Methodology
Using geographic information systems and spatial and space-time scan statistics for a population-based risk analysis of the 2002 equine West Nile epidemic in six contiguous regions of Texas
-
* Corresponding authors: Min Lian mlian@im.wustl.edu - Kenneth R Dixon ken.dixon@tiehh.ttu.edu
1 Division of Modeling and Geographic Information Systems, Institute of Environmental and Human Health, Texas Tech University/TTU Health Sciences Center, Box 41163; Lubbock, TX 79409, USA
2 Department of Family and Community Medicine, Texas Tech University Health Sciences Center School of Medicine; Lubbock, TX 79430, USA
3 Texas Department of State Health Services, Health Service Region 1, WTAMU Box 60968; Canyon, TX 79016, USA
4 Department of Medicine, Washington University School of Medicine, Campus Box 8504, St. Louis, MO 63108, USA
International Journal of Health Geographics 2007, 6:42 doi:10.1186/1476-072X-6-42
Published: 21 September 2007Abstract
Background
In 2002, West Nile virus (WNV) first appeared in Texas. Surveillance data were retrospectively examined to explore the temporal and spatial characteristics of the Texas equine WNV epidemic in 2002. Using Geographic Information Systems (GIS) and the Spatial and Space-Time Scan (SaTScan) statistics, we analyzed 1421 of the reported equine WNV cases from six contiguous state Health Service Regions (HSRs), comprising 158 counties, in western, northern, central and eastern Texas.
Results
Two primary epidemic peaks occurred in Epidemiological (Epi) week 35 (August 25 to 31) and Epi week 42 (October 13 to 19) of 2002 in the western and eastern part of the study area, respectively. The SaTScan statistics detected nine non-random spatio-temporal equine case aggregations (mini-outbreaks) and five unique high-risk areas imbedded within the overall epidemic.
Conclusion
The 2002 Texas equine WNV epidemic occurred in a bi-modal pattern. Some "local hot spots" of the WNV epidemic developed in Texas. The use of GIS and SaTScan can be valuable tools in analyzing on-going surveillance data to identify high-risk areas and shifts in disease clustering within a large geographic area. Such techniques should become increasingly useful and important in future epidemics, as decisions must be made to effectively allocate limited resources.