- Email: [email protected]
Geographic information systems as a tool for control program management for schistosomiasis in Egypt
Acta Tropica 79 (2001) 49 – 57 www.parasitology-online.com
Review
Geographic information systems as a tool for control program management for schistosomiasis in Egypt M.S. Abdel-Rahman a, M.M. El-Bahy a,*, J.B. Malone b, R.A. Thompson b, N.M. El Bahy c b
a Faculty of Veterinary Medicine, Cairo Uni6ersity, Giza, Egypt Department of Pathobiology, School of Veterinary Medicine, Louisiana State Uni6ersity, Baton Rouge, LA 70803, USA c Faculty of Veterinary Medicine, Monifyia Uni6ersity, Sadat City, Egypt
Abstract During a 4-year study a geographic information system (GIS) risk model was constructed for predicting the relative risk of schistosomiasis in Kafr El-Sheikh governorate, Egypt. A 1-year 1990– 1991 time series on diurnal temperature difference (dT) prepared from the advanced very high resolution radiometer (AVHRR) sensor on the NOAA-11 satellite was used to develop a regional risk model for the Nile delta based on thermal-hydrological domains. A May 15, 1990 Landsat TM scene (path 177, Row 38) was used to develop a local ‘village-scale’ environmental risk model based on higher resolution satellite sensor data (30 m picture element size at earth surface). Four of ten classes derived from a tasseled cap (Tcap) transformation of the Landsat TM scene were shown to be significantly related to a 5-year Schistosoma mansoni prevalence database from the Ministry of Health. A risk model was developed based on dT and the proportional area of the four Tcap classes in 5 km2 buffer zones centered on rural health unit (RHU) reporting units. Available historical data on S. mansoni and its snail host Biomphalaria alexandrina, as well as recent field collected data were gathered and incorporated as separate themes. Model validation was done using data collected on snail population bionomics–infection rates, water quality, underground water table and cercariometry at 13 hydrologically representative sites. The role of soil type, water table and water quality was studied at 79 of 154 rural health unit sites. The model permitted retrieval of relevant data by RHU point location. For the first time in Egypt, the Kafr El-Sheikh GIS schistosoma prediction model can support MOH efforts to make more accurate control program decisions based on environmental predilection sites of endemic Schistosomiasis mansoni. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Advanced very high resolution radiometer; Biomphalaria alexandrina; Egypt; Geographic information systems; Hydrology; Landsat; Models; Remote sensing; Schistosoma mansoni; Vegetation index
* Corresponding author. Tel.: + 202-5720478; fax: + 202-5725240. E-mail address: [email protected] (M.M. El-Bahy). 0001-706X/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 1 - 7 0 6 X ( 0 1 ) 0 0 1 0 2 - 4
50
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
1. Introduction Schistosomiasis, the most important endemic snail-borne disease in Egypt, has plagued its people since ancient times. Schistosoma spp. are patchy in distribution and often quite heterogeneous in terms of infection prevalence between individual sites in the same local area. This has been attributed to environmental effects (Malek, 1958), local snail–parasite genetic compatibility (Mulvey and Vrijenhoek, 1982) or to sociologic factors, especially agricultural practices and proximity to strong endemic foci (Webbe and El Hak, 1990). A few foci of high prevalence may serve as reservoirs for re-infection of well-controlled areas. Identification of infected foci for both human hosts and vector snails in Egypt still depends on ground survey methods for data collection, usually by a large number of people grouped in teams. Major limitations on the accuracy of current survey methods used by the Ministry of Health are related to comparability of sampling in relation to the season of the year, snail population– habitat dynamics, time since prior treatment and the experience, skill and accuracy of the workers on different field teams. The availability of computer-based geographic analysis methods and satellite sensor technology has made it possible to measure ecological factors that define the suitability of the environment for propagation of Schistosoma – snail host life cycles at a given site. Satellite remote sensing methods permit evaluation of spatial and temporal variables in the environment and make it possible to obtain standardized environmental data measures over large geographic areas, including temperature, moisture regime and vegetation index. This has led to emerging interest in the use of environmental satellite assessment tools to determine the distribution and predilection sites of vector-borne diseases and to efforts to develop predictive models that will enable design of more effective control programs. Egypt has unique advantages for remote sensing data applications, including the relatively stable, irrigation-based environment in agricultural areas and the frequency of clear, cloud-free days that allow for quality images to be recorded
throughout the year. Malone et al. (1994), indicated that diurnal temperature difference measurements (dT) reflect surface hydrological factors that are related to the availability of snail-intermediate hosts. This study provided the impetus for a 4-year study in the late 1990’s funded by the Egyptian Ministry of Health — USAID Schistosomiasis Research Project on the development of environmental risk assessment methods based on GIS and remote sensing technologies. A GIS model was constructed for schistosomiasis disease prediction in the Kafr El-Sheikh governorate in the northern Nile Delta that focused on the use of environmental data and new and archival data on prevalence and snail intermediate hosts (AbdelRahman et al., 1998; Malone et al., 1997a). The aim was to develop satellite-based environmental risk assessment methods and to initiate use of the resulting GIS model for integrated schistosomiasis control programs in Egypt.
2. Methods and materials
2.1. Area of study The Kafr El-Sheik governorate was selected for the study because of its high Schistosoma infection prevalence rates and the presence of a broad range of dT values that could be clearly divided into three thermal-hydrological domains: wet, moist and dry.
2.2. The ‘regional scale’ risk model Cloud-free digital images of Egypt, from the advanced very high resolution radiometer (AVHRR), were acquired from the global archives of the National Oceanic and Atmospheric Administration (NOAA, Washington, DC) NOAA-11 satellite and processed on a Hewlett-Packard 9000-825SRX UNIX workstation using the TeraScan software system (SeaSpace, San Diego, CA). Channel 4 thermal infrared day–night image pairs from the NOAA11 satellite were calibrated, georeferenced, co-registered and used to create daytime temperature maximum, night-time temperature minimum and
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
diurnal temperature difference maps (Huh, 1991). Channel 1 visible and Channel 2 near infrared data was use to produce eight NDVI maps. NDVI maps were used to create a template of cultivated areas; dT values were then calculated for 5× 5 pixel areas centered on 4l sites from 1935, 1983 and 1990 human infection prevalence surveys of the Nile delta. The annual time series of 12 dT maps in the cultivated zone (August, 1990– October, 1991) was used to confirm and extend positive results that dT maps can be used to define surface hydrological conditions that favor a high risk of schistosomiasis. Thermal patterns defined in dT maps were analyzed individually and as seasonal and annual composites for correlation with Schistosoma prevalence and field environ-
51
mental data and to create the regional scale risk model.
