Predicting Risk of Vector-borne Disease:
Mosquitoes, Malaria and Worms
Mosquito-borne infectious diseases including malaria, dengue and West Nile are among the chief concerns for global public health. Remotely sensed data is now being used to predict the risk of vector-borne disease worldwide.
Currently, the annual death counts due to vector-borne infectious diseases (those transmitted by arthropods, including mosquitoes) are massive. Malaria alone accounts for more than 300 million acute illnesses per year with at least one million deaths annually, according to 2004 figures from the World Health Organization (WHO). See Figures 1 and 2. Dengue, also a mosquito-borne infection, has become a major international public health problem. WHO estimates that worldwide, mostly in tropical and subtropical regions, there may be more than 50 million cases of dengue infection every year. The 1999 emergence of West Nile Virus (WNV) in New York has brought the threat of infectious disease to the forefront in the U.S. WNV, although not accounting for the same high morbidity and mortality figures as other vector-borne diseases, has caused an increase in epidemiological surveillance in the U.S. and become a public health concern.
Vector-borne diseases take a heavy toll on the health of the global population. However, we are in the age of novel and cost-effective control strategies as governments and private entities are increasing their motivation towards disease prevention. Spatial analysis and remote sensing technologies are rapidly developing to complement international disease-control strategies. For example, WHO has called for the development of innovative methods for malaria surveillance and control (Najera 1989). Such development requires definition of the environmental determinants and epidemiological parameters that affect the patterns of the malaria vector distribution, the host distribution, and the risk of disease transmission. Remote sensing (RS) is increasingly being used as a tool to collect data about the environment that is applicable to vector-borne diseases.
Recognizing the aspects of the environment that lend themselves to disease transmission can be tough business. For example, the larval stage of the malaria vector develops in an aquatic environment. One strategy for reducing the numbers of mature malaria mosquitoes is treating aquatic areas with mosquito control products. However, locating these prime mosquito habitats is often not feasible due to high cost and time constraints. In addition, the land area under consideration may be too large for mosquito control technicians to locate all of the mosquito habitats. By utilizing remotely-sensed data, scientists are able to collect information about the suitability of a specific landscape for the establishment of disease vectors. Remotely-sensed attributes of the landscape such as vegetation cover may act as indicators for disease risk.
|Figure 2: Malaria Distribution, courtesy of CDC. |
Note: This map shows countries with endemic malaria in yellow. In most of these countries, the malaria risk is limited to certain areas.
A number of studies have used an RS/GIS approach to analyze the risk of vector-borne disease. Several studies have then used the gathered data to design strategies to reduce the risk of vector-borne disease through mosquito control operations. For example, in California, both LANDSAT (USGS, EROS Data Center) and aircraft were used in a project to monitor rice fields for water quality, vegetation and mosquito abundance (Wood et al. 1991). The principal mosquito monitored was Anopheles Freeborni, also called the western malaria mosquito. Researchers observed that the Anopheles population peaked in August followed by a rapid decline in September when rice fields were drained before harvest. The remotely-sensed data provided information about the ‘greenness’ of the rice fields. The research showed that the fields that ‘greened up’ earlier supported higher populations of Anopheles Freeborni larvae. In addition, rice fields that tended to have slower crop growth supported fewer mosquito larvae. Based upon remotely-sensed data that was collected early in the mosquito season (June), researchers were able to correctly predict 81% of the rice fields that would support high mosquito numbers throughout the season. This early prediction of mosquito numbers occurred two months prior to peak mosquito abundance. Such early prediction provided by RS data is very beneficial to mosquito control operations by determining which rice fields need focus for efficient mosquito control.
If the goal for this RS/GIS study in Californian rice fields was to use RS data to gain a new perspective of mosquito dynamics, then this study achieved its purpose. However, we must consider the aspect of scale in these studies. The 30-m resolution LANDSAT images were useful in the rice field study to resolve mosquito habitat within the fields. However, in a more heterogeneous landscape where the scale of mosquito habitat can be much smaller than 30-m, an image of finer resolution is essential. For example, LANDSAT images may not be the most appropriate data for the study of urban malaria transmission or WNV (Castro et al. 2004). Important breeding sites for the vectors of urban malaria and WNV are often smaller than 10-m and would thus require a very high resolution image.
In 2003, an article was published comparing the usefulness of LANDSAT imagery with that of IKONOS to determine the location of mosquito larval habitats on a US military base in the Republic of Korea (ROK) (Masuoka et al. 2003). Malaria had reemerged in the ROK in 1993 and this study was aimed at reducing the risk of malaria transmission to U.S. personnel. If larval habitats could successfully be identified, necessary mosquito control measures (using larvicide in aquatic habitats to kill mosquito larvae) could be used to reduce the risk of malaria. Both IKONOS and LANDSAT images were used to identify mosquito larval habitat on the military base. Following larval collection and image classification, researchers calculated land cover estimates of mosquito larval habitat based on both types of RS imagery. Masuoka et al. found the IKONOS imagery was able to resolve small irrigation ponds more accurately than the lower-resolution LANDSAT imagery. They found that although these small ponds represented a minor portion of the total mosquito habitat, they were important mosquito breeding sites. The researchers concluded that for areas where small niche features represent the majority of mosquito habitat, high resolution imagery such as that of IKONOS is necessary for correct identification of the habitats, and for the planning and implementation of a mosquito control program on the U.S. military base.
|Figure 3: Schistosoma worm, courtesy of SCI and Environmental Health Perspectives|
Just recently, published results of several other spatial disease projects have become available. Two of these projects have focused on urinary schistosomiasis, an endemic disease in many sub-Saharan countries in Africa. Human contact with the Schistosoma parasite often occurs during domestic activities like clothes washing and bathing in surface waters. S. haematobium invades the veins around the urinary tract in infected individuals. Bulinus snails, the important intermediate hosts of the Schistosoma parasite, are found in small water sources. Figure 3 shows a Schistosoma worm.
