Retrospective space-time analysis methods to support West Nile virus surveillance activities

Abstract

The steep increase in human West Nile virus (WNV) infections in 2011-2012 in north-eastern Italy prompted a refinement of the surveillance plan. Data from the 2010-2012 surveillance activities on mosquitoes, equines, and humans were analysed through Bernoulli space-time scan statistics, to detect the presence of recurrent WNV infection hotspots. Linear models were fit to detect the possible relationships between WNV occurrence in humans and its activity in mosquitoes. Clusters were detected for all of the hosts, defining a limited area on which to focus surveillance and promptly identify WNV reactivation. Positive relationships were identified between WNV in humans and in mosquitoes; although it was not possible to define precise spatial and temporal scales at which entomological surveillance could predict the increasing risk of human infections. This stresses the necessity to improve entomological surveillance by increasing both the density of trapping sites and the frequency of captures.