Multi-species Evaluation of Human Risk of West Nile Virus Infection, Saskatchewan 2003
Corrigan, Rebecca L.A.
In an effort to monitor WNV across Saskatchewan, the provincial health authority and collaborators established an integrated surveillance system for WNV. Surveillance included both human and animal components, as well as environmental variables including precipitation, growing degree days (GDDs), ecozone and landcover. Post-season analysis of the integrated surveillance system included investigating which surveillance variables were successful in predicting human risk of WNV infection in the same areas, determining if clinical horse case data could be used to predict future human cases of WNV, and establishing if early-season seroconversion in sentinel horse herds could predict occurrence of human cases in the same areas. Cluster analysis revealed significant clusters of human and horse cases of WNV in space, time, and space-time; however, only space-time clusters of horse cases preceded human cases by one week in the same area. Significant spatial clusters of human and horse cases did not occur in the same areas. Both human and horse cases clustered during the same time period under temporal analysis. Spatial regression analyses were used to determine which components of surveillance predicted increasing risk of disease in humans and horses in the same regions where surveillance variables were collected. Environmental variables predictive of increasing risk of human cases of WNV included increasing maximum ODDs. Maximum precipitation had a protective effect for both human and horse cases. Both Culex restuans and Culex tarsalis mosquitoes were predictive of increasing risk of human and horse disease. Increasing percentages of positive corvid birds were also predictive of increasing percentages of human cases. The results of these analyses can be used to guide future years' surveillance efforts for WNV in Saskatchewan. Early-season serological surveillance of sentinel horse herds by detection of IgG antibodies using an ELISA test was predictive of human cases of WNV in the same regions. Detection of IgM antibodies in the same samples was not predictive of human risk. Seroconversion of sentinel horse herds was predicted by decreasing maximum precipitation, increasing percentages of Culex restuans and Culex tarsalis, and increasing percentage of the rural municipality covered by water. Early-season serological horse surveillance is an important surveillance technique to predict human risk of WNV infection.