Technological progress has led to the emergence of new computerized analysis tools, known as geographic information systems (GIS), whose use in the health field vary from the simple automated mapping of epidemiological data (Pyle 1994), to the sophisticated analysis of satellite images which demonstrate vector/environment relationships (Hugh-Jones 1989; Perry 1991; Rogers 1991; Malone 1992). GIS have already been widely used in other sectors such as the management of natural resources, agriculture, and rural and urban planning (Rideout 1992).
The purpose of this article is to show how, on the basis of specific examples drawn from WHO's experience with the ministries of health in Botswana, Senegal and Morocco, it is possible to meet the decision-making needs of countries through the use of GIS within tropical diseases control programs. This review illustrates how geographic information systems facilitate the monitoring and management of control programs and open new avenues for intersectoral collaboration.
Specific needs for health services vary according to the level of decision-making. The district medical officers, who in developing countries frequently move from one post to another, must be able to acquire a thorough knowledge of their area of responsibility rapidly and easily on the basis of data in their possession.
They are the key partners of the national disease control services, and must have the capacity to express the needs of their districts and to determine its health priorities. They are responsible for the reliability of the data communicated by the clinics, and consequently have specific needs with regard to training, increased awareness, and motivation of the local health personnel for data collection.
The regional medical officers are located at the interface of the national and local services. They must have accurate and reliable data for planning. At this level, they are often the chief coordinators of research projects carried out by foreign institutions, providing the link between research and action.
Epidemiological surveillance is an essential need for tropical diseases control programs (malaria, schistosomiasis, onchocerciasis, and so on). These programs must be capable of constantly generating updated information and should be useful in guiding field operations: when and where to intervene, which would be the most effective interventions, whether an intervention is feasible with the limited resources available, and so on.
At the central level, the statistics and epidemiology departments are responsible for determining long-term trends, assessing specific risk factors, and integrating data that are not directly supplied by the health services to support planning and management. Reliable but still readily understandable information should be available to decision makers.
GIS can be used not only for automatically producing maps, but it is unique in its capacity of integration and spatial analysis of multisource data: population, topography, hydrology, climate, vegetation (satellite pictures), access routes (roads and railways), public infrastructure (schools, main drinking water supply), and health infrastructure, including epidemiological data on diseases.
The GIS is often misperceived as a sophisticated technology, requiring satellite imagery that is often inaccessible to developing countries. In reality, GIS enclose a wide range of hardware and software covering a span of affordability and technical performance. This technological diversity offers great flexibility so that it can be implemented in most developing countries according to their needs.
Schistosomiasis Control in Botswana
The first infection of Schistosoma haematobium in Botswana occurred in Lobatse Railway in
1930. Transmission spread, and by 1978 S. haematobium infection was prevalent throughout the
country at rates ranging from 0.6% to 14%. S. mansoni infection was first detected in Ngamiland
District in 1965 with a prevalence rate of 3.1%. However, by 1983 a survey of primary school
children in Maun revealed an 80.3% prevalence of S. mansoni and 1.4% of S. haematobium.
Recognizing the public health importance of schistosomiasis, a control program was established in 1985 in Ngamiland District with two main objectives.
Simplified data collection and analysis procedures were used. The data collection forms were designed according to the objectives of the program and stored in DBF with the aid of a private NGO.
School Survey Results
Six school surveys were undertaken between 1986 and 1991. Of the 16,402 enroled children in
45 schools, 15,943 (97%) were examined during the first survey. Overall prevalence of S.
mansoni was 28.7%, while prevalence of heavy infection was 4.1%. By 1989, the survey was
carried out only in schools where the prevalence remained above 10%. Within these, overall
prevalence of S. mansoni was 8.4%, while prevalence of heavy infection had fallen to 0.2%.
The sixth school survey in 1991 was conducted again where previous prevalences remained above 10%; 5,381 children were examined. Prevalence of heavy infestation had decreased to 0.01%, with total prevalence at 6.7%. Only one primary school had a much higher rate, 14.68%. By 1993, however, the prevalence rate in that school had dropped to 8.3%.
