New York John Wiley, 1981.
(5) Kafadar K. Smoothing Geographical Data, Particular Rates of Disease. Statistics in Medicine 1996; 15(23):2539-60.
(6) Pike MC, Smith PG. Diseases clustering: a generalization of Knox’s approach to the detection of space time clustering. Biometrics 1968; 24: 541-556.
(7) Schlesselman JJ. Case-Control Studies. Design, Conduct, Analysis. New York Oxford. Oxford University Press, 1982.

Source: Prepared by Eng. Ramon Martinez, Mr. Manuel Vidaurre, Geog. Patricia Najera, Dr. Enrique Loyola, Dr. Carlos Castillo-Salgado and Mr. Charles Eisner from PAHO’s Special Program for Health Analysis (SHA).

Return to Index
Epidemiological Bulletin, Vol. 22 No. 3, September 2001