A Spatial Decision Support System for the Lubombo SDI
Malaria control, and development in general, has a spatial component which is ideally suited to be supported by a Geographical Information System (GIS) based Decision Support System.
A GIS provides a mechanism to integrate different data sets, analyse their spatial and statistical components, and model possible scenarios to provide a planning and management tool. Its effective use requires appropriate data, mechanisms for analysis and distribution, as well as expertise and technical support.
The Innovation Fund funded a project by the Malaria Research Programme of the Medical Research Council to develop:
- a Spatial Decision Support System (DSS)
- malaria prediction models for the Lubombo SDI.
The Spatial DSS consists of :
- A Malaria Information System
- A GIS Data Repository
- A web-based spatial and statistical query and analysis tool
The DSS provides support for:
- The extension of malaria control to southern Mozambique
- Changes in the distribution and incidence of malaria, as a result of the implementation of the malaria control programme
- the assessment of the intervention effects on development, with a special emphasis on tourism and job creation.
- broader decision-making on development issues within the region
- the production of malaria risk maps for tourists and tourism developers to ensure they are able to make decisions based on accurate, up-to-date information.
1. Malaria Information System
Accurate malaria case information is critical for the planning of malaria control as well as the assessment of the efficacy of the intervention in the region.
Malaria Information Systems have been set up in the malarious provinces of South Africa. These computerised systems allow the input, management and output of malaria case information which is central to disease management, research and the regional evaluation of the LSDI malaria control program
In South Africa and Swaziland, all cases are definitively diagnosed and definitive diagnosis will be implemented in phases in Mozambique beginning in 2004.
The MIS has been adapted for implementation in Mozambique and Swaziland. Information officers have been employed in Mozambique and Swaziland to assist the control programmes with the collection, management and distribution of malaria-related information and to provide technical support. Information officers will be hired and placed in each of the three malarious provinces in South Africa in 2004.
The data collected during routine operations and entered into the MIS consists of both in- and out-patient data of confirmed and clinically diagnosed malaria cases. The input screens mirror the data collection forms and the automatic-linking and drop-down list minimises data entry errors. Pre-designed outputs are provided in the form of maps, graphs or tables (i.e. number of can refills per week per person). This allows problems to be identified and addressed on an ongoing basis. Data capture takes place at each of the regional offices with the data being sent to the MRP where it is cleaned and analysed. The cleaned data and the results of the research are fed back to the partners who use the MIS for management purposes through the outputs provided.
The MIS provides the ability to identify areas of risk and to assist decision makers in directing resources and strategies. The ready access of data, its rapid entry, analysis and output in the form of graphs, tables and maps, allows for stream-lining of the data within the region, and for rapid management and evaluation of the situation on an on-going basis, enabling the cost effective use of resources. The SDI's GIS-based MIS and the resulting availability of spatially-referenced data, will have implications for the development of historically disadvantaged communities, for example, the electrification of health facilities, the optimal placement of roads, new schools, clinics and other infrastructural developments.
Importance of Malaria Incidence Data
Malaria risk assessment and control intervention efficacy in Mozambique is
currently based on malaria prevalence in children aged 2 < 15 years at
pre-determined sentinel sites. Baseline prevalences were all above 60% prior
to the implementation of malaria control, but have decreased to less than
10% in Zone 1 over 3 years. Falling prevalence ratios require bigger samples
to show significant decreases. Malaria incidence based on definitive diagnosis
at health facilities becomes the best measure of malaria risk and will be
the uniform measure used in the 3 countries. This is made possible by Global
Fund funding which will provide for definitive diagnosis capabilities at health
facilities in Mozambique and Swaziland. This system is already fully operational
in South Africa and is well under way to be developed in Swaziland and Mozambique
(See Figure 10 in Section A.1.4.2). The MIS is critical to assessing malaria
incidence.
Spatial Analysis of MIS data
A stand-alone spatial component of the MIS application is based on the Health
Mapper application developed by the World Health Organisation (WHO)and is being
customised to minimise end-user skill requirements and optimise access to spatial
data sets.
More extensive spatial analysis of MIS data is carried out by the GIS unit of the Malaria Research Programme of the MRC.
Implementing the GIS platform required various data sets, both spatial (map e.g. roads, clinics) and attribute (statistical e.g. malaria case data, population), hardware and software, and appropriately trained personnel. Spatial data has been collected for the region and includes administrative boundaries, roads, population, health facility locations, towns and schools. New sources are continually sought to ensure that current data at appropriate scales are provided.
The vector-based desktop GIS software package MapInfo was used to provide the spatial platform of the MIS. Microsoft Access provided the database platform as well as the front-end data entry and output screens. Due to the variety of skills levels in the partner-sectors, and the need for rapid data entry and retrieval mechanisms, data entry screens that mirrored the data forms were designed for each centre. These included automated data entry mechanisms on the pre-selection of a number of columns. The outputs were designed to meet the management needs and were provided in graphs, tables and maps.
The GIS data will standardise and strengthen the Health Information System (HIS) of the SDI partner-sectors. Acting as a management and research tool, the primary functions of the GIS-based MIS are to:
- Identify the current distribution and density of the population, as the current available demographic data are estimates only.
- Monitor changes in the distribution and incidence of malaria over space and time. Evaluate over time the effects of the new roads and tourist developments on the distribution of malaria.
