West Nile virus : from surveillance to prediction using Saskatchewan horses

Abstract

This thesis describes the West Nile virus (WNV) epidemic in horses by exploring all aspects: sub-clinical infection, development of clinical disease and case fatality. All of the collected data were then compiled to create predictive risk maps of WNV infection for the province of Saskatchewan. During the 2003 season, 133 clinical cases were documented with laboratory testing. Week of onset of clinical signs, gender, and coat color were significant predictors of whether the horse died or was euthanized due to severity of clinical signs. Studies of the serological response to vaccination and natural infection were examined to interpret the lab results from over 1100 samples taken from approximately 875 horses in 2003. A serologic study involving 212 horses on 20 farms determined the prevalence of sub-clinical infection (55.7% (95%CI, 44.9% to 65.8%)) and identified risk factors for infection. The study found risk of infection was highest in the Grasslands ecoregions compared to the Boreal Transition ecoregion. A case control study looked at risk factors for development of clinical disease. The study followed 23 case farms and control farms with a total of 300 horses sampled. This was the first field study to show that vaccination was efficacious in preventing the development of clinical signs. The inclusion of horse surveillance data in the Saskatchewan Health WNV Integrated Surveillance Initiative was useful; however, it was discontinued due to time constraints, logistics, and declining monetary resources. Since West Nile Virus is a mosquito-borne disease it is highly influenced by environmental changes, spatially and temporally. Discriminant analyses were used to partition Saskatchewan rural municipalities (RM) into categories of risk of infection with WNV based on acquired horse data and different environmental and meteorological data derived from both satellites or climate stations. The result was the creation of yearly predictive risk maps defining low to high risk of infection with WNV for each RM. The 2003 epidemic provided a novel opportunity to study an important zoonotic disease emerging in a new environment. The information gathered will further the knowledge base upon which decisions for prevention of infection and clinical disease are made.