Home Articles The use of GIS in research and control of Malaria

The use of GIS in research and control of Malaria

Thisula Abeysekera, D.M. Goonewardena, Gamini Jayasundera, Lal Muthuwatte, P.Kumr Kotta1, Thilak Senanayake2, Richard Carter3 Kamini N. Mendis, A.R. Wickremasinghe4
Malaria Research Unit, Faculty of Medicine, University of Colombo.
1SACEP, Anderson Road, Colombo 4.
2Mahaweli Authority of Sri Lanka, Polgolla.
3Division of Biological Sciences, University of Edinburgh, Scotland, U.K.
4Deparment of Community Medicine and Family Medicine, Faculity of Medical Science, University of Sri Jayawardenepura

Malaria is a disease transmitted by mosquitoes of the Anopheles species. Its distribution in Sri Lanka Corresponds to the climatic zones of the island with the dry zone carrying the highest case load, the intermediate less, and the wet zone being almost free of the disease except during epidemic periods. Malaria transmission varies seasonally corresponding toe monsoonal rainfall. The distribution of malaria is not homogeneous even within climatic zones, and scattered pockets of high transmission are found mainly in the dry zone. Malaria is affected greatly by environmental factor such s h u mediate and altitude which affect the survival and longevity of the vector mosquito, and the presence of water bodies which are its breeding sites. It has an association with large scale agricultural development projects due to creation of breeding sites and he migration of labour from non-malarious area, they being more prone to disease due to low levels of immunity.

The Malaria Research Unit (MRCU) of the Faculty of Medicine, University of Colombo has been conducting research on malaria and its control over the last 15 years. A geographic Information system (GIS) laboratory was established in the MRU in 1994. The objectives of the programme is to establish a GIS for health with emphasis on malaria. Since the establishment of the GIS laboratory at the MRU, research has focussed on two major projects: The first project is on the micro-epidemiology of malaria, where the unit the analysis was the household. The second project is on macro-epidemiology in which the objective is to determine environmental an land use pattern correlates of malaria at the Grama Niladhari Divisional level and beyond.

1. The Micro -Epidemiology Project.
The purpose of this study was to determine the association between the incidence of malaria and the distance to sources of water and the forest controlling for the type of house construction, the latter shown by us previously to be a risk factor for malaria. The study was carried out in 8 adjoining villages in Kataragama, situated in the Moneragala District of Sri Lanka. The study units consisted of 423 households and a population of 1875. Malaria data was collected over a period of 18 months from January 1992 to June 1993. All individuals were individuals were identified by a unique identification number and it was possible to link each malaria case to a particular house in the area. The detection of malaria was carried out at the field station located in the area by trained personnel. We assume that almost all the malaria infections that occurred in the area detected by us.

The study area was mapped using aerial photographs of 1:5,000 magnification, obtained from the Survey department and field checked by the project personnel. The maps were digitized using the Arc/Info software package. Coverages of roads, streams, the landuse pattern and houses were digitized separately and overlaid to obtain the final map. Data on individual households was linked to the point attribute table (PAT) final map. Data on individual households was linked to the point attribute table (PAT ) files of houses that are generated by Arc/Info. Two varieties i.e. the Incidence Rate and the House Type, were used for analysis. The incidence rate was calculated as the number of malaria cases in a household that occurred during the 18 month period of the study divided by the number of members of the household. Household with 2 persons or less were eliminated for statistical reasons. House type was classified into two according to the type of house construction namely the walls and roofs; ‘good’ houses had brick walls and tiled roof and the ‘poor’ house and mud walls and cadjan roofs. It was shown that the malaria incidence rate of ‘poor’ house was significantly higher (2.5 times) than that of ‘good’ house.