Community Survey Results
Mobile field teams carried out community surveys after consultation with the local authorities and
local health facilities. These surveys were designed to identify and treat adults and children not
attending schools (about 30% of the school age population) who were at risk of exposure to the
disease. Two community surveys were conducted in the villages in Ngamiland in 1985-86 and
1987-88.
The prevalence of infection in all villages in Ngamiland was much lower than in the schools (19% during the first survey). Following selective chemotherapy of affected individual wards and villages, only 2 of the 25 wards in Maun, and 1 of the 12 villages in Ngamiland, showed an increase in prevalence during the second survey. One explanation for this might be the mobility of the population.
The crucial issue which remains to be addressed is monitoring the trend in prevalence rate reduction in relation to changing environmental factors. GIS was seen as an tool to assist this study.
Integration of Environment and Health with GIS in Senegal
Seeking to control the drought in Senegal, two dams were built, one 27 km from the
regional capital city of St Louis, the other in Mali territory, which lies some 1200 km from
St Louis. These dams have led to ecological changes which were responsible for a severe
outbreak of schistosomiasis in Richard Toll in 1988.
The construction of the Diama dam was meant to stop salt water intrusion from the Atlantic Ocean into the Senegal River. Richard Toll is located 106 km from St Louis. It is by far the largest rural agroindustrial city in the country. Its main industry is cane sugar manufacture, which attracts people from the surrounding areas where intestinal schistosomiasis is endemic.
The population is currently expanding so quickly that alimentary provisions such as clean water supply are lacking. The only source of water for domestic use remains the irrigation canals of the factory, and the river.
Schistosomiasis prevalence is estimated at 60-90% in Senegal, varying by region, and all age groups are equally affected. A recent study carried out in 1993 in the district demonstrated that the disease is spreading outward into other regions, in spite of the preliminary control measures which were undertaken.
The Ministry of Health currently recognizes that adequate control requires a deep and precise understanding of the distribution of the disease according to environmental data. This focus provides a relevant justification for using GIS.
The Ministry of Health project has two main objectives.
Implementation of GIS
WHO's approach to the implementation of GIS in the control of tropical diseases is fundamentally
different from that taken by universities or research laboratories. The latter, by
virtue of their terms of reference, have the option of using sophisticated technological resources
such as remote sensing satellite imagery and mathematical modelling. As WHO's approach is
focused on the optimal use of resources already existing within a country, its process for
implementing GIS for health follows six main lines:
To ensure that the utilization of GIS is quickly undertaken, the system should be confined initially to a limited geographic area related to a priority disease/problem in the country. In Botswana and Senegal, starting from the expression of a specific need for schistosomiasis control, the GIS concept will be used in one region to integrate a whole set of data, a process which facilitates subsequent expansion while setting successive objectives.
It is necessary to identify the national GIS resources in other sectors. Specifically, the available boundary files, databases, and technical skills should be determined. GIS is generally not used in the health field, while other sectors, such as the management of natural resources, agriculture, and water resources, are already advanced in its application. Moreover, in some developing countries international agencies (e.g., UNITAR, UNDP, UNEP, FAO, World Bank) have helped to set up centres of excellence in GIS, cartography and remote sensing. These centres have trained staff and geographic databases (automated maps), and form the nucleus of the national network of GIS users (Sahel and Sahara Observatory 1991).
In Botswana, the National Committee for Mapping and Remote Sensing, supported by a GIS users group, defined the standard formats to be followed in GIS. The group recommended that for each GIS project set up in Botswana, data should be compatible with the ARC/INFO format (Nkambwe 1994).
Since several departments of various ministries (e.g., Water Resources/Department of Water Affairs; Planning/Department of Town and Regional Planning; Agriculture/Department of Wildlife and National Parks) were already using GIS, the central government designated the Department of Surveys and Land to produce a digitized map of the country at a scale of 1:250,000. This map would comply with the defined standards and would be available and used by all the ministries concerned. At the same time, a private company with technical capabilities and special interest in nature conservation initiated a project to digitize a map of northern Botswana.
