Use of Geographical Information System to Study the Epidemiology of Dengue Haemorrhagic Fever in Thailand By Department of Entomology, Armed Forces Research Institute of Medical Sciences, Bangkok, Thailand
Ratana Sithiprasasna, Kenneth J. Linthicum Kriangkrai Lerdthusnee
National Centre for Genetic Engineering and Biotechnology, Rama VI Road, Bangkok 10400, Thailand
Thomas G. Brewer Headquarters, Armed Forces Research Institute of Medical Sciences, Bangkok, Thailand
Abstract A Geographical Information System (GIS) was used to study the spatial distribution of dengue haemorrhagic fever (DHF) in Thailand. The National Oceanic and Atmospheric Administration (NOAA) satellite data, with a spatial resolution of 1 km, were used to produce a land cover map and calculate the percentage of forest cover in each province of Thailand. GIS was used as an analysis tool to map the distribution of DHF by creating overlays of epidemiological and digitized province data on a NOAA normalized difference vegetation index image of Thailand. The countrywide GIS database demonstrated that DHF incidence was not correlated to the area of forest cover in the provinces of Thailand. Global Positioning System (GPS) instruments were used to map villages involved in dengue epidemiological studies in Tak province. Differentially processed GPS data with a spatial resolution of approximately 1 metre were incorporated into a GIS for analysis and mapping. Databases associated with a village GIS included village name, house number, demographic data on house occupants, Aedes aegypti populations, Ae. aegypti immature breeding sites and seroepidemiology data on house occupants.
Introduction Dengue virus infection can result in a wide range of clinical symptoms, including asymptomatic infection, undifferentiated febrile illness, classical dengue fever, dengue haemorrhagic fever (DHF) and dengue shock syndrome (DSS), which are
associated with significant morbidity and mortality. Factors that are involved in the progression of classical dengue fever to DHF and DSS are not well known(1-3). Dengue infection occurs after the bite of a competent mosquito vector, primarily Aedes aegypti, that is infected with one of the four dengue virus serotypes. An
68
Dengue Bulletin – Vol 21, 1997
Use of Geographical Information System to Study the Epidemiology of DHF in Thailand
analysis of the incidence of DHF cases in Bangkok(4) and experimental studies on the vector efficiency of Ae. aegypti(5) suggest that the number of DHF cases is greatly affected by temperature-induced variation in the vectorial capacity of the vector. The spatial distribution of DHF is dependent upon the presence of Ae. aegypti or another suitable vector and viremic and susceptible human hosts. The Geographical Information System (GIS) databases have been recently used to monitor factors affecting disease transmission(6). Recent studies have demonstrated that satellite imagery, digitized landuse maps and global positioning data are promising for predicting changes in habitats of mosquito vectors as they affect disease transmission(7-10). In this study epidemiological, digital, satellite and Global Positioning System (GPS) data have been incorporated into GIS databases to better understand the spatial distribution of DHF in Thailand.
been shown to be highly correlated with green-leaf biomass and represent the photosynthetic capacity of the area measured. The NDVI data were calculated for NOAA satellite coverages of Thailand for ten dates between December 1992 and April 1993. Data from different dates were then composited by selecting the highest NDVI for each grid cell location. These data have been used as thematic raster base map for displaying DHF incidence by province. The raster map is displayed in a Latitude/Longitude Coordinate system and are georeferenced using the Indian 1975 Datum and Everest 1830 (1975 Definition) Ellipsoid. These data also have been used to calculate forest cover in provinces using unsupervised and supervised classification methods to create vector maps of forest cover (11). Provincial border boundary data for Thailand were digitized using MapInfo software(12). DHF case data were reported by provincial health offices and published by the Ministry of Health, Thailand(13,14). The province population data were provided by the Ministry of Interior, Thailand. The DHF incidence was calculated as the number of cases of reported DHF/100 000 population of the province. The position of houses in three villages in Tak province, Thailand, were mapped using a Trimble Geoexplorer GPS instrument as the rover unit. A Trimble Pathfinder Pro XL GPS instrument was run as a base station at our laboratory in Bangkok (130 45’ 58.878” N Latitude, 1000 32’ 08.504” E Longitude). Both rover and base station units were run simultaneously to allow differential correction of rover data using Trimble software(15). The regression between DHF incidence and per cent 69
Materials and methods Digital remote sensing data were produced by the advanced very high resolution radiometer (AVHRR) sensor on polar-orbiting satellites operated by the National Oceanic and Atmospheric Administration (NOAA). The AVHRR records visible, near-infrared and thermal channels of the electromagnetic spectrum. The characteristics of the NOAA AVHRR have been described previously by Linthicum et al(8). The normalized difference vegetation index (NDVI) analysed in this study is a transformation between data from the visible channel and the nearinfrared channel. The NDVI data have Dengue Bulletin – Vol 21, 1997
Use of Geographical Information System to Study the Epidemiology of DHF in Thailand
forest area were calculated with SPSS software(16).