Kariuki et al. (2004) used satellite imagery in their study of the spatial distribution of the Bulinus snails. In a companion study, Clennon et al. (2004) used the same image to study the distribution of human cases of urinary schistosomaisis and the transmission dynamics of S. haematobium. Demographic, parasitologic, and household location data were mapped for an area in the Coast Province of Kenya. All distances were calculated between houses and water sites to detect clustering of human schistosomiasis infection. Overall, the transmission pattern of urinary schistosomiasis was highly associated with the spatial distribution and abundance of Bulinus snails. The spatial and temporal variation of snail abundance generated by this research is vital for successful schistosomiasis control programs.
The Arthropod and Infectious Disease Laboratory (AIDL) at Colorado State University is utilizing IKONOS imagery for the study of disease vectors. As mentioned previously, the emergence of WNV has caused an increase in public health concern for mosquito vectors in the U.S. Colorado has been greatly affected by WNV, with 2947 human cases in 2003 and 225 human cases as of October 6, 2004 (Centers of Disease Control).
The study area for our project is located within Larimer and Weld Counties of Colorado along the Front Range of the Rocky Mountains. Two satellite images incorporating urban, suburban, riparian and agricultural areas are being used for a base map. The main goal is to identify whether researchers can correctly determine mosquito habitats within the study area. We trapped mosquitoes at a number of sample sites within the study area during the summer of 2004. The mosquito collection data will be statistically analyzed in relation to an unsupervised classification of the satellite images covering the study area.
|Figure 4: IKONOS imagery of the northern region of the study area in Colorado|
Several mosquito species are found within the study area. Some of these mosquitoes are considered nuisances and do not transmit disease. Others, like the Culex species, are able to transmit WNV. AIDL proposes that satellite imagery will serve as a useful tool for identifying the habitat of a number of mosquito species within the study area. If this assumption is correct, the imagery may someday be used within a model to streamline mosquito control operations by targeting pesticide application to the identified prime mosquito habitats. Such a model has the potential to limit over-application or under-application of insecticides. A model based upon satellite imagery could reduce transmission of mosquito-borne diseases like WNV as well as reduce costs and limit unnecessary environmental exposure to pesticides.
Globally, vector-borne diseases are an imminent threat to healthy human lives. The field of remote sensing has developed along with vector-borne disease research to address the problem of disease. As technology increases and satellite imagery becomes more refined, the science community must continue to examine the usefulness of RS in vector-borne disease research. The applications of RS are immense and human health is at stake. Vector-borne disease must be controlled to increase the quality of international human health, and the capabilities of RS will contribute to that control.
Castro, M., et al. 2004. “Integrated Urban Malaria Control: A Case Study in Dar es Salaam.” American Journal of Tropical Medicine and Hygiene, 71 (Suppl 2), 103-117.
Center for Disease Control. “West Nile Virus Statistics, Surveillance and Control.” (2004) Retrieved October 6, 2004 from http://www.cdc.gov/ncidod/dvbid/westnile/surv&control04Maps.htm
Clennon, J., et al. (2004). “Spatial Patterns of Urinary Schistosomiasis Infections in a Highly Endemic Area of Coastal Kenya.” American Journal of Tropical Medical Hygiene, 70(4): 443-448.
Kariuki, H., et al. (2004). “Distribution Patterns and Cercarial Shedding of Bulinus Nasutus and Other Snails in the Msambweni Area, Coast Province, Kenya.” American Journal of Tropical Medical Hygiene, 70(4): 449-456.
Masuoka, P., et al. 2003. “Use of IKONOS and Landsat for Malaria Control in the Republic of Korea.” Remote Sensing of Environment 88, 187-194.
Najera, J. 1989. “Malaria and the Work of WHO.” Bulletin of the World Health Organization, 67:229-243.
Wood, B., et al. “Distinguishing High and Low Anopheline-producing Rice Fields Using Remote Sensing and GIS Technologies.” Preventive Veterinary Medicine, 11, 277-288 (1991).
World Health Organization, “Dengue Fact Sheet.” (2004) Retrieved October 6, 2004 from http://www.who.int/mediacentre/factsheets/fs117/en/
World Health Organization, “Roll Back Malaria Fact Sheet.” (2004) Retrieved October 6, 2004 from http://www.rbm.who.int/cmc_upload/0/000/015/372/RBMInfosheet_1.htm
Thanks to: Dr. Chet Moore and Keith Olson.