WHO Report Benefits of GIS in Public Health

WHO Report Benefits of GIS in Public Health - WORLD HEALTH...

Info iconThis preview shows pages 1–21. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 2
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 4
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 6
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 8
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 10
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 12
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 14
Background image of page 15

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 16
Background image of page 17

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 18
Background image of page 19

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 20
Background image of page 21
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: WORLD HEALTH ORGANIZATION ORGANISATION MONDIALE DE LA SANTE CONSULTATION ON EPIDEMIOLOGICAL AND STATISTICAL METHODS OF RAPID HEALTH ASSESSMENT Geneva 26-29 November 1990 DISTFl. : LIMITED DISTR :LIMITEE ESM/CONS.RA/90.l3 ENGLISH ONLY THE BENEFITS OF THE APPLICATION OF GEOGRAPHICAL INFORMATION SYSTEMS IN PUBLIC AND ENVIRONMENTAL HEALTH Hank J. Soholtenl & Marion J.C. de Lepper2 1. National Institute of Public Health and Environmental Protection P.O. Box 1 3 '72 0 BA Bilthoven The Netherlands 2. Department of Regional Economics Free University Amsterdam This dowment I5 not Issued to the general DUDIIC, and all rlghts are reserved by the World Health Organization (WHO) The document may not be rewewed. abstracted quoted, reproduced or translated, to part or In whole, without the pnor wrutten permussron of WHO. No part of thus document may be stored In a retrieval system or transmitted in any form or by any means - electronic, mechanical or other wrthout the pram written DEFMISSIOH of WHO. The weWs expressed m documents by named authors are solely the responslballty of those authors. Ce document n’est pas destiné a farm distrrbué au grand public at tous les dronts y afférents sont réserves par l'OrganIsatuon mondlale de la Santé (OMS), II ne peut étre commenté, re’sumé, clté. reproduut ou tradmt. partiallement an en totallte. sans une autorisatlon oréalable écnte de I'OMS. Aucune name ne dolt étre chargée dans un systérne de recherche documen. tatre ou diffusee sous quelque forme ou par queloue moven oue ce sort - électromoue, mécamque. ou autre - sans une auto- rlsanon préalable Ecrlte de POMS. Les opamons exprlmées dans les documents par des auteurs crte‘s nommément n'engagent oue lesdlts auteurs. Contents 1 “40301 (20 Introduction The objectives of GiS Types of data storage Hardware and software In a GIS environment Types of information systems Appiications and users of GIS Applications of GIS In Health Research Conclusions References 1 introduction Health and ill~health always have and always will have a spatial dimension. Already more than a century ago, epidemiologists and other medical scientists began to explore the power of maps in helping the human mind to assimilate and understand the spatial dynamics of disease. One of the most tamous early users of maps in medical science was Dr. John Snow (18134858) - a London anathetist and Queen Victoria's obstetrician. Snow had the idea that cholera — the classic epidemic disease of the nineteenth century - might be spread by contaminated water supplies By using maps showing the geographical distribution of cholera deaths in the Soho area of London in 1854, he demonstrated that associations between cholera deaths and contaminated water supplies resulted in a striking geographical distribution. Nowadays, Aids is projected to be of critical importance in the remainder of this century. Consequently. attempts to forecast and predict its developments are urgently needed. Not surprisingly, Aids is increasingly subject of research. On our current state of knowledge. research on Aids has however to date failed to incorporate the spatial dimension of the disease. Although a vast amount of literature - describing many different aspects of the disease - has already been produced. the important geographical aspect seems to have been largely passed over. As Kabel (1990) points out: " Rarely can one find an attempt to model the spread of Aids incorporating the basic spatial dimensions of human existence. Most modelling seems to be focused completely within the temporal domain." Kabel describes a recently introduced methodology - called Spatial Adaptive Filtering - which enables the examination of the variability of Aids in time and space. One of Kabel's main lines of argument is that modelling the geographic distributions of Aids is useful toward both educational intervention and the planning of health care delivery systems. Mapping can play an important role in both areas as it is an excellent means of communication. in order to be of use to resource planners predictions of Aids should include a spatial component. These two examples describe the use of spatial information in health research. For most people using spatial information is the same as producing maps. whereby the map is used to show the geOgraphical distribution of some medical condition. Cliff and Haggett (1988) give a more than excellent review of the use of maps in health research and the problems and benefits. Based on their work a number of aspects related to the use of spatial information can be formulated. The most important point they make. is that map-making constitutes much more than presenting the geographical distribution of various diseases. Rather, it is a methodology that enables the handling of the spatial dimension of information. Another aspect related to map- making they bring fomrard is: "While maps are sometimes used to good effect, in other instances they tend to obscure evidence, to suggest false concentrations and to start false trails. The spatial distribution of diseases remains one of the oldest of puzzles and yet one of the most contemporary... they need to be handled with as much care and critical attention as any other source of evidence." The authors describe how, by combining cartographic and statistical arguments the value of maps can be sharpened and errors of interpretatioc reduced. Given the fact that health and ill-health contain a spatial dimension, it goes without saying that the science geography has always played a role in health research. Geographical work in this area known as medical geography embraces both the study of geographical variations in the provision of health care and the distribution of disease. Obviously. this includes the study of the relationships between health and environment, which main aim is to understand the relationships between the distribution and diffusion of diseases and the environment (eg. climate, vegetation, water and air quality). The problem is that it is almost impossible as a geographer to Specify cause and effect. "Nonetheless. knowledge about spatial variation in the incidence of disease is of significant value to the broader field of etiology and to medical science in general" {Groenewegen 21.0., 1987). From this it clearly follows that the handling of spatial information can be defined as a