2.3. Regional field 6alidation studies Using dT maps derived from the 1990–91 monthly time series, 13 villages were identified in representative wet, moist and dry thermal-moisture domains (Fig. 1) in the Kafr El Sheikh governorate for field study. Waterways within a 1km radius, centered on each village, were mapped, divided into collection stations 50–100 m apart and examined for snail number, distribution and snail infection rates. One infected water contact site per village was identified for a more intensive monthly study of snail population bio-
Fig. 1. Regional risk map for Schistosoma mansoni in Kafr El Sheikh based on diurnal temperature difference (dT) and NDVI (to eliminate non-cultivated areas) prepared from a composite of a 12 month time series of AVHRR day – night image pairs. White circles are locations of rural health units (RHU) included in the control program Schistosoma infection prevalence surveillance. Purple squares are RHU sites used for field validation studies on water contact point snail microhabitat characteristics, soil type, water table and snail – schistosome transmission dynamics. The large white square represents the 5 ×5 km area, centered on an example RHU, that was used to extract mean dT and tasseled cap (Tcap) values for each of the 159 RHU.
52
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
nomics –infection rates, water quality, underground water table and cercariometry. Maps on water table and soil type for the entire governorate were prepared.
2.4. The local ‘6illage scale’ GIS Risk Model The village scale risk model was developed, with additional field study data, to validate and extend regional scale risk model results. A Landsat TM satellite data scene (WGS Path 177, Row 38 from May 15, 1990, 185×185 km), covering the Kafr El Sheikh governorate, was spectrally enhanced and classified using ERDAS Imagine (Atlanta, GA) to create various band combinations and classification indices, including Channel 5, Channel 6 thermal infrared data, Channel 1/5/ 7, NDVI and the tasseled Cap (Tcap) transformation by use of ERDAS image analysis procedures and algorithms (Malone et al., 1998; Thompson et al., 1997). Combining data available from the regional model with Landsat TM-derived data, digitized paper map data and additional field study data, the digital resource data available for the local scale GIS risk model included: 1. landsat TM data for May 15, 1990, which served as the basemap (30 m resolution at earth surface, which approximates a 1:50 000 scale classical paper map); 2. mean dT data extracted from 5×5 km areas centered on survey points in the AVHRRderived regional risk maps based on dT; 3. feature data from a digitized 1:250 000 scale paper map covering Kafr El Shiekh (US Army, 1953); 4. project generated soil type– water table– salinity maps updated in 1996 and 1997 by survey data at 79 RHU sites; 5. 1: 10,000 maps of canals, drains and summer crops of 28 selected areas from the Egyptian Survey Authority; 6. prevalence datasets, including (a) 1991– 95 MOH control program prevalence data for 154 rural health units (RHU) in Kafr El Sheikh, (b) published 1935, 1983 and 1990 survey data (41 sites) for the Nile delta, and (c) the Schistosomiasis Research Project EPI-1,2,3
survey for eight village-ezbat clusters in the Kafr El Sheikh governorate; 7. survey data on snail host distribution, Schistosoma infection rates and seasonal bionomics from current surveys and data archives of the MOH Snail Control Unit.
3. Results and discussion
3.1. Regional scale GIS model studies A comparison was made of 1935, 1983 and 1990 national survey prevalence values for 41 survey sites in the Nile delta, MOH snail distribution survey data for the Nile Delta, and dT values extracted from an annual composite map prepared from the monthly 1990–91 AVHRR sensor time-series. Results confirmed that dT was related to the regional thermal-hydrology regime and risk of S. mansoni in the Nile delta (Abdel-Rahman et al., 1997, 1998; Malone et al., 1996, 1997a, 1997b, 1997c, 1998). In a study of district level snail host distribution in the Nile delta, significantly larger numbers of Biomphalaria alexandrina were found in 1993–95 annual spring surveys by the MOH Snail Control Unit at sites with low (wetter) dT values. These findings were unique in focusing on satellite dT thermal domains, as an indicator of the suitability of surface hydrology regimes for parasite propagation and development. Annual and seasonal composite map dT patterns were proposed to reflect stable landform, soils and climate–irrigation–water table factors that influence environmental suitability for Schistosoma (Fig. 1).
3.2. Local ‘Village le6el’ GIS model studies in the Kafr El Sheikh go6ernorate Field data from the 13 sites studied to validate the regional model in Kafr El Sheikh were consistent with earlier observations on the inverse relationship between dT values and S. mansoni prevalence (PB 0.05), snail population abundance, and snail infection rates (Abdel-Rahman et al., 1997). Values for dT were then extracted for 5× 5 km areas centered on each of 154 RHU sites
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
in the Kafr El Sheikh governorate and used as a component, with other databases, for construction of a ‘village level’ GIS model designed to utilize satellite sensor-derived environmental risk factors, with current MOH control program infection prevalence data and snail control data, for control program management by the MOH. Results showed that Tcap values derived by unsupervised classification of the Landsat TM satellite image were statistically related to infection prevalence at 154 RHU sites listed in the 5-year (1990–95) school-based control program prevalence database. Tcap values are calculated using data from six of the seven Landsat channels (excluding Channel 6 thermal infrared data) to produce wetness, greenness and brightness (dryness) indices; each pixel is assigned a vector value based on the variable proportions of signals for each these three indices (ERDAS Field Guide, 1995). Ten map classes ranging from maximum wetness through greenness and maximum dryness were created by classification of the Tcap transformation image. Tcap values were analyzed for statistical relationships to RHU prevalence for 5 × 5 km areas centered on 51 randomly selected villages (Fig. 2a). Class 3 and 4 (water), Class 5 (wet vegetation) and Class 10 (dry) were significantly related to prevalence and were used in further analysis. A linear regression prediction model was constructed using: (1) a 5-year MOH rural health unit school control program prevalence database; and (2) Landsat Tcap data classes as independent variables alone and in combination with dT values. Use of the four Tcap classes combined with dT produced a linear regression model that explained 74% of the variation in RHU prevalence (r 2 value = 0.74; P B0.05). Results suggested that environmental features detected by Tcap, in combination with dT, can be used to predict S. mansoni prevalence in irrigated agricultural areas of the Nile delta. The local ‘village scale’ GIS was constructed for schistosomiasis in the Kafr El Sheikh governorate using ArcView3 GIS and ERDAS Imagine software and included image maps, feature maps and Dbase IV files on field data and selected data values extracted from maps. The GIS included thematic layers based on a channel 7,5,1 combina-
53
tion map from the Landsat TM scene which served as the basemap (Fig. 2b), the dT maps prepared from the monthly time series of 1990–91 AVHRR data, Tcap ten-class (Fig. 2a) and fourclass maps used for model development above, and the 1991– 95 RHU school prevalence database from MOH. Using tasseled cap, dT and RHU databases, model predicted RHU prevalence values were statistically derived. Attribute data on dT values and dT/TCap model predicted prevalence values (within 95% confidence limits) prepared for 5× 5 areas centered on RHU’s were extracted and included as attribute data in the master RHU prevalence database. The Tcap/dT model was developed using data from 51 villages, then tested against data from 25 randomly selected villages. The model’s 95% confidence prediction interval contained 84% of the annual prevalence rates of the 25 villages. Environmental features detected by dT and tasseled cap have been included in a GIS control program management model (Fig. 3) for evaluating the risk of S. mansoni under irrigated agricultural conditions in the Kafr El Sheikh governorate. The model and ArcView software query functions are intended for use by control program managers to rapidly assess risk factors and make control program decisions at any given RHU locality in the governorate map, based on the reported level of infection in annual MOH surveys. The GIS model constitutes an easily understood, dynamic ‘database with maps’ that can be updated with additional annual control program data to accumulate a time series archive that contains all previous available data by location.
