An Investigation of West Nile Virus Surveillance Activities in Ontario, 2002 - 2008
This thesis is an investigation of the utility of data on dead corvids found by the public and tested in surveillance for West Nile virus (WNv) in Ontario, Canada. The aim of this thesis is to improve understanding of the predictive ability of dead wild corvids in the surveillance of WNv for human infections. Surveillance data obtained through citizen reports of found dead wild corvids (American Crows, Ravens, Magpies and Jays), some of which were submitted for WNv testing, are examined for their relative time-to-detection and consistency in time and space compared with trapped mosquito and human clinical case data between 2002 - 2008. Phone call reports obtained from citizens across Ontario during 2002 were also explored for their association with sociodemographic factors, and for their timeliness and reliability in WNv detection compared with test-positive corvids. Based on results from survival analysis, the dead corvid surveillance program identified WNv within public health units in Ontario more quickly, and were more predictive of human cases, than mosquito testing during the first few years of surveillance. During the later years of the study period, mosquito testing showed faster time-to-detection. Regional and sociodemographic factors influenced the speed of detection, depending on the surveillance modality. Clusters of early seasonal WNv-positive mosquitoes identified using scan statistics were spatially alike to, but not predictive for clusters of early seasonal human clinical cases. A cluster of early WNv-positive corvids preceded a spatially similar cluster of human cases after influencing geographic and sociodemographic factors were accounted for. Rates of phone call reports of dead corvids were also associated with sociodemographic factors within small areas across the province. Ultimately, the phone calls alone were not as timely or specific in comparison with the testing of corvids in the identification of areas with WNv. The data obtained through the help of citizens in Ontario provided a timely and effective approach to surveillance of a disease that was identifiable through observation of their surrounding environment. However, knowledge of the underlying factors influencing citizen reporting rates is important to allow informed adjustments for cluster detection to reduce confounding bias.