Results and discussion The distribution of the DHF incidence in 1996 in the provinces of Thailand is shown in Insert A. The relative size of the yellow circle represents the incidence of DHF as indicated in the legend. The darkest green colors on the NOAA/NDVI satellite data-generated base map represent forested areas of Thailand. Approximately 70 000 cases of DHF were reported in 1996. This suggests that there were at least 7 000 000 dengue infections in Thailand during 1996 based upon a conservative ratio of 1 DHF case/100 dengue infections (17). The DHF incidence ranged from 0 to 220/100 000 population. The highest incidence occurred in Tak province in the central-western part of the country bordering Myanmar. High incidence was also observed in several provinces bordering Laos and Cambodia in eastern Thailand. Lowest incidence of transmission was observed in the northernmost and southern-most provinces. It appeared to the authors that most of the DHF occurs in provinces of the central plains and Korat plateau in the east, with the exception of Tak province. Most of these provinces have little forest cover. The relationship between the DHF incidence and forest cover is shown in Figure 1. There is no correlation between disease incidence and forest cover (R2 = 0.001, F = 0.05, P = 0.83). The primary mosquito vector of DHF in Thailand, Ae. aegypti, is a species that is closely associated with humans and breeds in artificial containers associated with human habitation, and it is not ecologically linked 70
to forests. Furthermore, most human habitations in Thailand are not associated with forest. For these reasons and from the distribution of DHF illustrated in Insert A, the authors suspected that there might be a negative correlation between forest cover and DHF incidence. Malaria incidence in Thailand, where the mosquito vector is closely linked to forests (Anopheles virus), has been strongly correlated with forest cover(6). The relation-ship between DHF transmission and ecolo-gical variables might be better evaluated by analysing temporal data with GIS. Figure 2 shows a small portion of the GIS map and database created for one of the three villages that was mapped with a GPS instrument. The GPS data were subjected to differential processing to create a map with a spatial resolution of one metre. These data were then imported into a GIS and a new database created that included village name and house number (portion of data shown), and demographic data on house occupants, adult and immature Aedes aegypti populations, Ae. aegypti immature breeding sites and seroepidemiology data on house occupants (data not shown). These GIS relational databases are being used as a powerful tool to monitor the status of efforts to control Ae. aegypti breeding sites and to evaluate the impact of this control effort on dengue and DHF transmission.
Dengue Bulletin – Vol 21, 1997
Use of Geographical Information System to Study the Epidemiology of DHF in Thailand
Figure 1. Relationship between per cent forest cover and dengue haemorrhagic fever incidence/100 000 population for the year 1993 for all provinces of Thailand.
Figure 2. Portion of GIS map and database for a village in Tak Province, Thailand.
House locations shown by circle with a spatial resolution of 1 metre.
Dengue Bulletin – Vol 21, 1997
71
Use of Geographical Information System to Study the Epidemiology of DHF in Thailand
Historically, maps describing the spatial distribution of DHF and other diseases have been limited to hand-drawn representations on pre-existing maps. GIS, with new advances in image processing and GPS to georeference databases, provides a new and powerful tool to efficiently store, retrieve and interpret DHF databases for epidemiology, ecology and control studies.