3.3. Other databases Field validation data on snail populations (distribution, abundance and infection rates), cercariometry, water quality, depth to water table and soil texture–soil chemistry characteristics at 13 sites in representative moisture domains were added as a separate database. Data from a separate field validation study at 79 of the 159 RHU sites, the same sites as used for soil surveys, was added as a third database to record water table,
54
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
Fig. 2. (a) Tasseled cap classification of the Kafr El Sheikh study area. Four classes of ten were shown to be significantly related to the 1991 – 95 RHU prevalence database. The area occupied by the four classes in a 5 × 5 km area centered on 75 RHU (one brightness class, two wetness classes, and one greenness class) were used, with dT, to develop the environmental risk prediction model. The greenness class is thought to represent lush vegetation in wet fields in May after spring planting, mainly rice. (b) ArcView3 GIS printout of the Landsat TM, Band 7-5-1 basemap (equivalent to a 1: 50,000 map) overlaid with points linked to a master database that includes: (1) the 1991 – 95 RHU infection prevalence database; (2) selected environmental characteristics extracted from 5 × 5 km areas centered on 159 RHU points; (3) project-generated water table, salinity and snail survey data, and (4) model-predicted prevalence values. Cites and towns are clearly visible as dry (white) areas in cultivated green zones. Purple squares are linked to a 13-site field validation database.
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
55
Fig. 3. Schematic representation of the structure of the MOH GIS control program management model for schistosomiasis.
soil type and soil chemistry data linked to GIS village point locations. Available MOH snail control maps/records were included as an archive of separate individual scanned hand-drawn survey maps by the MOH Snail Eradication Office of the Kafr El Sheikh Governorate. The latter, with the 79-site snail survey data, may be used to assess risk factors related to snail populations (distribution, abundance and infection rates) and water contact points.
3.4. Pro6isional use of the GIS risk assessment model The GIS can be used by the MOH: (1) as a control program management database; and (2) as an investigative tool to identify environmental risk factors and control program requirements at sites that fall outside the expected patterns of response to control measures. MOH control program policy (El Khoby, 1998) follows guidelines that RHU with infection prevalence rates exceeding 20% in the most recent school survey should receive treatment for at-risk 10– 14 year old school children, followed by blanket treatment with molluscicide (Baylucide) of all waterways
within a 1 km radius of affected villages. If the reported prevalence exceeds 30%, treatment with praziquantel is given to all individuals in all age groups inhabiting affected villages. An example scenario of the use of the environmental risk model, by control program managers in Egypt, as a decision support system is shown in Fig. 3. All RHU with annual reported prevalence values reported to be above the 20% threshold by the field team and any RHU prevalence reports that are much higher than the ‘predicted’ prevalence of the environmental risk model based on dT/Tcap and prior RHU prevalence databases would be flagged for further investigation of reasons for the variance. If the predicted dT/Tcap prevalence rate was consistent with the high prevalence rates observed (e.g. within one standard deviation), then underlying environment risk factors favoring propagation and transmission of the S. mansoni– Biomphalaria alexandrina life cycle would be considered the probable reason. If model predicted prevalence is lower than that most recently observed in the field, then other factors can be assumed to be present, for example, social behavior, unusually heavy contamination of water contact sites or a large number of infected water contact points.
56
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
Village snail survey maps prepared in the recent past or in response to the annual prevalence report can be electronically scanned and added to the village snail control map/record archive as TIF or JPEG records. These maps, although hand drawn by snail control teams, provide a record of: (1) the location of major water contact points frequented by village inhabitants; (2) village waterway survey points in canals and drains (50– 100 m apart) in which infected snails were found in three dip net samples; (3) a record of molluscicide treatment. High prevalence rates, not consistent with high environmental risk, may prompt investigation by snail control teams for evidence of unusual infected snail exposure that may be corrected (i.e. high number of water contact sites, proximity to heavy human fecal/urine contamination sources, etc.). ArcView software supports point and click retrieval of the scanned hand drawn maps, digital photographs as well as detailed database records that correspond to point locations. If neither snail survey data nor predicted environmental risk support the unusually high observed prevalence, then greater consideration may be given to a review of the effectiveness of the village control program survey teams, the sampling plan of school children/village populations and/or treatment procedures.