methodology with strong links to geography. cartography and statistical science. Needless to say, this methodology may be applied by scientists of all disciplines whose areas of interest contain a spatial component. At the same time that automation processes of industrial production took place. attempts to automate the handling of spatial information were introduced: significant advances have occurred during the last decade. In this article, we Wlll give an overview of the socalled Geographical Information Systems (GIS). These GIS can be deiined as automated information systems based on the methodology of geographic data handling (both spatial analysis and cartography) and the methodology of geographical database management and query. Hardware and software developments are briefly reviewed in SectiOn 4, types of spatial information system are outlined in Section 5 and the variety of application fields. organizations and users are discussed in Section 6 We begin, however, by clarifying the objectives of GIS and by outlining the baSlC types of data and data storage. In section 7 focus will be on the extent to which health research has succeeded in making uss of the possibilities oftered by GIS, that is the role that Geographical Information Systems have and could have played so far in studies focusing on health issues. 2 The objectives of GIS Geographical inlormation Systems represent a technology designed to acnieve particular objectives. In recent years, a variety of GIS products to assist the management and manipulation of spatial and non-spatial data have arrived on the market and users worldwide have begun to gain familiarity With these new systems. Experience suggests that there can be no doubt but that the application of GIS is making Significant contributions in facilitating the availability, integration and presentation of information A GlS can be defined as a set of tools for collecting, storing, retrievmg, analyzing and displaying spatial data. As a technology, a GIS is not necessarily limited to the confines of one independent system. lt may well have several components, each with a particular objective. A GIS must therefore accomplish the followmg four main tasks: Firstly, there is 3‘9 storage, management and integration of large amounts of spatially referenced data. A spatially referenced database contains two types of information, locational and attribute data (Figure l). Locational 0r spatial data are two or three dimensional co-ordinales of points (nodes), lines (segments) or areas (polygons). Non- locational or descriptive data. on the other hand, refer to the features or attributes of points. lines or areas. Data is obtained from a wide variety of sources and one of the most important features of GIS is the facility to integrate data, eg converting data values to a common spatial framework. The second main task of GIS is to enable spatial retrievals. These spatial query possrbllities are very flexible and still very powerful. It is possible to retrieve data for defined areas (polygons) like a municipality or postal code area. but there are also a number of possmilities that would appear to be extremely relevant for public and environmental health purposes (Openshaw and Charlton 1990), like: radial retrieval on a single point, radial retrieval for a search region defined as a buffer around a site's perimeter, retrieval of data lying in a corridor focused on a linear map feature (e.g. an overhead wire), retrieval of data for a by the user delined area, retrieval of data for search ares defined by manipulating existing digital map databases (e.g. relationship of cancer rates With geology patterns). The third main task of GIS is to provide methods that enable analyses which relate specifically to the geographic component of the data. The analysis techniques may be simple or more sophisticated. At the simplest level, for example. data about different spatial entities such as soil type (per square kilometre) and land use (by local administrative area) can be combined by overlay analysis. At an intermediate level. GlS may allow statistical calculations of the relationship between data sets to be computed or distances between entities may be used to determine the route that must be followed to move as quickly as possible from one location to another. The most sophisticated analysis occurs when modelling is introduced. In this context, there are a variety of analytical opportunities. it is possible, for example, to use atmospheric modelling techniques to discover which areas might be affected by pollution resulting from an explosron at a particular hazardous installation (e.g. Chernobyl), given certain wind and weather conditions. Alternatively. modelling methods can be used to determine the‘impact of locating a large public taCiIity (e.g. a hospital) at different sites in a City region. The fourth main task of GIS involves d:splaying data on map forms of high quality. Maps no longer have to be drawn by hand; they are an implicit product of all the work that is carried out within a GIS. However. for many different purposes, other forms of display (e.g. graphs and tables) may also be required, often for use in combination with maps. The RIA (RUimtelijke informatie via Automatisering) system, developed by the National Phys:cal Planning Agency in the Netherlands (Scholten and Meijer 1990) is an good exampte of alternative methods of displaying spatial information in an interactive and user-friendly way. Figure 1: Locational and attribute information 3 Types of data storage The distinction has already been drawn between locational and attribute data. it is important to distinguish further between three forms in which locational data can be incorporated within a GIS: raster. vector and quadtree storage (Figure 2). Raster or grid storage This form of storage tor locational data involves a regular grid of cells being laid over an area. Attribute data are collected for each grid cell which may measure for example 500 by 500 square meters. This means that the whole area is covered by a group of cells each at which has an attribute value. Within a grid-oriented system of this type, it is often the case that only limited use is made of the attribute data. Satellite photographs, in which a considerably smaller grid Size is used, provide raster information. In a satellite photograph, a single value is attached to each cell. in this way. tactual data can be collected in a very efficient way. Curran (1985) indicates how remote sensing data is frequently contained in raster based GlS. Vector storage _ The storage of locational data in vectors gives a very precise representation ol reality. in this way. pornts, lines and areas are incorporated as Cartesian co-ordinates in the computer. Whilst lines can only be represented as a series at cells in a raster structure. in the vector storage method, the exacr middle paint of a line can be identified. An area is represented by a group of cells With an angular boundary in the raster structure, whereas the precise boundary at an area can be included when a vector structure is used. This precise storage oi x- and y-coordinates usually generates larger data sets. Within a vector system. the relationship