7. Hayes RO, Maxwell EL, Mitchell CJ and Woodzick TL. Detection, identification and classification of mosquito larval habitats using remote sensing scanners in earth-orbiting satellites. Bull. WHO. 1985, 63: 361-374. 8. Linthicum KJ, Bailey CL, Davies FG and Tucker CJ. Detection of Rift Valley fever viral activity in Kenya by satellite remote sensing imagery. Science, 1987, 235: 1656-1659. 9. Linthicum KJ, Bailey CL, Tucker CJ, Gordon SW, Logan TM, Peters CJ and Digoutte JP. Observations with NOAA and SPOT satellites on the effect of man-made alterations in the ecology of the Senegal River basin in Mauritania on Rift Valley fever virus transmission. Sistema Terra 1994, 3: 44-47. 10. Pope KO, Sheffner EJ, Linthicum KJ, Bailey CL, Logan TM, Kasischke ES, Birney K, Njogu AR and Roberts CR. Identification of central Kenyan Rift Valley fever virus vector habitats with Landsat TM and evaluation of their flooding status with airborne imaging radar. Remote Sensing of the Environment. 1992, 40: 185-196. 11. TNTmips, Version 5.6 of the Map and Image Processing System 1997. Micro Images Inc., Lincoln, Nebraska. 12. MapInfo, Version 4.1 of MapInfo User’s Guide, 1996. MapInfo Corporation, Troy New York. 13. Division of Epidemiology, Ministry of Public Health, Thailand, 1994. 1993 Annual Summary. Epidemiological Surveillance Report. 14. Division of Epidemiology, Ministry of Public Health, Thailand, 1997. 1996 Preliminary Annual Summary. Epidemiological Surveillance Report. 15. PFINDER Software, PFINDER Software User Guide, 1995 Trimble Navigation Ltd, Sunnyvale, CA. 16. SPSS. SPSS for Windows, Base System User’s Guide, Release 6.0 (SPSS Inc, Chicago, IL, 1993). 17. Gubler DJ, Dengue. 1987. In The Arboviruses: Epidemiology and Ecology, Vol. 2, ed. Monath, T.P. pp.223-260. Boca Raton, FL: CRC Press.
References 1. Halstead SB. The pathogenesis of dengue: Molecular epidemiology in infectious disease. Am. J. Epidemiol. 1981, 114:632-648. 2. Gubler DJ, Reed D, Rosen L, and Hitchcock JC. Epidemiologic, clinical and virologic observations on dengue in the Kingdom of Tonga. Am J. Trop. Med. Hyg. 1978, 27: 581-589. 3. Rosen L. The emperor’s new clothes revisited, or reflections on the pathogenesis of dengue haemorrhagic fever. Am. J. Trop. Med. Hyg. 26: 337-343. 4. Burke DS, Jatanasen S, Watts DM and Tang DB. Correlation between cool season environmental temperatures and dengue haemorrhagic fever th (DHF) case rates in Bangkok, Thailand. 10 Intl. Congr. Trop. Med. Malaria. 1980, 56: 35-36. 5. Watts DM, Burke DS, Harrison BA, Whitmire RE and Nisalak A. Effect of temperature on the vector efficiency of Aedes aegypti for dengue-2 virus. Am. J. Trop. Med. Hyg. 1987, 36: 143-152. 6. Nualchawee K, Singhasivanon P, Thimasarn K, Dowreang D, Linthicum K,Sithiprasasna R and Rajbhandari PL. Correlation between malaria incidence and changes in vegetation cover using satellite remote sensing and GIS techniques. In Proceedings of the International Geoscience and Remote Sensing Symposium (IGARSS), August 1997. Institute of Electrical and Electronic Engineers, New York. 1997 (in press).
72
Dengue Bulletin – Vol 21, 1997
Insert A Distribution of dengue haemorrhagic fever incidence/100 000 population for the year 1996 in different provinces of Thailand plotted on provincial boundary layer in a GIS. NOAA/NDVI satellite image showing forest cover in dark green colour used for base map. Water shown in dark black colour.
73
Dengue Bulletin – Vol 21, 1997