3.5. The potential for further model de6elopment In parallel with development of the environmental risk model for Kafr El Sheikh, a separate project funded by the Schistosomiasis Research Project focused on the construction and evaluating use of a GIS database based on school-based versus RHU-based reporting systems in MOH control programs (Mohamed, 1998). Combining this GIS database and the environmental risk GIS would provide a comprehensive system for MOH control of schistosomiasis in a single governorate, with potential expansion of similar systems to other governorates. The more detailed data available for 1996 in the school-based system, as compared to the RHU-based reporting system, can be used as an independent dataset for validation and
further development of the environmental risk model. Directions for further development of the environmental risk component of the combined GIS control program model, may include incorporation of Landsat TM-based soil moisture maps by algorithms described by Crombie et al. (1999) in which soil moisture was related to filariasis risk in the Nile Delta. In that study, dT was found to give generally similar results as methods of satellite-based risk assessment methods for filariasis (Thompson et al., 1996), a finding which supports use of dT in studies reported here for schistosomiasis. Based on the successful use of the GIS risk assessment model in Kafr El Sheikh, the model may be expanded to enable development and use of similar control program models in other governorates. Eight agroclimatic zones have been defined for Egypt (Aboukhalid et al., 1975) that reflect variation in temperature (typically a 4– 6°C temperature difference occurs between upper and lower Egypt), soil type, elevation, water table and other factors that occur in the Nile River Valley. Incorporation of agroecological and climatedriven systems that describe suitability of a given site for schistosomiasis may be useful additions to the model, for ultimate application of GIS control program management systems on a national scale. Acknowledgements This article is dedicated to the memory of the late MS Abdel-Rahman, a medical scientist, veterinarian and academician whose ‘one medicine’ philosophy and devotion to progress in science and education in Egypt provided the team leadership, scientific judgment and collegial insights that made these studies possible. This work was supported in part by the National Institutes of Health, NIAID Grant R15 AI-28192-01 and the Ministry of Health/USAID Schistosomiasis Research Project, Grant 08-04-69. The authors thank M. Shafik, Snail Control Unit, Ministry of Health, Cairo, Egypt, A.H. Wafaa, Endemic Disease Control, Ministry of Health, Cairo, Egypt and Y.A. Hassanein, Endemic Disease Control, Ministry of Health, Cairo, Egypt for their support in completing this research project.
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
References Abdel-Rahman, M.S., El-Bahy, M.M., El-Bahy, N.M., Malone, J.B., 1997. Development and validation of a satellite based geographic information system (GIS) model for epidemiology of Schistosoma risk assessment at snail level in Kafr El-Sheikh governorate. J. Egypt Soc. Parasitol. 27, 299– 316. Abdel-Rahman, M.S., El-Bahy, M.M., Shafik, M., Malone, J.B., El-Bahy, N.M., Wafaa, A.H., Hassanein, Y.A., Mohamed, S.A., 1998. Health maps: a GIS risk model for schistosomiasis control in Kafr El-Sheikh governorate Egypt (abstract). In: SRP International Conference on Schistosomiasis, 15 – 19 March. Aboukhalid, A., Arar, A., Balba, A.M., Bishay, B.G., Kadry, L.T., Rutema, P.E., Taher, A., 1975. Research on crop use, salt affecting soils and drainage in the Arab Republic of Egypt. Food and Agriculture Organization, Middle East Regional Office, Cairo. Crombie, M.K., Gillies, R.R., Arvidson, R.E., Brookmeyer, P., Weil, G.J., Sultan, M., Harb, M., 1999. An application of remotely derived climatological fields for risk assessment of vector-borne diseases: a spatial study of filariasis prevalence in the Nile Delta. Egypt Photogram. Eng. Rem. Sens. 65, 1401 – 1409. El Khoby T. 1998, Vice Director, Ministry of Health, personal communication. ERDAS Field Guide 1995, ERDAS Inc, Atlanta, GA. Huh, O.K., 1991. Limitations and capabilities of the NOAA satellite advanced very high resolution radiometer (AVHRR) for remote sensing of the earth’s surface. Prev. Vet. Med. 11, 45 – 62. Malek, E., 1958. Distribution of the intermediate hosts of bilharziasis in relation to hydrography, with special reference to the Nile Basin and the Sudan. Bull. WHO 18, 691– 734. Malone, J.B., Huh, O.K., Fehler, D.P., Wilson, P.A., Wilensky, D.E., Holmes, R.A., Elmagdoub, A.L., 1994. Temperature data from satellite imagery and the distribution of schistosomiasis in Egypt. Am. J. Trop. Med. Hyg. 50 (7), 14– 722. Malone J.B., Abdel-Rahman M.S., Huh O.K., El-Bahy M.M., Bavia M., Portwood B.J., El-Bahy N.M., Khater E.S., 1996. Geographic information systems and public health hazards of schistosomiasis in the Nile delta. Natural
.
57
Coastal Environmental Hazards, Report of a Workshop held at Louisiana State University, Feb.22 – 23, 1996, Coastal Studies Institute Technical Report TR-654, Louisiana, USA. Malone, J.B., Abdel-Rahman, M.S., El-Bahy, M.M., Huh, O.K., Shafik, M., Bavia, M., 1997a. Geographic information systems and the distribution of Schistosoma mansoni in the Nile Delta. Parasitol. Today 13, 112 – 119. Malone, J.B., Yilma, J.M., El-Bahy, M.M., Huh, O.K., AbdelRahman, M.S., Hansen, J.W., 1997b. Geographic information systems and snail borne disease in Africa. In: Proc. Workshop on Medical and Veterinary Malacology in Africa, Harare, Zimbabwe, 22 – 26 September, 1997, pp. 119 – 136. Malone, J.B., Abdel-Rahman, M.S., El Bahy, M.M., 1997c. A geographic information system for control of schistosomiasis in Kafr El Sheikh governorate, Egypt. Am. J. Trop. Med. Hyg. Suppl. 57, 170. Malone, J.B., Thompson, R.A., Huh, O.K., Braud, D.H., Abdel-Rahman, M.S., El-Bahy, M.M., Shafik, M., Wafaa, A.H., Hassanein, Y.A., 1998. Predicting schistosomiasis risk in Egypt using remote sensing-GIS surrogates of the environment. In: SRP 1998 (final) International Conference on Schistosomiasis. March. Cairo, Egypt, p. 58. Mohamed S.A. 1998, PhD Thesis, Cairo University, Giza, Egypt. Mulvey, M., Vrijenhoek, R., 1982. Population structure in Biomphalaria glabrata: examination of an hypothesis for the patchy distribution of susceptibility to schistosomes. Am. J. Trop. Med. Hyg. 31, 1195 – 2000. Thompson, D.L., Malone, J.B., Harb, M., Faris, R., Huh, O.K., Buck, A.A, Cline, B.L., 1996. Bancroftian filariasis distribution in the southern Nile Delta: correlation with diurnal temperature difference maps from satellite imagery. Emerg. Inf. Dis. 2, 234 – 235. Thompson, R.A., Malone, J.B., Braud, D.H., Abdel-Rahman, M.S., El Bahy, M.M., Wafaad, A.A., Hassanein, Y.Y., 1997. Prediction of Schistosoma mansoni prevalence in the Nile Delta using Landsat TM and AVHRR diurnal temperature difference data. Am. J. Trop. Med. Hyg. 57 (Suppl.), 318. US Army, 1953. Al Mansurah, NH 36-1, AMS Series P502, Edition 1-AMS, 1:250 .000. Webbe, G., El Hak, S., 1990. Program on control of schistosomiasis in Egypt, 1985 – 1988. Trans. R. Soc. Trop. Med. Hyg. 84, 394 – 400.