between locational and attribute data is of great importance. Each element is related to a record in the database with the same, unique identification number. Quadrree storage The quadtree storage method falls between the grid and vector storage methods. In this method, the data are stored in grid cells of variable size. Within larger homogeneous areas. a large cell size is used whereas towards the edges of the area, the cell size diminishes to form a precise picture. An area is therefore covered by considerably lewer quadtree cells of varying size than is the case in the regular grid storage method. The speed with which analyses can be carried out with a quadtree structure is high, whilst the original precision of the data is retained rather well. Figure 2 The three forms of data storage 4 Hardware and software in a GIS environment The initial GIS products date from the 19705. a decade in which the typical hardware configuration comprised a central computer surrounded by memory and storage disks and a number of peripheral dewces. Time share systems enabled a large number of terminals with lines attached to the mainframe to be used at the same time. At the beginning of the 19803. this centralized approach was extended by connecting minioomputers to the central mainframe in order to carry out certain processes. This was the period in which GIS came of age. Very large databases began to be assembled and the need for processor capacity increased enormously. In the middle of the decade, the personal computer (PC) arrived, although its impact for GIS meant little more than an extra terminal in many cases. Nevertheless, it goes without saying that the PC has become central to the popularization of GIS. In the second hall of the 19805, the PC played an important role in simple GIS tasks such as the automatic production of maps. It is exactly this function that has brought GlS to the attention of many people, whilst the basic concept of a common central database tends to be forgotten. Attempts to transter luily-fledged GIS irom the maintrame onto the PC have been commercially successful. even though in many cases their perlormance leaves something to be desired. Of much greater importance in the mid-19805 was the further development of minicomputers and work stations connected to a network. At this time. the larger organizations making use of GIS, realized that the central maintrame option was not the solution to a number of GlS tasks (Figure 3). it was recognized that each separate task required its own processor capacrty or its own working environment. Throughout the 19805. it became clear that hardware vendors were meeting these demands perfectly well by means of further optimization of minicomputers using servers and the increased processor capacity ot work stations. and that hardware costs were decreasing significantly. Identification of the main tasks oi a GIS enVIronment allows us to specify a corresponding set of software demands. Database management software One of the most significant advances in GIS software development came at the end of the 19705 with the introduction of the concept of the relational database. In the relational model, different data sets are linked together by the use oi common key fields. For example. attribute data available for two different sets of Spatial units (areas) wrll require a third sat of information to show how the two spatial bases fit together. This type of structure can also be used to construct spatial databases in which lines are linked together to represent polygons. The creation, maintenance and accessmg ot a database requires a Data Base Management System (DBMS). In order to handle very large quantities of iniormation. a relational DBMS is a necessrty tor most GIS applications (see Frank 1988). It also serves the very important tunction of separating the data user from the technicalities of the computing system and facflitates data manipulation and analysis. Many of the proprietary GIS have their own systems to handle basic storage. management and analysrs operations (e.g. iNFO is the relational DBMS of ARC/INFO: ORACLE is the relational DBMS ot ARGlS). Analysis software The software required to perform certain analytical tasks varies according to the nature of the problem. the quantity of information available and the objectives of the organizations involved. Figure 3: Decentralized GIS tasks A variety of analytical tools are now available within GIS. The overlay procedure, for example, has been widely used for combining different data sets in order to identify areas or Sites With reqwred characteristics (see, for example, Dangermond 1983). Buffering, address matching and network analysis are additional tools adopted in planning applications of GIS. However, the development of analytical functionality within GIS has tended to neglect the important benefits that modelling procedures can contribute through data transformation, integration and updating; simulation; optimization, impact assessment; and forecasting (see: Birkin et al., 1987; Openshaw, 1990). The specialized and complex nature of modelling algorithms and the restrictions imposed by processor capacny have both been influential in keeping the modelling component separate. One of the challenges confronting the next generation of GIS is to improve the integration of modelling and GIS so as to provide researchers, decision-makers and planners with enhanced model based decasion support systems. Access and presentation software One of the most important functions of GIS development is frequently the provasion of access to information by a variety of different users. In the context of planning, individuals in different departments of the same organization (Public Works. Social Services, Transport, Parks, etc.) may require access to the same database. Similarly, users in different national, regional and local organizations may wish to access the database simultaneously. The National On-line Manpower information System (NOMIS) in the United Kingdom IS a good example of a GIS which stores up-to-date information about employment, unemployment, job vacancies, migration and population, and enables users at remote Sites to obtain raw counts, tables, graphs and maps of data at a variety of spatial scales (Townsend et at 1986. 1987). The technological advances in computer hardware which have been occurring over the past 20 years have had a direct impact on the presentation of information in GIS. There have been striking advances in automated mapping, is. the development of programs enabling maps to be produced automatically (see: Croswell and Clark, 1988). The experience gained over this period has meant that very professional products are now the norm, and micro-based mapping software (ATLAS‘Graphics, GIMMS, MiCFlOMAP, for example) provides high quality output. 