PARENT SESSION
Contributed Oral Session 94: Toxicology and Disease : Viruses and Epidemics
Wednesday, August 10, 1:30 PM - 5:00 PM, Meeting Room 514 A, Level 5, Palais des congrès de Montréal
Dynamic epidemiological models for West Nile virus: Assumptions and controls.
Wonham, Marjorie*,1, Lewis, Mark1, Renclawowicz, Joanna2, van den Dreissche, Pauline3, 1 University of Alberta, Edmonton, Alberta, Canada2 Polish Academy of Sciences, Warsaw, Poland3 University of Victoria, Victoria, British Columbia, Canada
ABSTRACT- Emerging and re-emerging infectious diseases of humans and wildlife are increasing in frequency. A powerful tool in understanding their ecology and control is classical epidemiological modeling. However, model development requires making certain biological assumptions about organism behaviour and pathogen transmission dynamics, which in turn determine the model's mathematical structure and therefore influence its predictions. Thus, for the same disease system, multiple biologically reasonable models may predict different outcomes. We consider this problem by exploring six S-I-R models developed by different authors for West Nile virus and two similar pathogens, Japanese and St. Louis encephalitis. We identify key differences among models in their treatment of mosquito and bird life cycles and vital dynamics (e.g., stage structure), vertical and horizontal disease transmission dynamics (e.g., biting rates, transovarial transmission), and time (e.g., continuous vs. discrete; constant vs. fluctuating rate parameters; time lags). We illustrate how these differences scale up to differences in the structure of R0, the disease basic reproduction ratio, and therefore to implications for disease spread and control. Modelling decisions to include a larval mosquito stage, mosquito loss of viremia, and bird recovery to and loss of immunity, all have important analytical implications that we illustrate in the speed and magnitude of of numerical outbreak simulations. Using realistic parameter values, we also show that because of their different structures, some of these models predict a disease outbreak (R0 > 1) but others do not, and the threshold mosquito level for disease outbreak varies over an order of magnitude. We discuss these results in the context of model development, selection, and validation, and management implications.
Key words: emerging infectious disease, outbreak threshold, North America, corvids