Review
Geographic information systems as a tool for control program management for schistosomiasis in Egypt M.S. Abdel-Rahman a, M.M. El-Bahy a,*, J.B. Malone b, R.A. Thompson b, N.M. El Bahy c b
a Faculty of Veterinary Medicine, Cairo Uni6ersity, Giza, Egypt Department of Pathobiology, School of Veterinary Medicine, Louisiana State Uni6ersity, Baton Rouge, LA 70803, USA c Faculty of Veterinary Medicine, Monifyia Uni6ersity, Sadat City, Egypt
Abstract During a 4-year study a geographic information system (GIS) risk model was constructed for predicting the relative risk of schistosomiasis in Kafr El-Sheikh governorate, Egypt. A 1-year 1990– 1991 time series on diurnal temperature difference (dT) prepared from the advanced very high resolution radiometer (AVHRR) sensor on the NOAA-11 satellite was used to develop a regional risk model for the Nile delta based on thermal-hydrological domains. A May 15, 1990 Landsat TM scene (path 177, Row 38) was used to develop a local ‘village-scale’ environmental risk model based on higher resolution satellite sensor data (30 m picture element size at earth surface). Four of ten classes derived from a tasseled cap (Tcap) transformation of the Landsat TM scene were shown to be significantly related to a 5-year Schistosoma mansoni prevalence database from the Ministry of Health. A risk model was developed based on dT and the proportional area of the four Tcap classes in 5 km2 buffer zones centered on rural health unit (RHU) reporting units. Available historical data on S. mansoni and its snail host Biomphalaria alexandrina, as well as recent field collected data were gathered and incorporated as separate themes. Model validation was done using data collected on snail population bionomics–infection rates, water quality, underground water table and cercariometry at 13 hydrologically representative sites. The role of soil type, water table and water quality was studied at 79 of 154 rural health unit sites. The model permitted retrieval of relevant data by RHU point location. For the first time in Egypt, the Kafr El-Sheikh GIS schistosoma prediction model can support MOH efforts to make more accurate control program decisions based on environmental predilection sites of endemic Schistosomiasis mansoni. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Advanced very high resolution radiometer; Biomphalaria alexandrina; Egypt; Geographic information systems; Hydrology; Landsat; Models; Remote sensing; Schistosoma mansoni; Vegetation index
* Corresponding author. Tel.: + 202-5720478; fax: + 202-5725240. E-mail address: [email protected] (M.M. El-Bahy). 0001-706X/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 1 - 7 0 6 X ( 0 1 ) 0 0 1 0 2 - 4
50
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
1. Introduction Schistosomiasis, the most important endemic snail-borne disease in Egypt, has plagued its people since ancient times. Schistosoma spp. are patchy in distribution and often quite heterogeneous in terms of infection prevalence between individual sites in the same local area. This has been attributed to environmental effects (Malek, 1958), local snail–parasite genetic compatibility (Mulvey and Vrijenhoek, 1982) or to sociologic factors, especially agricultural practices and proximity to strong endemic foci (Webbe and El Hak, 1990). A few foci of high prevalence may serve as reservoirs for re-infection of well-controlled areas. Identification of infected foci for both human hosts and vector snails in Egypt still depends on ground survey methods for data collection, usually by a large number of people grouped in teams. Major limitations on the accuracy of current survey methods used by the Ministry of Health are related to comparability of sampling in relation to the season of the year, snail population– habitat dynamics, time since prior treatment and the experience, skill and accuracy of the workers on different field teams. The availability of computer-based geographic analysis methods and satellite sensor technology has made it possible to measure ecological factors that define the suitability of the environment for propagation of Schistosoma – snail host life cycles at a given site. Satellite remote sensing methods permit evaluation of spatial and temporal variables in the environment and make it possible to obtain standardized environmental data measures over large geographic areas, including temperature, moisture regime and vegetation index. This has led to emerging interest in the use of environmental satellite assessment tools to determine the distribution and predilection sites of vector-borne diseases and to efforts to develop predictive models that will enable design of more effective control programs. Egypt has unique advantages for remote sensing data applications, including the relatively stable, irrigation-based environment in agricultural areas and the frequency of clear, cloud-free days that allow for quality images to be recorded
throughout the year. Malone et al. (1994), indicated that diurnal temperature difference measurements (dT) reflect surface hydrological factors that are related to the availability of snail-intermediate hosts. This study provided the impetus for a 4-year study in the late 1990’s funded by the Egyptian Ministry of Health — USAID Schistosomiasis Research Project on the development of environmental risk assessment methods based on GIS and remote sensing technologies. A GIS model was constructed for schistosomiasis disease prediction in the Kafr El-Sheikh governorate in the northern Nile Delta that focused on the use of environmental data and new and archival data on prevalence and snail intermediate hosts (AbdelRahman et al., 1998; Malone et al., 1997a). The aim was to develop satellite-based environmental risk assessment methods and to initiate use of the resulting GIS model for integrated schistosomiasis control programs in Egypt.
2. Methods and materials
2.1. Area of study The Kafr El-Sheik governorate was selected for the study because of its high Schistosoma infection prevalence rates and the presence of a broad range of dT values that could be clearly divided into three thermal-hydrological domains: wet, moist and dry.
2.2. The ‘regional scale’ risk model Cloud-free digital images of Egypt, from the advanced very high resolution radiometer (AVHRR), were acquired from the global archives of the National Oceanic and Atmospheric Administration (NOAA, Washington, DC) NOAA-11 satellite and processed on a Hewlett-Packard 9000-825SRX UNIX workstation using the TeraScan software system (SeaSpace, San Diego, CA). Channel 4 thermal infrared day–night image pairs from the NOAA11 satellite were calibrated, georeferenced, co-registered and used to create daytime temperature maximum, night-time temperature minimum and
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
diurnal temperature difference maps (Huh, 1991). Channel 1 visible and Channel 2 near infrared data was use to produce eight NDVI maps. NDVI maps were used to create a template of cultivated areas; dT values were then calculated for 5× 5 pixel areas centered on 4l sites from 1935, 1983 and 1990 human infection prevalence surveys of the Nile delta. The annual time series of 12 dT maps in the cultivated zone (August, 1990– October, 1991) was used to confirm and extend positive results that dT maps can be used to define surface hydrological conditions that favor a high risk of schistosomiasis. Thermal patterns defined in dT maps were analyzed individually and as seasonal and annual composites for correlation with Schistosoma prevalence and field environ-
51
mental data and to create the regional scale risk model.