5 Types of information systems A good deal of discussron has taken place concerning the definition of a geographic information system. The GIS literature contains many terms that are used as synonyms for GIS which include spatial information system: gee-data system; geographic data system; land use information sysiem (Clarke, 1986; Parker, 1988). This plethora of meanings arises in part because GIS is a very young selence and has important relationships with other physical and social science disciplines that involve handling spatial data. These include remote sensing, photogrammetry, cartography, surveying, geodesy, environmental scrence, regional science, planning and, of course. geography. The definition of GIS, given earlier in this article. is therefore one ot a number of definitions that appear in the literature (see e.g.: Burrough, 1986; Marble et al., 1983; Calkins and Tomlinson, 1984: Berry. 1986). In this discussion, for convenience, we use a rather rigid classification framework to illustrate the important differences between spatial information systems which fall under the GIS umbrella. Figure 4 gives an overview oi the three main groups of spatial information systems that may be distinguished. Computer Aided Design (CAD) systems CAD systems are graphics systems which are used by industrial designers. architects and landscape architects to support and display their work. CAD has replaced the drawing board. Whilst early CAD systems were purely automated drawing systems. later packages have provided enhanced facilities for qualitative and quantitative design analysis as well as database facilities in which information can be stored and in which a large number of symbols can be used. CAD systems allow the possmility of automated drawing, the manipulation ot drawings (changes of scale, location, zooming in, rotating and editing) and the presentation of this information in a professional format. Graphics software development has been under considerable influence from the world of CAD, and GlS presentation software has tried to incorporate CAD features. The developments in GlS have tended to lag behlnd those taking place in the CAD envrronment it is important to recognize that the basic concepts in the two worlds are different. 618 IS totally involved With the concept of the database whereas CAD is more concerned with the deSlgn process and the accompanying use of symbols. Both the hardware and software in the CAD environment are focused on presentation, whereas this principle applies less within the framework of GlS. Automated Mapping (AM) was developed initially as another application of computer graphics technology alongsrde CAD. However. AM was extended to permit the storage and retrieval of associated locational and attribute data linked to the graphics. thus creating Automated MappinglFacrlities Management (AM/FM) for use as a speCialist application in the management of utilities. Figure 4: Spatial information systems Land use Information Systems (LIS) The objective of this type of information system is to function as an administrative system for the management of geographic data on land use and in this sense. LIS is in several ways similar to AMIFM. Many different demands are often made of LIS which have a direct influence on the way in which data should be stored. In the example of real estate information, concern over the legal obligations related to the precision of data may be of paramount importance. However, with regard to information on pipelines or cables. tor instance. there may be less wony about legal matters but much more concern that the exact location of a plpe or cable can be established quickly. In this type of information system. the central focus Is the development of a very detailed database. Moreover there are tools available in this type of system which allow data with a very high degree of accuracy (double precision) to be stored, managed, integrated, updated and displayed. Relatively few spatial or geographical analyses are carried out Within this type of system. Geographical Information Systems (GIS) Whilst us provide powerful tools for local planning authorities and public infrastructure agencies operating at a very detailed, micro scale, GIS tend to suppon analysis, planning and evaluation at a more macro scale. They are used in a variety of contexts to assist the research required to formulate and evaluate central and local government policy With respect to different aspects of physical and environmental planning on the one hand, and economic or strategic planning on the other. Such information Systems adopt both a database approach and a set of tools to assist data coilection, management, updating and presentation The distinguishing feature of GIS in comparison with US and CAD is the availability in GIS of the spatial analytical tools that have been discussed in a previous section. These three categories of spatial information system are certainly not mutually exclusive as far as data are concerned. Many public utilities and local authorities have, for example, begun to use CAD systems to illustrate their information because CAD systems fulfill the demands for precision. However, the problem that confronts this group of users is the addition of information and its relationship to assoCiated attribute information. A local authority or municipality may input its inventory of street lamps into a CAD system which possesses a number of qualities which are indispensable in the desrgn process. for example. However, it is clear that this information may be related to other land use information and thus the integrated use of the information requires a database approach for which a LIS or GIS is required. 6 Applications and users of GIS Despite efforts to distinguish different forms of spatial information systems, it is commonplace to use the term GIS to refer to all types of Spatial information systems. Applications of GIS are many and varied tsee Department of Envrronment 1987). The following lists exemplify some of the fields of application of GiS which may help to confirm the distinguishing features of categories of information systems that have been identified. it) CAD and automated cartography - ciVIl engineering - construction - architecture - landscape architecture (ii) LIS and facility management - public utilities (water, gas, electricity. telephone) - real estate (land and property) - management of infrastructure (roads, raiiways, water supply. etc), housing, listed buildings, industry (iii) GIS - traffic and transport planning - agricultural planning - environment and natural resource management - recreation planning - location/allocation decisions - spatial planning (land use) - service planning (educatiOn, police, health, social, etc.) - marketing - public and enVIronmental health A number of categories of users can be identified on the basis of the objectives of the organization to which these users belong. A number of types of organizations can be classified which differ from eacnother according to the type of activity that each performs Four main types of organization may be distinguished. Firstly, there is the research institute where research may be carried out to find solutions to problems or answers to questions posed by external paymasters. Data collection and manipulation takes place and descriptive. explanatory and predictive analyses are undertaken Secondly. there are administrative institutions such as public utilities or property registration agencies. Here. the obiective is to manage information in such a way that the process of acquiring and manipulating data is made as simple as possible. The management of waste disposal system of sewerage pipes for a local administrative area is one example where accurate and quick answers are requrred on the basis of the information stored in the GIS to questions such as: 'Where are the oldest parts of the system?'; 'What IS the total length of pipe involved?’; 'At what depth are the pipes buried?