After identifying the available resources, the role of WHO, assisted by IDRC, was to support the initiatives in progress, both financially and through institutional collaboration. The mapping of Northern Botswana (1:250,000) was completed so that it could be used not only by the health structures, but by all ministries. An agreement was reached with the Department of Surveys and Lands that, after verification, the digitized cartographic database would be approved as an official map of Botswana. Similarly, training of the health staff involved in GIS was based on collaboration between the Department of Environmental Sciences of the University of Botswana, which provides appropriate training for the staff of certain ministries (e.g., Ministry of Local Government, Lands and Housing), a consultant of the ESRI conservation program attached to a private company, and WHO for the health component.
In Senegal, the Ecological Monitoring Centre (CSE) is the principal partner of the Ministry of Health. The CSE is a centre of excellence which was initially supported by the international agencies. It is now independent. Its terms of reference are the supply of reference data on national resources, the monitoring of indicators of environmental status, the management of a database integrated in a GIS, and the dissemination of information on the environment to planners and decision-makers. Within the GIS health project for the river valley, it is responsible for integrating the database of the Ministry of Water Resources (SIGRES project: Geographic Information System for the Management of Water Resources in Senegal) with all the other available data, and transferring them to the Mapinfo format for MacIntosh.
In Morocco, as in Botswana, there are two national committees working in the field of GIS: the national committee on mapping and the national committee on remote sensing. The task of these two committees is to coordinate activities between these two areas of study. This collaboration highlights the importance of developing GIS in other ministries (e.g., the Ministry of Mines and Geology, Water and Forestry, Water Resources, the State Under-Secretariat for the Environment, the Ministry of the Interior, the Royal Police Force). To cope with the demand for basic data, the Department of Land Conservation, Land Registry and Cartography has been entrusted with the task of digitizing all maps at a scale of 1:250,000. The automated maps will be available to any ministry requesting them, as is already the case with the paper maps. The Royal Centre for Spatial Remote Sensing, a public body working primarily in the field of remote sensing, serves as the national reference centre. It is particularly well endowed with regards to equipment, and it also provides training.
The participants in the GIS project are located at different levels within the health system. To ensure national commitment, the initiative to set up a GIS must be taken at the higher levels of the Ministry of Health. The decision-makers can then swiftly perceive the potential benefits of GIS and can participate in setting up the network of collaborators, which is one of the cornerstones for the success of the project. The Permanent Secretary of the Ministry of Health in Botswana, the Director of Public Health in Senegal, and the Director of the Health Programs in Morocco fully participate in the projects on GIS for health in their respective countries.
The practical implementation and monitoring of the project is specifically entrusted to a GIS task force. The members belong to the division of statistics or epidemiology: epidemiologists, statisticians, computer scientists and geographers, according to the availability of local manpower. In Botswana, the core group is built around the unit of epidemiology, while in Senegal the leadership was taken by the director of health statistics. In Morocco the team will be driven by geographers. These task forces are the preferred contact points of the national GIS centres of excellence. The local health services, in particular the district chief medical officers, are the first suppliers of data to the system and the first users. In Botswana two districts are integrated in the initial phase, in Senegal five.
When choosing the hardware and software to be used within the GIS projects, several considerations must be taken into account. Collaboration is stressed in this approach to the use of GIS for health in general, and for tropical diseases in particular. The GIS must remain a means of analyzing information and of making good use of what is available; it is neither necessary nor desirable for departments of statistics or epidemiology to turn into cartography units. Accordingly, preference is given in the initial stages to equipment for visualizing information rather than equipment for data input. For each project, three elements have been taken into account for the choice of software:
In Botswana, for example, the ARC/INFO format was fixed as standard for the government's GIS data. The complexity of using PC ARC/INFO led to the utilization of ArcView, which is much simpler. (ARC/INFO and PC ARC/INFO are registered trademarks of ESRI, Inc., Redlands, CA, USA.) It can be used for visualizing geographic data directly in the ARC/INFO format and for attaching specific information to them. By choosing this software, it is possible to respect the high standards established in the country for GIS, while adapting to the needs expressed by the health sector. It ensures that during the initial phase of the project, the various participants (epidemiology unit, national program for control of parasitic diseases and the two district medical officers) work with the same tools. This offers the possibility, if necessary, of switching to more powerful software and collaborating with research laboratories which may wish to supplement the analyses with studies integrating satellite imagery.