2.3. Regional field 6alidation studies Using dT maps derived from the 1990–91 monthly time series, 13 villages were identified in representative wet, moist and dry thermal-moisture domains (Fig. 1) in the Kafr El Sheikh governorate for field study. Waterways within a 1km radius, centered on each village, were mapped, divided into collection stations 50–100 m apart and examined for snail number, distribution and snail infection rates. One infected water contact site per village was identified for a more intensive monthly study of snail population bio-
Fig. 1. Regional risk map for Schistosoma mansoni in Kafr El Sheikh based on diurnal temperature difference (dT) and NDVI (to eliminate non-cultivated areas) prepared from a composite of a 12 month time series of AVHRR day – night image pairs. White circles are locations of rural health units (RHU) included in the control program Schistosoma infection prevalence surveillance. Purple squares are RHU sites used for field validation studies on water contact point snail microhabitat characteristics, soil type, water table and snail – schistosome transmission dynamics. The large white square represents the 5 ×5 km area, centered on an example RHU, that was used to extract mean dT and tasseled cap (Tcap) values for each of the 159 RHU.
52
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
nomics –infection rates, water quality, underground water table and cercariometry. Maps on water table and soil type for the entire governorate were prepared.
2.4. The local ‘6illage scale’ GIS Risk Model The village scale risk model was developed, with additional field study data, to validate and extend regional scale risk model results. A Landsat TM satellite data scene (WGS Path 177, Row 38 from May 15, 1990, 185×185 km), covering the Kafr El Sheikh governorate, was spectrally enhanced and classified using ERDAS Imagine (Atlanta, GA) to create various band combinations and classification indices, including Channel 5, Channel 6 thermal infrared data, Channel 1/5/ 7, NDVI and the tasseled Cap (Tcap) transformation by use of ERDAS image analysis procedures and algorithms (Malone et al., 1998; Thompson et al., 1997). Combining data available from the regional model with Landsat TM-derived data, digitized paper map data and additional field study data, the digital resource data available for the local scale GIS risk model included: 1. landsat TM data for May 15, 1990, which served as the basemap (30 m resolution at earth surface, which approximates a 1:50 000 scale classical paper map); 2. mean dT data extracted from 5×5 km areas centered on survey points in the AVHRRderived regional risk maps based on dT; 3. feature data from a digitized 1:250 000 scale paper map covering Kafr El Shiekh (US Army, 1953); 4. project generated soil type– water table– salinity maps updated in 1996 and 1997 by survey data at 79 RHU sites; 5. 1: 10,000 maps of canals, drains and summer crops of 28 selected areas from the Egyptian Survey Authority; 6. prevalence datasets, including (a) 1991– 95 MOH control program prevalence data for 154 rural health units (RHU) in Kafr El Sheikh, (b) published 1935, 1983 and 1990 survey data (41 sites) for the Nile delta, and (c) the Schistosomiasis Research Project EPI-1,2,3
survey for eight village-ezbat clusters in the Kafr El Sheikh governorate; 7. survey data on snail host distribution, Schistosoma infection rates and seasonal bionomics from current surveys and data archives of the MOH Snail Control Unit.
3. Results and discussion
3.1. Regional scale GIS model studies A comparison was made of 1935, 1983 and 1990 national survey prevalence values for 41 survey sites in the Nile delta, MOH snail distribution survey data for the Nile Delta, and dT values extracted from an annual composite map prepared from the monthly 1990–91 AVHRR sensor time-series. Results confirmed that dT was related to the regional thermal-hydrology regime and risk of S. mansoni in the Nile delta (Abdel-Rahman et al., 1997, 1998; Malone et al., 1996, 1997a, 1997b, 1997c, 1998). In a study of district level snail host distribution in the Nile delta, significantly larger numbers of Biomphalaria alexandrina were found in 1993–95 annual spring surveys by the MOH Snail Control Unit at sites with low (wetter) dT values. These findings were unique in focusing on satellite dT thermal domains, as an indicator of the suitability of surface hydrology regimes for parasite propagation and development. Annual and seasonal composite map dT patterns were proposed to reflect stable landform, soils and climate–irrigation–water table factors that influence environmental suitability for Schistosoma (Fig. 1).
3.2. Local ‘Village le6el’ GIS model studies in the Kafr El Sheikh go6ernorate Field data from the 13 sites studied to validate the regional model in Kafr El Sheikh were consistent with earlier observations on the inverse relationship between dT values and S. mansoni prevalence (PB 0.05), snail population abundance, and snail infection rates (Abdel-Rahman et al., 1997). Values for dT were then extracted for 5× 5 km areas centered on each of 154 RHU sites
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
in the Kafr El Sheikh governorate and used as a component, with other databases, for construction of a ‘village level’ GIS model designed to utilize satellite sensor-derived environmental risk factors, with current MOH control program infection prevalence data and snail control data, for control program management by the MOH. Results showed that Tcap values derived by unsupervised classification of the Landsat TM satellite image were statistically related to infection prevalence at 154 RHU sites listed in the 5-year (1990–95) school-based control program prevalence database. Tcap values are calculated using data from six of the seven Landsat channels (excluding Channel 6 thermal infrared data) to produce wetness, greenness and brightness (dryness) indices; each pixel is assigned a vector value based on the variable proportions of signals for each these three indices (ERDAS Field Guide, 1995). Ten map classes ranging from maximum wetness through greenness and maximum dryness were created by classification of the Tcap transformation image. Tcap values were analyzed for statistical relationships to RHU prevalence for 5 × 5 km areas centered on 51 randomly selected villages (Fig. 2a). Class 3 and 4 (water), Class 5 (wet vegetation) and Class 10 (dry) were significantly related to prevalence and were used in further analysis. A linear regression prediction model was constructed using: (1) a 5-year MOH rural health unit school control program prevalence database; and (2) Landsat Tcap data classes as independent variables alone and in combination with dT values. Use of the four Tcap classes combined with dT produced a linear regression model that explained 74% of the variation in RHU prevalence (r 2 value = 0.74; P B0.05). Results suggested that environmental features detected by Tcap, in combination with dT, can be used to predict S. mansoni prevalence in irrigated agricultural areas of the Nile delta. The local ‘village scale’ GIS was constructed for schistosomiasis in the Kafr El Sheikh governorate using ArcView3 GIS and ERDAS Imagine software and included image maps, feature maps and Dbase IV files on field data and selected data values extracted from maps. The GIS included thematic layers based on a channel 7,5,1 combina-
53
tion map from the Landsat TM scene which served as the basemap (Fig. 2b), the dT maps prepared from the monthly time series of 1990–91 AVHRR data, Tcap ten-class (Fig. 2a) and fourclass maps used for model development above, and the 1991– 95 RHU school prevalence database from MOH. Using tasseled cap, dT and RHU databases, model predicted RHU prevalence values were statistically derived. Attribute data on dT values and dT/TCap model predicted prevalence values (within 95% confidence limits) prepared for 5× 5 areas centered on RHU’s were extracted and included as attribute data in the master RHU prevalence database. The Tcap/dT model was developed using data from 51 villages, then tested against data from 25 randomly selected villages. The model’s 95% confidence prediction interval contained 84% of the annual prevalence rates of the 25 villages. Environmental features detected by dT and tasseled cap have been included in a GIS control program management model (Fig. 3) for evaluating the risk of S. mansoni under irrigated agricultural conditions in the Kafr El Sheikh governorate. The model and ArcView software query functions are intended for use by control program managers to rapidly assess risk factors and make control program decisions at any given RHU locality in the governorate map, based on the reported level of infection in annual MOH surveys. The GIS model constitutes an easily understood, dynamic ‘database with maps’ that can be updated with additional annual control program data to accumulate a time series archive that contains all previous available data by location.