‘: and ’How many houses are connected to these pipes?'. The third type of organization is the government agency whose objective is to formulate policy recommendations. For this purpose, concept design and evaluation takes place. Commercial enterprises are the fourth type of organization. Their aim is to maximize their profits by sailing goods and services. information is collected and manipulated within an integrated modelling/GIS envnonment, for example to establish optimum locations for new retail outlets. The type of GIS that is adopted and applied therefore varies between each category of organization and between organizations of different size and function Within the same category. However, it is possible to identify particular groups of inditnduals across the spectrum of organizations whose occupational characteristics with respect to GIS are distinctive: information specialists, researchers, research coordinators, poliCy preparers, decision makers and third parties In each of the four categories of organization, information speCIalists are required to acquire and manage data. computer hardware and software. The information specialist usually works with the raw data and requires a large GIS (e.g. ARC/INFO, ARGIS. SYSTEMS) which is flexible and able to be connected to other systems. Researchers, on the other hand, tend to be confined to their own institutes or to commercial companies. They work either with raw data or data that has been partially processed or transformed. They demand user-friendliness from a GIS, analytical features (as with SPANS, for example) and appropriate interfaces that allow transfer of information to other packages for modelling and other purposes. Research coordinators are concerned with the interrelationship between the different prooucts of the organization and therefore work with manipulated data and require a simple, user-friendly GIS. Policy formulation is usually the responsrbility of a government agency and the main requirement of a GIS in this context is that it should be easy to use. The same applies to the decision makers in administrative. government and commercial organizations, whose job it is to translate information into policy statements, and to third parties, who simply utilize information provided by government agenCies or research institutes. The successful implementation of a GIS therefore depends on a careful preliminary assessment of the type of GIS required to meet the demands to be satisfied by the various users of the system. Ideally, there needs to be a detailed functional specification which outlines the needs that a GIS has to meet. Whilst GIS packages are now available for the range of GIS users that we have identified, there remain cons.derable difficulties, particularly in large organizations. in establishing the configuration of hardware and software and in enabling the easy exchange of information between the machines and packages involved. it is important to ensure that there is a GIS management framework which takes into account the separation of functions between user departments and the central data store. Figure 5 shows the different user groups and their varying demands. Figure 5: User groups and their demands 7. The Application of GIS in Health Research In the foregoing sections of this article. different aspects that characterize Geographical Information Systems as well as its possible users and applications have been discussed. Major themes brought forward included the various functions a GIS can perform, i.e. the automated collection/capture, storage. checking, integration, manipulation, retrieval, analysis and display of data, both geographical and attribute data. In the present section emphasis will be on applications of GIS in health research, an area which has proved to lend itself to a GIS approach. By briefly describing some case-studies. here focus will be on the extent to which health and health-related research has succeeded in making use of the possibilities offered by GIS. To put it differently, the question here being addressed IS what role Geographical Information Systems have and could have played so far in studies focusing on health issues. There is a large number of applications available which may fall under a GIS umbrella. but in most cases it is questionable whether or not these applications constitute GIS or rather still the "old-fashioned" hand-made maps Automated mapping is the most well-known field of GIS in health research at present. In loregomg years a large number of applications of atlases has been produced. A complete. geographical, review is given in Cliff and Haggett (1988). Figure 6 shows an example of an automated map (Bopp. 1989). These atlases show how interesting patterns of spatial distribution of health and ill- healfh can often be revealed srmply by mapping. However, it goes without saying that it is riot sufficrent to stop at that point in epidemiological research. Once the characteristics of a mapped distribution of a disease are described, questions as to why a certain geographical distribution of a disease 15 prevalent will raise. In this sense therefore, maps may provide an indication as to where on the map further research may be useful. Interpreting spatial patterns and looking for relationships lies within the domain of spatial analysis. Spatial analysis embraces GIS functions with which computer mapping is not equipped to deal. as has become clear in the case study on childhood cancer in Northern England. which was set up as a result of the currently considerable interest with regard to cancer clusters and their possible association with envrronmental factors. Figure 6: Example 01 an automated map 13 In an attempt to search tor the relationships between geographical correlates of children With cancer in the Northern region of England, a set of spatial analysis methods has recently been integrated and automated - the socalled Geographical Analysis Machine (Openshaw. Charlton, Craft & Birch, 1988). One impetus to the development of GAM has been that prevrous epidemiological attempts for finding eVidence oi cancer clusters provided uncertain and spurious results of the data. by virtue oi the arbitrary geographic units used tor analysis such as districts or wards. GAM provides a baundary-free method and is an attempt to move away from arbitrary discretations of geographic space and represents a method that uses both point and area iniormation. Area intormation (based on Enumeration Districts - these are the smallest building blocks for which Census data are publicly released - as so called ‘Small Area Health Statistics' - in the United Kingdom) is used to ascertain the number of children "at risk", and random selections are made from these data in order to determine expected point distribution of cases. The link from areas to points is made