A statistical model to predict episodes of Ross River Virus infection in Northern New South Wales, Australia

Document Type

Conference publication

Publication details

Morgan, G, Beard, J, Jong, K & Brooks, L 2004, 'A statistical model to predict episodes of Ross River Virus infection in northern New South Wales, Australia', paper presented to the Sixteenth Conference of the International Society for Environmental Epidemiology (ISEE), Washington Square campus of New York University (NYU), 1-4 August.

Peer Reviewed



Ross River Virus (RRV) infection is the most prevalent mosquito-borne disease in Australia. RRV is active in northern New South Wales (NSW), Australia, and causes significant morbidity with over 50% of those contracting the virus having symptoms persisting beyond six months. Our study of the Northern Rivers Area (NRA) of NSW from 1992 to 2001 describes trends and cyclical variations in monthly counts of RRV, monthly mosquito numbers, and various environmental factors. We used time series analysis, taking account of overdispersion and serial correlation, to investigate associations between monthly counts of RRV and environmental factors. We developed statistical models based on temperature, rainfall (or the Southern Oscillation Index) and the previous patterns of RRV to predict RRV incidence in the NRA. Our models effectively predicted the regular seasonal peaks in RRV in the NRA, as well as the extreme episodes in 1996, 1999 and 2001. We validated our short term and long term predictive models prospectively against a further 2 years of data and effectively predicted RRV patterns in the NRA over this period. Our long term predictive model can be used to estimate RRV episodes up to 6 months in advance. Our analysis suggests that predictive models can be developed for arboviral diseases, based on environmental factors. These models can be used to predict extreme episodes of disease and to target limited public health resources. They might also be used to investigate the impact of climate change on disease prevalence.

Additional information

Abstract published in: Epidemiology, vol. 15, no.4, pp. S129-S130.