After natural disasters inflict wreckage and death on a population, what often follows is the threat of diseases. After floods, standing freshwater can become a breeding ground for mosquitoes. Although the chance of a mosquito-born epidemic following a disaster in the United States is slim, disaster-response policies require that we monitor such risk.
Vector-borne diseases — infectious diseases transmitted by arthropods
— are re-emerging as public health threats in the United States. While
some of these diseases have been present for over a century, others are
newly introduced into the country. For example, since 1999, an increasing
number of mosquitoes on the East Coast have been caught carrying the West
Nile virus. But an even more dangerous arboviral disease — one that kills
one-third of infected humans — lurks in the wetlands of the Gulf Coast.
When floods hit, coupling the epidemiologist's knowledge of disease outbreaks
with geographic information systems (GIS) and remote sensing technology
could help natural disaster relief workers prevent the spread of disease
Scientists at the United States Geological Survey (USGS) and the Centers for Disease Control and Prevention (CDC) are undertaking a study in the Mobile, Ala., area to determine the feasibility of using remote sensing and GIS technologies as part of a disaster response plan. Such technology may help to establish the location and extent of a disaster as well as those areas that contain the hosts of pathogens that cause disease. A key response will also involve locating all populations at risk and working to reduce the likelihood of exposure to disease.
[This picture of a freshwater swamp habitat of eastern equine encephalomyelitis in North Carolina was taken after Hurricane Fran hit the state in 1996. Courtesy of R.S. Nasci, Centers for Disease Control and Prevention.]
The major post-disaster disease risk in the Mobile study area is from eastern equine encephalomyelitis (EEE), an often-fatal inflammation of the brain and spinal cord. Carried by mosquitoes, the disease occurs in focal locations along the eastern seaboard, the Gulf Coast and some inland Midwestern locations of the United States. Small outbreaks of human disease have occurred in the United States, and outbreaks among horses can be a common occurrence during the summer and fall. Humans develop symptoms four to 10 days after being bitten by an infected mosquito. These symptoms begin with a sudden onset of fever, general muscle pains and a headache of increasing severity. Many individuals will progress to more severe symptoms such as seizures and coma. About one-third of all people with clinical encephalitis caused by EEE will die from the disease. Of those who recover, many will suffer permanent brain damage, with many of those requiring permanent institutional care.
The EEE virus also can produce severe disease in horses; some birds, such as pheasants, quail, ostriches and emus; and even puppies. The whooping crane, an endangered species, is highly susceptible to EEE. Cases in horses usually precede those in humans, making horse cases a good surveillance tool.
After a flood event, the numbers of mosquitoes may dramatically increase. Culiseta melanura mosquitoes transmit the EEE virus among avian hosts. But the EEE virus may escape from the wetland via birds or such bridge vectors as the freshwater mosquito Coquillettidia perturbans and the eastern salt marsh mosquito Aedes sollicitans. These species feed on both birds and mammals and can transmit the virus to humans, horses and other hosts.
The disease investigator's challenge after a flood is to locate areas of likely habitat for these and other mosquito species, then determine their abundance and whether they are carrying a virus.
Locating the areas that contain the hosts of disease agents is a multi-faceted problem. Remote sensing technology can help to determine the geographic extent and severity of a disaster. Comparing modern imagery with archival data helps to determine the magnitude of the damage. In the case of this study, we are particularly interested in the extent of flooding — resulting from a Gulf Coast hurricane, for example. If clouds are not obscuring the ground, high-resolution images (pixels 15 meters in size or smaller) from satellites or aircraft can detect these flood boundaries. Should cloud cover be a problem, radar sensors can penetrate the cloud cover and provide good discrimination of the land and water boundary despite their coarser spatial resolution.
[Entomologist Charles Anderson with the Louisiana Department of Health and Hospitals stands beside a water-filled pit created when hurricane Andrew uprooted trees in southern Louisiana in 1992. Large numbers of larval habitats can form when trees are uprooted. Courtesy of C.G. Moore, Centers for Disease Control.]
The remote sensing data is then combined with other geospatial information in a GIS. For the Mobile area, we have constructed a GIS database that contains multi-resolution land cover data, digital imagery from Landsat 7 and digital orthophoto quads, National Wetlands Inventory data, digital elevation data, digital soils information, hydrography and transportation information, and census data.
From these data, we can determine which areas mosquitoes will most likely inhabit, such as vegetated fresh-water wetlands and areas of bog or muck soil. Ideally, seven to 10 days after the peak flood, imagery would be acquired to see if the flooding has created or removed habitats favorable for mosquitoes. Additionally, the imagery would show if areas of potential habitat are accessible via ground transportation, or if watercraft must be used. With this information, the field entomologists can deploy their mosquito traps in areas where mosquitoes are most likely to live. The captured mosquitoes are then tested to see if they carry disease pathogens.
The locations of any pools containing virus-infected mosquitoes are then analyzed in the GIS with respect to the distribution of the human population. Clearly, infected mosquitoes located close to large numbers of susceptible humans pose a greater risk than mosquitoes that are miles from most of the population. While this analysis can be done with existing census data, in an emergency situation adjustments must be made for temporary population redistributions, such as to shelters at high schools. If limited resources are available for intervention and mitigation — spraying of insecticides to kill adult mosquitoes, application of larvicides — then the areas of mosquito habitat that pose the greatest risk can be targeted first.
For additional reading, visit www.cdc.gov/ncidod/dvbid/arbor/arbdet.htm