via the ED centroids, grid references accurate to 100 meters, prowded for each areal unit. The method then counts the observed number of cases falling Within a circle oi given size and location and compares this with the expected value. It the count significantly exceeds the expectation then the circle is plotted on the map. Multiple overlapping Circles arise because the potential centres of circles are locations on a fine grid and because a wide range oi circle sizes are considered. Clearly, this method provades the possibility ‘o distingmsh ‘real' from 'illusory' clusters USing the GAM method in the case of Nonhem England. some striking results were tound. one oi them being that a cluster was spotted that other techniques prevrously had tailed to iind. This cluster lacked the prewously favored causative lactor ot low-level radiation discharge from a nuclear installation. Another cluster oi cases was found near the famous Sellatield power station (this one had also been detected by techniques prewously used by others.) The new identified cluster proved that not only cancer incidences vary geographically, but it also gave reason to believe that associations may exist between various aspecrs of physical and human enwronment other than those previously hypothesized. Although it has not yet been undertaken, a follow-up on the testing oi the existence of the localized aggregations of disease by means of GAM might be to extend the research by using a GIS framework which would allow for a more intenswe search for possible multivariate associations via map-based analySis. That a GIS approach certainly otters benefits in epidemiological research may further be illustrated by a case study on analyzing cancer of the larynx in North West England (Gatrell and Dunn, 1990) GIS tumed out to improve the possrbilities of descriptive mapping of disease. espeCially Since large data sets such as those from national and regional cancer registries were used. The ability to execute multiple queries on a data set, together with the use of high resolution graphics, had greatly facilitated manipulation of the data, such that subsets of cases by cancer Site, sex and age could be readily extracted and analyzed. Apart from that, GIS ippeared very helpful in attempting to assess relationships of disease to environmental variables. Especially the facilities offered by GIS enabling users to define their own areas of interest (areas around incinerators) appeared very valuable. In the study results of GIS operations such as 'butiering’ and 'poiygon overlay' were used as inputs tor statistical modelling of spatial variations in the incidence of diseases. in this case cancers of the larynx. Public and political concern about possible links between the emission of dioxin trom commercial and municipal incinerators and human ill-health has been the main incentive to this epidemiological study. Earlier studies by the same North West Regional Research Laboratory in Lancaster. United Kingdom. had suggested a possmle link between cancer oi the larynx and proximity to an industrial waste incinerator. The empirical study taken up consequently tried to assess whether there is any assomation between the geographical distribution of laryngeal cancer and proxrmity to an industrial waste incrnerator and included testing the hypotheSIs that cancer incidence is significantly higher around incmerators than elsewhere. in tirst instance the study was restricted to one of the 19 District Health Authorities that constitute the Regional Health Authority - where an industrial incrnerator that had operated in the District for several years had been closed. Data on cancers registered between 1974 and 1983 were used. First oi all, some simple descriptive mapping was undertaken to see it there was any Visual evidence of localized clustering of cases in the vicinity of the incinerator. Such mapping was made possmle by converting the unit postcodes oi cases to Ordnance Survey Grids, accurate to 100 meters; the computerized Central Postcode Directory (a directory wherein unit postcodes are associated with an Ordnance Survey grid reference) was used for this purpose. Point distribution maps showed that the distribution of most cancers mirrored the distribution of population as a whole. The exception to this was laryngeal cancer. Here, out of only 58 cases found. five were Within about two kilometres of the closed industrial incinerator. No conclusions could be drawn from this, but the mapping did provide a first indication on where turther investigation was needed. The results of the work within the one District - which came down to the assumption that indeed proximity to incmerators and the inCIdence of cancer of the larynx are interrelated - lead to a Similar second study in a larger geographical area. Now, much larger data sets were used, namely on cancer of the larynx of the entire North West Regional Health Authority. The cancer data, including age, sex and cancer site, year of notification, unit postcode and 0.8. grid reference, were imponed into a GIS whicn uses both vector and raster types of data storage as well as various spatial analysis techniques, together with digitised electoral ward boundaries and data from the 1981 Population Census. The latter included the age~sex data needed to compute expected incidence of cancers. A GIS operation called 'poini-in-poiygon-overiay‘ on the point coverage of cancers and the ward polygons allowed to attach a ward code to each case. Observed number of cases, by sex, cancer Site and ward, were then counted by a Database Management System (dbms) software program. Expected occurrences per ward were calculated by applying lung and laryngeal cancer rates for the North Wesr Regional Health Authority, based on sex and ten-year age groups. to the age-sex disaggregated population in each ward. Standardised Registration Ratios (SFiFl's) - i.e. measures of relative risk - of cancer of the larynx were determined and those wards with Significant SF‘iFi‘s were displayed. After having done all this, the step that followed was most interesting. The GIS was used to define areas possibly at risk from pollution by hospital incinerators. Areas were defined to be circles of two kilometres radius and were located around 44 hospitals as well as around 44 at random chosen control sites. For both the set of hospital circles and the set of control circles a number of operations were then performed. Each set of circles was overlain separately with both the full point coverage of observed cases and the polygon coverage of ward boundaries. Again, point-in-polygon operations gave a count by area (circle). The second overlay clipped out those wards or parts of wards which each Circle intersects. The GIS automatically calculates the area of each ward which falls within the Circle boundary. An estimate of the expected number oi cases was then found by multiplying the proportion of each ward area that lied within the circle by the expected value for the whole ward: summing those values prowded the number of expected cases for each circle. SRR's were then determined and displayed. The results of the GIS operations were consequently put into a statistical model of cancer incidence based