- Provide distribution data for tourist and tourism planning. Maps indicating malaria distribution are essential to inform tourists, investors and decision-makers of the areas of high- and low-risk, as well as changes that occur over time. It is envisaged that the implementation of the spraying programme will dramatically reduce the disease incidence in the LSDI area with a related benefit for tourism and agricultural development. Maps displaying malaria across the region currently do not indicate small-scale variations, thereby necessitating the acquisition and display of data at a finer resolution to effectively promote tourism. The movement of people with the implementation of the tourism and infra-structural developments could result in changes in the distribution of malaria incidence and risk. Maps indicating changes over time will need to be made available to tourists, potential developers, and decision-makers working in the region.
- Assist in malaria control and planning of water resource developments. The agro-tourism focus of the SDI includes the development of water resource projects in endemic malaria areas, bringing a very real danger of a dramatic increase in malaria incidence in the surrounding areas. Mapping these sites in relation to malaria risk areas and population distribution will indicate areas that are likely to require additional control measures.
- Community based malaria vector control in southern Mozambique. Optimally allocating community based spray operators required identifying the distribution of the population in the areas to be sprayed in Zone 1. This ensured the optimal use of the vehicles in order to supply the malaria control programme with sufficient insecticide and equipment, transport spray operators to areas inaccessible on foot, and to provide health facilities with adequate drugs. Recording the details of the insecticide applications allowed the programme to be monitored and the spraying status to be displayed on an on-going basis.
2. Spatial Data Repository
The availability of GIS data for the LSDI region has traditionally been limited and when available often in different formats (digital, CAD and hardcopy), and housed with different organizations, making it difficult, if not impossible to create composite maps of the region. The aim of the GIS data repository was therefore to collate these datasets into a single resource which can be accessed by end-users in the different sectors of the LSDI.
The data has been grouped into the following categories:
- COMMUNITY
- TOURISM
- ENVIRONMENT
- Greater St. Lucia Wetland Park (GSLWP)
- MALARIA
- BACKDROP MAP LAYERS
Within each of these categories there is data specific to the three countries,
as well as the anchor project of the LSDI, the GSLWP. Most of this data was
provided in ArcView shapefile format and had to be converted to MapInfo. The
original shapefiles have been stored for request purposes.
3. Web-Mapping Application
Introduction
The web-mapping application, LubomboMapping (www.LubomboMapping.org.za), is a vehicle for the dissemination of spatial information over the internet to decision-makers in the Lubombo region, and a tool whereby users can interactively interrogate, analyse and display the data to support decision-making.
The importance of spatial data analysis for planning and management
decisions is well documented in numerous application areas. However, its use
has typically been limited due to the cost of data and software, the need for
extensive expertise and the difficulties in sourcing spatial data sets. The
innovation of web-GIS applications overcomes these limitations in that a user
only requires internet access and no previous experience to begin to view and
analyse spatial data sets that have been sourced and made available by the web
application. The aim of LubomboMapping is to provide a tool for the analysis
of spatial data for the Lubombo SDI to support decision-making on malaria control
as well as broader development issues in the region.
Development of the Application
An extensive review was made of available web-GIS software. Software packages were compared according to speed, reliability, robustness, versatility and performance. MapInfo’s MapXtreme rated the highest overall and was purchased. A consultant group based in Pietermaritzburg, InterMap, was selected to train MRC staff in developing web-GIS applications using MapXtreme software, the MapX object model and the Hahtsite programming environment. A server at the MRC, Cape Town office was set up as a webserver and MapXtreme was installed and configured.
A review of existing internet-based GIS applications was conducted in order to assess the functionality and data sets offered by these applications. In order to establish user requirements, meetings were held with Lubombo SDI and GSLWPA personnel to establish their needs in terms of decision support and planning. Workshops were also held with malaria researchers to assess their requirements.
The spatial data sets in the repository were categorized and incorporated into the application as data layers to be turned on or off. A prototype application was developed using a Java mapplet to provide basic GIS tools and HAhtsite code embedded in html to allow access to MapX objects. Microsoft Acess 2000 is used as the database back-end application to store data sets required for the search function, which is based on Microsoft ASP.
The draft application was demonstrated to MRC colleagues for appraisal and recommendations. The recommendations forthcoming were then incorporated into the application. The GIS tools selected for inclusion in the application were:
- Zoom, allowing the user to display spatial data at different scales, showing more detail as the user drills down.
- Pan, which allowed the user to drag the map to view different areas of the map displayed.
- Distance tool which measure the distance between selected points.
- Labelling tool toggles the labels on and off
- Print button allows the user to print a customized map.
- Map Layers allows the user to select which data layers on the map to leave visible
- Info tool allows the user to select a feature such as a tourist facility and view any stored information associated with that point such as contact details or malaria prophylaxis advice.
- A search option allows a user to search for a town, tourist facility, school or health facility and then locate it on the map in relation to other features like roads, reserves, police stations, schools, landcover, malaria incidence.
Data layers selected from the data repository that can be displayed as different layers in the application include:
- Community: population, schools, police stations
- Tourism: Tourist facilities, game reserves, arts and crafts
- Environment: landcover, land use,
- Census 2001: population, household services mapped thematically
- Malaria: official malaria risk guidelines, malaria cases at small-scale.
The draft application was then demonstrated to selected institutions
to assess user requirements. LubomboMapping
undergoes continual extension and refinement as new data sets are sourced and
feedback from users indicates the need for further analytical tools.