Two types of data need to be included in the GIS: strictly geographic data and information relating to these data, known as "attributes." The geographic data include such items as administrative boundaries, the location of towns and villages, roads and railways, rivers and the main areas of vegetation. The attributes may be general (category of road or type of village), or more specific (population at different dates, availability of a clinic, hospital, school, and so on). The epidemiologic data are reported either for the village where the clinic is located or for the clinic's area of influence if it can be identified. These data are derived from specific surveys conducted previously or from routine reports.
The project in Botswana was initiated using data from exhaustive school surveys carried out by the Schistosomiasis Control Program between 1986 and 1991. In Senegal, routine data collected from health posts will serve as a basis for the analysis. In Morocco, surveillance data from the schistosomiasis control program will be integrated.
One of the key components for the utilization of GIS is the ability to ensure appropriate training. It has been demonstrated in Botswana and Senegal that training should be provided for everyone involved in a GIS project, and should be adjustable to the subsequent responsibilities of each person. A large part of the training should be devoted to the practical use of the tools. Decision-makers could be given shorter training, consisting of a general presentation of the concept of GIS and of the data to be incorporated. The potentials and limitations of GIS for decision-making should be stressed. It is desirable for decision-makers to be able to use the software selected in order that they appreciate its usefulness and its limitations. Joint training for district medical officers and for members of the GIS task force ensures that everyone has the same basic knowledge.
Training is carried out through national centres of excellence, using geographic and epidemiologic data from the area of study. The involvement of these centres is essential. The presentation of concepts and the practical work refer more specifically to health problems, and it may be necessary to bring an external consultant for the initial training. Later, the members of the GIS task force will be able to take over the training.
The constant simplification of the handling of GIS software may conceal both the importance of data in GIS and the limitations of GIS. For the control of tropical diseases, all applications begin at the periphery. If the data generated at the periphery have not be reliably collected and grounded, no level of sophistication of the technology at the central level can improve it.
The level of reliability of geographic data vary according to the scale of the map. A village, for example, appears as a dot on a map with a scale of 1:200,000 (1 cm on the map represents 2 km on the ground) and a clinic building cannot be individually identified. On a map at a scale of 1:50,000, a village occupies a whole area within which the health facilities may be visible. It seems that a basic map at 1:200,000 or 1:250,000 corresponds to the management needs of disease control programs conducted at country level because it makes it possible, through the enlargement and reduction functions (zoom in and zoom out), to obtain an overall view if necessary, but one that is still sufficiently detailed at district level.
It is necessary always to bear in mind the scale of the basic map, for even if the functions of the software make it possible to enlarge the maps almost infinitely, the precision of the maps is determined at the time of input (a pool 100 m in diameter — or theoretically 0.5 mm on the map — can never be individually identified at 1:200,000).
In a given country, the first ministry or agency that tends to use GIS at the national level with reliable geographic data must meet the cost of inputting these data. If the expenditure has been high, the ministry will tend to try to protect the use of these data like an asset. It is therefore necessary to create a clear definition of the procedures for information exchange. A financial contribution could be charged for the digitized maps, as is commonly done for paper maps. What is paid for is not the information in itself, access to which is unrestricted, but the work of inputting this information. Provided that the amount charged is reasonable, this seems one of the safest ways to avoid duplication of the work, while providing some return on the basic investment. Each agency can then supplement the basic map according to its own data, integrate information from different sources, and conduct its own analyses.
