West Nile virus has been a focus of public health efforts in the United States since its introduction in 1999. Humans infected with WNV may experience a range of symptoms, from flu-like febrile illness to neuroinvasive disease, often termed West Nile Neuroinvasive disease (WNND). WNND is a serious, and at times fatal, condition in populations exposed to WNV. In 2012, the US has experienced an epidemic year with 5,245 reported cases of WNV, 51% of which were WNND cases, and an estimated incidence of 1.67 per 100,000 people. WNND cases have an estimated fatality rate of 3-15% highlighting the impact on public health. Because there is currently no vaccine against WNV for humans, the best preventative measure possible is control of the mosquito vector population guided by the surveillance of virus activity. In California sentinel chicken flocks, mosquito infection prevalence and abundance, and the infection prevalence in dead birds are the surveillance methods used by mosquito control agencies to target control efforts to reduce the risk of transmission to humans. For the proposed project, these methods will be analyzed to address the overall hypothesis that the risk of WNV infection for humans can be accurately predicted over diverse ecological contexts using spatio-temporal modeling. The information gained from the proposed project is intended to be used by public health agencies in influencing surveillance strategies to more effectively detect virus activity and tailor control efforts to preent transmission to the human population where and when the risk is greatest. Data on the surveillance methods from four California counties for the years 2004-2011 will be used along with climatic and population data to test the overall hypothesis. A mixed-effects Poisson regression model will be used to estimate the lead-time and distance at which each of the surveillance methods is most predictive of human WNND cases. In order to determine the time and place that the human population is at highest risk, a predictive Poisson regression model will be constructed using various ecological and population-based predictor variables, such as surveillance measures, human population density, climate, and mosquito-host biting preference. This model will be constructed with the intent that it be used by mosquito control agencies across the nation to guide control efforts. In addition, a probabilistic statistical model will be uilt in order to characterize the different combinations of climatic and population factors which are most conducive to transmission of WNV to humans.
Public health agencies charged with preventing human infection with West Nile virus rely on a variety of surveillance information to guide control of th mosquitoes that transmit the virus. To increase the efficacy of interventions and minimize the use of pesticides, it is important to target control efforts in the places and times that are likel to have the greatest impact in reducing future disease. This project will estimate the collective contributions of surveillance methods for estimating human disease risk in various ecological contexts and translate these associations into a predictive model that will inform local, state, an national surveillance and response guidelines.