Distances between individual houses and water bodies and the edge of the forest were obtained using the NEAR command IN Arc/Info. There was no association between the incidence of malaria and the distance from a house to the forest edge after controlling for house type. There was a significant negative correlation between the incidence of malaria and the distance from a house to a water body only in the case of ‘poor’ house. A buffer zone of 200 meters around water bodies was simulated, from which, we eliminated the ‘poor’ houses and relocated them outside the buffer zone. The reduction in the incidence of malaria that this intervention would yield was estimated to be approximately 30%. 2. Macro-Epidemiology Project
This study uses government administrative units such as Grama Niladhari (GN) areas and Divisional Secretary (DS) divisions as the unit of analysis. This study was done in the Moneragala Division of the Moneragala District a major part of which is situated in the dry zone and is endemic for malaria. The purpose of this study was to describe the malaria situation in the DS division, monitor the pattern of malaria transmission over time, identify environmental factors influencing the transmission of the disease, and possible in what way, using routinely collected data which were obtained from various government departments such as, the departments of Census & Statistics, Landuse Planning Division of the Ministry of Lands, Irrigation and Mahaweli Development, Anti-malaria campaign of the Health Department, the meteorology Department, etc.,. The transmission of malaria in this district at a D.S. division level shows an increase from north to south, corresponding to the agroecological zones of this area which shifts from a high altitude intermediate rainfall zone to a low altitude low rainfall (dry zone) area. Further investigation of a D.S. division (Moneragala D.S. Dvision) revealed certain G.N. arean with high incidence rates. Environmental factors which might influence trasmission such as landuse pattern were considered in the analysis. The maps on landuse were considered in the analysis. The maps on landuse were considered in the analysis. The maps on landuse were obtained from the Landuse Policy Planning Division (LUPPD) in Moneragala where landuse map of 1:20,000 were available for certain D.S. divisions. The Moneragals D.S. division, was digitized in our unit using ARC/INFO. The different crop types were labeled according to the classification used by the LUPPD. Separate GN separate GN sections were clipped out and percentage of the different land use areas such as areas under homesteads, forest, paddy, rubber etc., were calculated using the Statistics command. These percentages were used to correlate with the malaria incidence in the same GN areas for 1992. There was a significant negative correlation between the malaria incidence and area of land under rain-fed paddy cultivation (r=.51185, p =0.0427) and a positive correlation with the area of dense forest (r=0.69986, p=0.00012) in the Moneragala DS Division. The correlation between malaria incidence and the area of land occupied by homesteads was of borderline significance (r=0.43998, p=0.0677).

Rainfall data was obtained from the Meteorology Department. This was in the IDRISI format and separate image were available monthly. These images were converted to the Arc/Info format, the district with its administrative boundaries overlaid, and the average rainfall for separate divisions were obtained again using the Statistics command. This analysis is still in progress.

This paper demonstrates the use of GIS and spatial analysis in health, with special reference to malaria. The microepidemilogy, study provided us with the malaria situation in a particular area at a household level. It was possible to identify houses with the heaviest caseload for that period of time. Using this information derived from a research programme, some relevant and important recommendations could be made with respect to the cont. of malaria. For instance that control efforts should focus attention on high risk groups, i.e. certain household or localities with a heavy caseload. Such an approach will probably be more cost effective than employing malaria control methods in a conventional manner in which total or ‘blanket’ coverage of the population is involved. Results of the micro-epidemilogy study also emphasizes the ability of a computerized GIS to provide quick and reliable information for planning purposes. For example, the imposition of a buffer zone around a particular geographic landmark, eg., water bodies, and the estimation of the impact of excluding house of poor construction type from it by simulation, to estimate the potential impact of such an intervention. An economic analysis is underway to determine the relative cost benefit of updating house of “poor” type to the “good” type in high in high risk areas, and relocating such houses to low risk areas, and to compare it with the cost of carrying out routine malaria control as practiced currently, taking into account the reduction of malaria which would result from the interventions suggested above. Such analyses will also enable the health administrators in allocating/reallocating health care resource more equitable.

Many problems were encountered in the use of routinely collected data. The malaria situation a given area is expressed by the malaria incidence rate. This measure which consists of the number of cases in the area divided by the population of the area indicates the seriousness of the problem in a demarcated area and also makes it comparable with other, maybe adjoining area. Sri Lanka has a good infra structure for the collection o health statistics compared to most developing countries. Malaria case data is routinely collected from govt. health care institutions were the case have been confirmed by blood filming, and is available monthly from regional malaria control officers situated in districts where malaria is a health problem. Although this gives an idea of the case load and the monthly changes in these areas, this could be improved on with the inclusion of patients who are treated in the private sector, which could account for a high proportion of cases in some areas, and proper location of patients according to their area of residence rather than which institution treatment was sought. This would give a more accurate figure for demarcated administrative areas for the purpose of carrying out analyses on correlating factors. Other than for the case data, the accuracy of data on population especially in small areas tend to differ between different departments. If data collection was spatially based this problem might be overcome with all sectors having data for the same administrative boundaries ranging from the GN (the smallest administrative unit) to District level. For the control of malaria and indeed for all vector borne diseases it is important that many sectors such as environment, irrigation ,. Urban development collaborate with the health sector, and a common data format as that provided by GIS could be a invaluable tool, provided accurate and relevant data bases are available for its full potential to be reached.