on the theOry of spatial point processes. Briefly, it entailed setting up a formal statistical model which states that the intensity of cases of the larynx is a function first of 'background' intensity (population at risk) and, second. of distance from the putative source of pollution. The model tests whether proximity to the pollution source is a significant influence on the distribution of this cancer. This method clearly enables separating out "point hazard“ effects from the "natural" clustering we expect due to population distribution. Although the number of applications of GIS in health research is growing very fast, at present GIS is still a working environment which has to be introduced further. To give an example of a geographical study which could have benefited from GlS we will briefly elaborate on a study on Chernobyl fall-out and perinatal mortality in England and Wales (Bentham, 1990). Due to other studies that had concluded that radioactive fall-out from Chernobyl (26 April 1986) may have caused an increase in perinatal mortality in West Germany and the USA and due to the tact that marked geographical variations in contamination from Chernobyl in England and Wales seemed to exist, a geographical study further investigating this matter has been taken up in the United Kingdom. The search for evidence for the existence of a relationship between an increase in perinatal mortality and radiation was the focus of attention. Whether increases in perinatal monality rates in the period following Chernobyl were relatively higher in regions with higher radiation doses than elsewhere was the main research question. The highest doses radioactive fall-out from Chernobyl in England and Wales were in the counties of Cumbria, Clwyd and Gwynedd where there was heavy raintall during the passage of the radioactive cloud. Observed and expected numbers of stillbirth. neonatal deaths and perinatal deaths in the three counties in the year following Chernobyl were calculated. Expected numbers were calculated from national rates multiplied by the ratio of rates in the three counties to national rates during the years 1981 to 1985. Mortality data of these counties were adjusted to national rates so that the hypothesized increase in mortality rates would not be falsely accepted due to the tact that the regions studied retlected any general tendency for mortality rates to be different from national rates. Evidence for an effect of Chernobyl would be it increases in perinatal mortality had been greater in regions with higher radiation doses and less elsewhere, but no such evidence was found. In fact. it was observed that perinatal mortality in these areas did not rise relative to the national average in the year following Chernobyl. The significant deficit of perinatal mortality in Cumbria, Clwyd and Gwynedd raised the question whether this was a particular local event or whether it had wider significance. Consequently. a geographical study of a wider range of areas was undertaken. Since deposition may be a poor gurde to radiation doses received by the population, consumption of contaminated milk was taken into account to investigate the pattern of perinatal mortality in the year following Chernobyl for any relationship with geographical radioactive deposition. Estimates of contaminated milk were provided for and observed and expected numbers of perinatal deaths were derived by the same methods used in the analysis of Cumbria, Clwyd and Gwynedd. But the negative finding ot the earlier study was confirmed: there is no evidence that radiation from Chernobyl caused a rise in perinatal mortality in England and Wales. It is important to stress here that although the Chernobyl study did not actually make use of (318 it prowdes a good example of how GIS can be helpful in such a study. Several aspects of the study favor using a GIS methodology. Some examples of the operations that (318 could have performed in this case are the followmg: - Large sets of spatially referenced data were used which within a GIS can be quickly and easily handled. integrated and manipulated. - Also some spatial analysis functions that a GIS can perform could have been used. At a very Simple level data about different spatial entities such as perinatal mortality rates by local administrative area and fall-out doses per user-defined area can be combined by overlay analysis. - Lastly, map-making might have been useful because a map is worth a thousand words and constitutes therefore an excellent displaying device. The most important conclusion to be drawn and lesson to be learned from the above exemplary case studies is that there is no doubt but that applying a GIS approach in health research can be of great value. Not only offers GlS possrbilities to perform various tasks fundamental in such research far more qutckly and with less effort, it also provides health researchers with new reliable and scientifically valid methods for handling their spatial information. By using GlS. functions can be performed with which automated statistical and/or computer mapping packages are not equrpped to deal. In this sense. GlS may lead the way to "new" insights into "old" information and may by this even contribute to a better understanding of the health problems we nowadays face. 8 Conclusions One of the most important questions arising in public and environmental health concerns the type of instruments that can be utilized to devise quick. reliable and scientifically valid methods of rapid assessment. to be of help in health research and the planning, monitoring and evaluation of health programmes. Measures to improve health and the quality of life are now receiving greater attention along with the need to protect and improve the environment. New techniques must therefore enable public authorities to gain insights into the consequences of decisions relating to investment in environmental management and public health. It is also necessary to provrde capacities to review the current situation in which new developments are proposed. To get more grip on all kind of public health problems, it must be possible for research institutes to have access to different data sources. in order to achieve these objectives. information is vital for effective guidance in this rapidly changing context. All relevant information must be stored. managed. made available and presented in a suitable form for use at different stages in the research and planning process. The information may be of various types. extensrve in quantity. variable in quality and referring to areal units of different size. In our opinion Geographical information Systems can provide excellent frameworks within which these activities can be undertaken because the main tasks of GlS are: - The management. integration and (spatial) query. There is no discussion posSible that at this moment GIS offers the most adequate solutions for database management, data integration and spatial query. Openshaw, Charlton et al. (1990) demonstrated how powerful a GIS environment is in comparison to a normal (relational) database approach. - The analytical possibilities. (318 offers an enVironment with different spatial analysis tools. which give all kind of opportunities