3.3. Other databases Field validation data on snail populations (distribution, abundance and infection rates), cercariometry, water quality, depth to water table and soil texture–soil chemistry characteristics at 13 sites in representative moisture domains were added as a separate database. Data from a separate field validation study at 79 of the 159 RHU sites, the same sites as used for soil surveys, was added as a third database to record water table,
54
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
Fig. 2. (a) Tasseled cap classification of the Kafr El Sheikh study area. Four classes of ten were shown to be significantly related to the 1991 – 95 RHU prevalence database. The area occupied by the four classes in a 5 × 5 km area centered on 75 RHU (one brightness class, two wetness classes, and one greenness class) were used, with dT, to develop the environmental risk prediction model. The greenness class is thought to represent lush vegetation in wet fields in May after spring planting, mainly rice. (b) ArcView3 GIS printout of the Landsat TM, Band 7-5-1 basemap (equivalent to a 1: 50,000 map) overlaid with points linked to a master database that includes: (1) the 1991 – 95 RHU infection prevalence database; (2) selected environmental characteristics extracted from 5 × 5 km areas centered on 159 RHU points; (3) project-generated water table, salinity and snail survey data, and (4) model-predicted prevalence values. Cites and towns are clearly visible as dry (white) areas in cultivated green zones. Purple squares are linked to a 13-site field validation database.
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
55
Fig. 3. Schematic representation of the structure of the MOH GIS control program management model for schistosomiasis.
soil type and soil chemistry data linked to GIS village point locations. Available MOH snail control maps/records were included as an archive of separate individual scanned hand-drawn survey maps by the MOH Snail Eradication Office of the Kafr El Sheikh Governorate. The latter, with the 79-site snail survey data, may be used to assess risk factors related to snail populations (distribution, abundance and infection rates) and water contact points.
3.4. Pro6isional use of the GIS risk assessment model The GIS can be used by the MOH: (1) as a control program management database; and (2) as an investigative tool to identify environmental risk factors and control program requirements at sites that fall outside the expected patterns of response to control measures. MOH control program policy (El Khoby, 1998) follows guidelines that RHU with infection prevalence rates exceeding 20% in the most recent school survey should receive treatment for at-risk 10– 14 year old school children, followed by blanket treatment with molluscicide (Baylucide) of all waterways
within a 1 km radius of affected villages. If the reported prevalence exceeds 30%, treatment with praziquantel is given to all individuals in all age groups inhabiting affected villages. An example scenario of the use of the environmental risk model, by control program managers in Egypt, as a decision support system is shown in Fig. 3. All RHU with annual reported prevalence values reported to be above the 20% threshold by the field team and any RHU prevalence reports that are much higher than the ‘predicted’ prevalence of the environmental risk model based on dT/Tcap and prior RHU prevalence databases would be flagged for further investigation of reasons for the variance. If the predicted dT/Tcap prevalence rate was consistent with the high prevalence rates observed (e.g. within one standard deviation), then underlying environment risk factors favoring propagation and transmission of the S. mansoni– Biomphalaria alexandrina life cycle would be considered the probable reason. If model predicted prevalence is lower than that most recently observed in the field, then other factors can be assumed to be present, for example, social behavior, unusually heavy contamination of water contact sites or a large number of infected water contact points.
56
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
Village snail survey maps prepared in the recent past or in response to the annual prevalence report can be electronically scanned and added to the village snail control map/record archive as TIF or JPEG records. These maps, although hand drawn by snail control teams, provide a record of: (1) the location of major water contact points frequented by village inhabitants; (2) village waterway survey points in canals and drains (50– 100 m apart) in which infected snails were found in three dip net samples; (3) a record of molluscicide treatment. High prevalence rates, not consistent with high environmental risk, may prompt investigation by snail control teams for evidence of unusual infected snail exposure that may be corrected (i.e. high number of water contact sites, proximity to heavy human fecal/urine contamination sources, etc.). ArcView software supports point and click retrieval of the scanned hand drawn maps, digital photographs as well as detailed database records that correspond to point locations. If neither snail survey data nor predicted environmental risk support the unusually high observed prevalence, then greater consideration may be given to a review of the effectiveness of the village control program survey teams, the sampling plan of school children/village populations and/or treatment procedures.
3.5. The potential for further model de6elopment In parallel with development of the environmental risk model for Kafr El Sheikh, a separate project funded by the Schistosomiasis Research Project focused on the construction and evaluating use of a GIS database based on school-based versus RHU-based reporting systems in MOH control programs (Mohamed, 1998). Combining this GIS database and the environmental risk GIS would provide a comprehensive system for MOH control of schistosomiasis in a single governorate, with potential expansion of similar systems to other governorates. The more detailed data available for 1996 in the school-based system, as compared to the RHU-based reporting system, can be used as an independent dataset for validation and
further development of the environmental risk model. Directions for further development of the environmental risk component of the combined GIS control program model, may include incorporation of Landsat TM-based soil moisture maps by algorithms described by Crombie et al. (1999) in which soil moisture was related to filariasis risk in the Nile Delta. In that study, dT was found to give generally similar results as methods of satellite-based risk assessment methods for filariasis (Thompson et al., 1996), a finding which supports use of dT in studies reported here for schistosomiasis. Based on the successful use of the GIS risk assessment model in Kafr El Sheikh, the model may be expanded to enable development and use of similar control program models in other governorates. Eight agroclimatic zones have been defined for Egypt (Aboukhalid et al., 1975) that reflect variation in temperature (typically a 4– 6°C temperature difference occurs between upper and lower Egypt), soil type, elevation, water table and other factors that occur in the Nile River Valley. Incorporation of agroecological and climatedriven systems that describe suitability of a given site for schistosomiasis may be useful additions to the model, for ultimate application of GIS control program management systems on a national scale. Acknowledgements This article is dedicated to the memory of the late MS Abdel-Rahman, a medical scientist, veterinarian and academician whose ‘one medicine’ philosophy and devotion to progress in science and education in Egypt provided the team leadership, scientific judgment and collegial insights that made these studies possible. This work was supported in part by the National Institutes of Health, NIAID Grant R15 AI-28192-01 and the Ministry of Health/USAID Schistosomiasis Research Project, Grant 08-04-69. The authors thank M. Shafik, Snail Control Unit, Ministry of Health, Cairo, Egypt, A.H. Wafaa, Endemic Disease Control, Ministry of Health, Cairo, Egypt and Y.A. Hassanein, Endemic Disease Control, Ministry of Health, Cairo, Egypt for their support in completing this research project.
