Space-time predictive model for nephropathia epidemica in Western Europe

The most common hantavirus in Europe, Puumala virus (PUUV), is specifically hosted by bank voles (Myodes glareolus) and causes a mild form of haemorrhagic fever with renal syndrome, called nephropathia epidemica (NE) in humans. In the last decade, significant changes in incidence patterns of NE have been described in Western Europe. A sudden increase in NE incidence and changes in epidemic periodicity ask for models allowing early prediction of local NE outbreaks in this region1. There is however explicit space-time variation in terms of local clusters with varying PUUV infection prevalence in bank voles and NE incidence in humans. Spatial variation in NE risk has been associated mainly with preferred bank vole habitat, local physicochemical and climate conditions, affecting both vole ecology and virus survival. Temporal variation is driven by bank vole population dynamics where peaks in bank vole abundance and consequent PUUV epizootics are followed by NE epidemics in humans2. In the presented study we hypothesize that a single set of explanatory variables can describe the observed space-time patterns found in Western Europe. Hence, we aimed to construct an ecological-based regression model for NE, allowing prediction of local NE outbreaks in the coming year throughout the West-European region (NUTS3 and NUTS4 resolution). A Bayesian Generalized Linear Mixed model is used, which is based on long-term NE datasets from Belgium, France, Germany and the Netherlands where geo-referenced NE cases were registered (between 2000-2012). Explanatory variables (climate, land use, soil texture variables and remote sensing derivatives) have been selected based on previous modelling work. The remote sensing derivatives have been selected as proxy’s for ecological variables that were previously associated with bank vole resources, survival and reproduction.