As opposed to research where the reliability of the health data, especially those related to the diagnosis of a disease, is extremely important in the process of risk factor identification, control programs based on surveillance systems might benefit from the integration of routine data into a GIS for a better management. Frerichs (1991) states that information feedback is the very key to the success of surveillance systems, and is facilitated by the utilization of information technology. He stresses the importance of graphs in the presentation of results and refers to the value of mapping. GIS do in fact meet this need for feedback to the district level by enabling routinely collected data to be presented easily and attractively. This process is by no means new and merely replaces the conventional map displayed in the offices of district medical officers on which the health posts are represented by pins of various colours. Nevertheless, GIS provides an advantage, inasmuch as it facilitates what Sandiford calls the "ritualization" of the interpretation of routine data (Sandiford et al. 1992). It is not the fact that the nurse regularly takes the patient's temperature that will enable the surgeon to detect a postoperative infection, but the fact that the information is plotted on a chart and regularly examined at the patient's bed.
In Senegal, for example, by strengthening the management information system, the GIS will make it possible to analyze routine information. Sandiford (1992) states that people are generally somewhat hesitant to use routine data under the pretext that they are not reliable. Looking at this from the viewpoint of "action-led information systems," that is, systems in which it is not worth collecting the information unless it leads to a decision, it becomes apparent that the lack of quality of the data is often due to their underutilization. As soon as the information is used, the errors and anomalies are rapidly corrected through the feedback process.
This review indicates that GIS should form an integral part of surveillance systems, for it is one of the few tools meeting the need of monitoring the distribution of a disease in space. GIS surveillance of tropical diseases have been mentioned elsewhere. In Israel in 1992, the surveillance of imported malaria cases, combined with the identification of anopheles breeding sites, led to the precise identification of the intervention areas and thus enabled malaria transmission to be kept within bounds (Kitron 1994).
In South Africa (le Sueur personal communication) the cartographic representation of geographically referenced databases of the malaria control program makes it possible to locate the high-risk areas in space and time. In the onchocerciasis eradication program in Guatemala, GIS are used to identify communities at risk and as tools for assisting in ivermectin distribution (Richards 1993).
In Morocco, the same approach will enable the schistosomiasis control program to concentrate its efforts on areas where the disease is still present, while maintaining active surveillance throughout the country, to attain the objective of eradicating the disease. The WHO/UNICEF program for the eradication of guinea-worm disease also relies on the mapping of endemic villages (Clarke 1991).
As was mentioned above, the simplification of the handling of GIS may conceal the importance of the data, and furthermore the significance and limitations of mapping. It may be very tempting to map just any variable. Experience has shown that even the basic epidemiological concepts such as prevalence tend to be forgotten when mapping is carried out (for example comparisons based on an absolute number of cases). The greatest risk when analyzing superimposed data is the deduction of causal relationships from mapping. It is important to stress that the simple visualization of data under no circumstances permits the conclusion of a cause-and-effect relationship between the various phenomena observed. Nevertheless, the GIS makes it possible in this context to demonstrate spatial relationships which might lead to subsequent in-depth epidemiological research (Scholten et al. 1991).
Given the current limitations of GIS for spatial epidemiological analysis, GIS still has an important role to play in the management of control of tropical diseases as it meets a real need in decision-making. When Frerichs (1991) claims that most developing countries are investing in administrators rather than in epidemiologists, and that the latter should therefore have the means to present rapid and readily understandable results to the administrators, it seems clear that the use of GIS meets a real need in decision-making.
The time of decision-makers is becoming increasingly valuable and they cannot afford to waste it by wading through pages of reports to extract the substance. They need brief presentations, focused on pertinent comparisons. When such reports are available, they often present conventional comparisons, such as the value of an indicator in terms of a theoretical reference, often defined at the central level, or a comparison of a variable from one year to the next. It is often more efficient in terms of management to make comparisons between similar health care structures or neighbouring regions (Sandiford 1992). There are two objections to such comparisons: first, the level of aggregation of the data, which has already been discussed above, and second, the difficulty of sharing the information. The very basis of the GIS can remove these two obstacles and provide a unique opportunity for meeting the needs of the decision-makers.
Back to Contents -
To next section
This file was created 23 February 1996