M.S. Abdel-Rahman et al. / Acta Tropica 79 (2001) 49–57
References Abdel-Rahman, M.S., El-Bahy, M.M., El-Bahy, N.M., Malone, J.B., 1997. Development and validation of a satellite based geographic information system (GIS) model for epidemiology of Schistosoma risk assessment at snail level in Kafr El-Sheikh governorate. J. Egypt Soc. Parasitol. 27, 299– 316. Abdel-Rahman, M.S., El-Bahy, M.M., Shafik, M., Malone, J.B., El-Bahy, N.M., Wafaa, A.H., Hassanein, Y.A., Mohamed, S.A., 1998. Health maps: a GIS risk model for schistosomiasis control in Kafr El-Sheikh governorate Egypt (abstract). In: SRP International Conference on Schistosomiasis, 15 – 19 March. Aboukhalid, A., Arar, A., Balba, A.M., Bishay, B.G., Kadry, L.T., Rutema, P.E., Taher, A., 1975. Research on crop use, salt affecting soils and drainage in the Arab Republic of Egypt. Food and Agriculture Organization, Middle East Regional Office, Cairo. Crombie, M.K., Gillies, R.R., Arvidson, R.E., Brookmeyer, P., Weil, G.J., Sultan, M., Harb, M., 1999. An application of remotely derived climatological fields for risk assessment of vector-borne diseases: a spatial study of filariasis prevalence in the Nile Delta. Egypt Photogram. Eng. Rem. Sens. 65, 1401 – 1409. El Khoby T. 1998, Vice Director, Ministry of Health, personal communication. ERDAS Field Guide 1995, ERDAS Inc, Atlanta, GA. Huh, O.K., 1991. Limitations and capabilities of the NOAA satellite advanced very high resolution radiometer (AVHRR) for remote sensing of the earth’s surface. Prev. Vet. Med. 11, 45 – 62. Malek, E., 1958. Distribution of the intermediate hosts of bilharziasis in relation to hydrography, with special reference to the Nile Basin and the Sudan. Bull. WHO 18, 691– 734. Malone, J.B., Huh, O.K., Fehler, D.P., Wilson, P.A., Wilensky, D.E., Holmes, R.A., Elmagdoub, A.L., 1994. Temperature data from satellite imagery and the distribution of schistosomiasis in Egypt. Am. J. Trop. Med. Hyg. 50 (7), 14– 722. Malone J.B., Abdel-Rahman M.S., Huh O.K., El-Bahy M.M., Bavia M., Portwood B.J., El-Bahy N.M., Khater E.S., 1996. Geographic information systems and public health hazards of schistosomiasis in the Nile delta. Natural
.
57
Coastal Environmental Hazards, Report of a Workshop held at Louisiana State University, Feb.22 – 23, 1996, Coastal Studies Institute Technical Report TR-654, Louisiana, USA. Malone, J.B., Abdel-Rahman, M.S., El-Bahy, M.M., Huh, O.K., Shafik, M., Bavia, M., 1997a. Geographic information systems and the distribution of Schistosoma mansoni in the Nile Delta. Parasitol. Today 13, 112 – 119. Malone, J.B., Yilma, J.M., El-Bahy, M.M., Huh, O.K., AbdelRahman, M.S., Hansen, J.W., 1997b. Geographic information systems and snail borne disease in Africa. In: Proc. Workshop on Medical and Veterinary Malacology in Africa, Harare, Zimbabwe, 22 – 26 September, 1997, pp. 119 – 136. Malone, J.B., Abdel-Rahman, M.S., El Bahy, M.M., 1997c. A geographic information system for control of schistosomiasis in Kafr El Sheikh governorate, Egypt. Am. J. Trop. Med. Hyg. Suppl. 57, 170. Malone, J.B., Thompson, R.A., Huh, O.K., Braud, D.H., Abdel-Rahman, M.S., El-Bahy, M.M., Shafik, M., Wafaa, A.H., Hassanein, Y.A., 1998. Predicting schistosomiasis risk in Egypt using remote sensing-GIS surrogates of the environment. In: SRP 1998 (final) International Conference on Schistosomiasis. March. Cairo, Egypt, p. 58. Mohamed S.A. 1998, PhD Thesis, Cairo University, Giza, Egypt. Mulvey, M., Vrijenhoek, R., 1982. Population structure in Biomphalaria glabrata: examination of an hypothesis for the patchy distribution of susceptibility to schistosomes. Am. J. Trop. Med. Hyg. 31, 1195 – 2000. Thompson, D.L., Malone, J.B., Harb, M., Faris, R., Huh, O.K., Buck, A.A, Cline, B.L., 1996. Bancroftian filariasis distribution in the southern Nile Delta: correlation with diurnal temperature difference maps from satellite imagery. Emerg. Inf. Dis. 2, 234 – 235. Thompson, R.A., Malone, J.B., Braud, D.H., Abdel-Rahman, M.S., El Bahy, M.M., Wafaad, A.A., Hassanein, Y.Y., 1997. Prediction of Schistosoma mansoni prevalence in the Nile Delta using Landsat TM and AVHRR diurnal temperature difference data. Am. J. Trop. Med. Hyg. 57 (Suppl.), 318. US Army, 1953. Al Mansurah, NH 36-1, AMS Series P502, Edition 1-AMS, 1:250 .000. Webbe, G., El Hak, S., 1990. Program on control of schistosomiasis in Egypt, 1985 – 1988. Trans. R. Soc. Trop. Med. Hyg. 84, 394 – 400.