for questions related to the combination of information based on different geographical scales. There are tools for the exploration of the data. the search for spatial patterns, and also possibilities for modelling (e.g. interaction models. diffusion models). - The mapping possibilities. Making maps will be rather simple by using (318. it is not surprising that large organizations try to bundle their efforts when dealing with enviromental and health policies and research. Environmental and health problems - which are diverse and complex a cross national boundaries and many need to be dealt with internationally. Besrdes the implementation of a GIS itself takes a lot of effort: the main issues being how to bring together different data sources. how to improve the knowledge of spatial analysis. and how to handle issues such as data accuracy. data errors. legal aspects of data and the like. One of the first examples of European initiatives to bundle efforts was the COFllNE project, which was set up by the European Community in 1985. The main impetus to the CORINE programme was the need to gather. coordinate and ensure the consistency of information on the state of the environment and natural resources. The Programme had two aims. namely: - to verify the usefulness of a permanent information system on the state of the environment for Community environmental policy. to check the technical feasibility of creating such a system. and to identify the conditions required for its installation and functioning: - to supply information useful for Community environmental policy. Concerning the first aim. the CORINE programme results show that a permanent information system on the state of the Community environment is necessary and technically feasible. The second of the programme's aim has also been successfully attained. Data on the priority topics were collected, supplemented by a series of basic data. and organized in an operational (318. Therefore the Council of Europe Ministers took the decision to transform the CORINE prototype into a permanent information system. The COFlINE experiences and results have encouraged other European initiatives aimed at the ioining of efforts with regard to environmental and health problems. The Warld Health Organization Regional Office for Europe decided in 1989 to start the HEGIS project. a permanent geographical information system for public health and environment. based on the .. .— cooperaiion of national focal points, collaborating WHO-centres and research institutes (see Appendix 1 for more information). it goes without saying that GIS will not solve any of our daily problems, these we ourselfs will have to solve. Nevenheless, to our pom! of view GlS offers a perlect werking environment to describe. to Investigate. to monitor or to iorecast. References Bentham, Graham (1990). Chernobyl Fall-out and Perinatal Mortality in England and Wales. in: Bentham. G., Haynes, R. a.o. (eds). 1990, Proceedings Fourth International Symposrum in Medical Geography, Universnty of East Anglia. Nonmch Berry, J.K. (1986) Learning computer-assisted map analysrs. Journal of Forestry, October. 39-43 Beurden, A.U.C.J. van and H.J. Scholten. The environmental geographical information systems of the Netherlands and its organizational implications. in: Harts. J.J.. H.F.L. Ottens and H.J. Schoiten (eds), 1990, EGIS'QO Proceedings, EGIS Feundation. Utrecht Birktn, M., Clarke. G.P.. Clarke, M. and Wilson, A.G. (1987) Geographical information systems and model-based locational analysis: ships in the ntght or the beginnings of a relationship?, Working Paper 498. School of Geography, University of Leeds Bopp. M. (1989). Kanographie als instrument in der epidemiologie: methodologische aberlegungen ft'ir den neuen Schwetzer Krebsatlas. in: 802 Praeventivmed 34 pp. 108-114 Burrough, PA. (1986) Princupies oi Geographical Information Systems for Land Resources Assessment, Monographs on Soul and Resources Survey 12. Clarendon Press Calkins. H.W. and Tomlinson, RF. (1984) Basic Readings in Geographic Information Systems, SPAD Systems Ltd, Williamswtle. New York Clarke. K.C. (1986) Recent trends in geographic information system research. Geo-Processmg, 3, 1-5 Cliff, AD. and P Haggett (1988), Atlas of Disease Distribution. analytic approaches of epidemiological data, Oxford. Basil Blackwell Ltd Cross, Anna F. (1990), Using a Geographical Information System to explore the spatial incidence of childhood cancer tn Northern England, in. Harts. J.J., H.F.L. Ottens and H.J. Scholten (eds), 1990. EGIS'90 Proceedings. EGIS Foundation. Utrecht Crossweil, PL. and Clark. SR. (1988) Trends in automated mapping and geographic information system hardware, Photogrammetnc Engineering and Remote Sensing, 54(11), 1571-1576 Curran. P.J. (1985) Principles of Remote Sensing, Longman Dangermond, .J. (1983) Selecting new town sites in the United States using regional databases, in Teicholz, E. and Berry, B.J.L. (eds) Computer Graphics and Environmental Planning, Prentice Hall lnc., Englewood Cliffs Department of Environment (1987) Handling Geographic information. HMSO. London Frank, AV. (1988) Requirements for a database management system for a GIS. Photogrammetnc Engineering and Remote Sensing. 54(11), 1557-1554 Gatreil, A. & C. Dunn (1990), GIS in Epidemiological Research: Analyzing Cancer of the Larynx in North West England. in: Harts. J.J.. H.F.L. Ottens and H.J. Scholten (eds), 1990, EGIS'90 Proceedings, EGIS Foundation. Utrecht Groenewegen, PP, J.P. Mackenbach and M.H. Stijnenbosch (1987), Geography of Health and Health Care in the Netherlands. in: Groenewegen, P.P. J.P. Mackenbach and M.H. Stiinenbosch (eds), Nederlandse Geografische Studies. 34, 37-55; Utrecht Kabel, R. (1990), Predicting the next map wrih spatial adaptive filtering. in: Bentham. (3.. Haynes. R. as. (eds), 1990, Proceedings Fourth International Symposrurn in Medical Geography, Universrty of East Anglia, Norwrch Marble. D.F. and Peuquet, D.J. (1983) Geographic information systems and remote sensrng. Manual of Remote sensing. 2nd Edition, Amencan Society tor Photogramrnetry and Remote Sensing, Falls Church, Virginia Openshaw, 8.. M. Charlton, A.W. Craft & J.M. Birch (1988} Investigation of Leukaemia clusters by use of a Geographical Analysis Machine. In: The Lancet, Febr. 6, 1988 Parker, HD. (1988) The unique qualities of a geographic information system: a commentary, Photogrammetric Engineering and Remote Sensing, 54(11), 1547-1549 Scholten, H.J. and Melier, E. (1990) From GIS to RIA: a user-friendly microcomputer- orientated regional information system for bridging the gap between researcher and user, in: Ekistics, 58, 289-244 Townsend. A., Blakemore, ML, Nelson. R and Dodds. P. {1987) The NOMIS database: availability and uses for geographers, Area, 19. 43-50 -21.. Geographical Infomation point line polygon grid Figure 2: The three forms of data storage VECTOR ...
View Full Document

This note was uploaded on 01/21/2012 for the course HUMBIO 156 taught by Professor Katzenstein,d during the Fall '11 term at Stanford.

Page1 / 21

WHO Report Benefits of GIS in Public Health - WORLD HEALTH...

This preview shows document pages 1 - 21. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online