Premium
A geographic study of West Nile virus in humans, dead corvids and mosquitoes in Ontario using spatial scan statistics with a survival time application
Author(s) -
ThomasBachli A. L.,
Pearl D. L.,
Berke O.,
Parmley E. J.,
Barker I. K.
Publication year - 2017
Publication title -
zoonoses and public health
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.87
H-Index - 65
eISSN - 1863-2378
pISSN - 1863-1959
DOI - 10.1111/zph.12350
Subject(s) - west nile virus , cluster (spacecraft) , geography , scan statistic , geographic information system , demography , cartography , warning system , early warning system , statistics , biology , virus , virology , mathematics , sociology , computer science , programming language , aerospace engineering , engineering
Summary Surveillance of West Nile virus ( WN v) in Ontario has included passive reporting of human cases and testing of trapped mosquitoes and dead birds found by the public. The dead bird surveillance programme was limited to testing within a public health unit ( PHU ) until a small number of birds test positive. These dead corvid and mosquito surveillance programmes have not been compared for their ability to provide early warning in geographic areas where human cases occur each year. Spatial scan statistics were applied to time‐to‐event survival data based on first cases of WN v in found dead corvids, mosquitoes and humans. Clusters identified using raw data were compared to clusters based on model‐adjusted survival times to evaluate whether geographic and sociodemographic factors influenced their distribution. Statistically significant ( p < .05) space–time clusters of PHU s with faster time to detection were found using each surveillance data stream. During 2002–2004, the corvid surveillance programme outperformed the mosquito programme in terms of time to WN v detection, while the clusters of first‐positive mosquito pools were more spatially similar to first human cases. In 2006, a cluster of first‐positive dead corvids was located in northern PHU s and preceded a cluster of early human cases that was identified after controlling for the influence of geographic region and sociodemographic profile.