Ross River virus forecasting in four Victorian local government areas

Ross River virus (RRV) is Australia’s most commonly reported arbovirus. Over the 2016–17 summer, Victoria experienced its largest RRV outbreak on record. Since 2013, the Victorian Department of Health and Human Services (DHHS) has been working with infectious disease modellers to develop predictive models to forecast human RRV notifications, improve understanding of the drivers of RRV and provide early warning to inform mosquito control and other interventions. This analysis aimed to refine an existing predictive model for RRV notifications, implement it in four local government areas (LGAs) and investigate similarities in predictive variables in each location. Ross river human case data were obtained from the DHHS, while hydrological and meteorological data were collected from a number of sources. Four LGAs were selected for analysis (Mildura, Campaspe, Horsham and Greater Shepparton) on the basis that their patterns of notifications over the last 10 years broadly represented those observed across all Victorian LGAs. Data from July 2005–June 2012 were used to train negative binomial regression models of monthly RRV notifications and data from July 2012–June 2017 were used for validation. Similar variables were found to perform well for prediction across the four LGAs, including variables relating to surface water, humidity and the El Nino cycle. It was surprising how consistently a small set of key variables performed across regions, considering expected differences in the ecology of RRV. Evapotranspiration appears highly predictive and was included in most models. An interactive application to report model results was developed and piloted during the 2017–18 season, enabling trends in forecasts to be compared across LGAs. The predictive models are contributing to situational awareness of RRV activity within DHHS and will facilitate appropriately timed and targeted control actions, including mosquito control and public health messaging focussed on high-risk LGAs in future seasons.