Ripe for an outbreak: predicting ideal conditions for Ross River virus. (Science Selections).

Ross River virus disease (also called epidemic polyarthritis) afflicts thousands of Australians each year with symptoms--joint pain, rash, fever, muscle pain, and fatigue--that may last from several weeks to years. Like West Nile fever and dengue fever, this mosquito-borne disease, the most common of its type in Australia, is on the rise. In recent years, outbreaks have become more common near urban areas, and this reportable disease is significantly impacting industry, agriculture, and tourism, as well as residents in endemic areas.

As more and more arthropod-borne viral diseases spread throughout the world, researchers are striving to improve their ability to predict outbreaks. Now, in a novel and promising approach, researchers at the Centre for Public Health Research at Queensland University of Technology are applying computer modeling techniques borrowed from the fields of economic theory and corporate strategy to study the effects of environmental conditions on disease incidence [EHP 109:1271-1273].

Researchers Shilu Tong and Wenbiao Hu used a time-series methodology called autoregression integrated moving average (ARIMA) modeling to quantify for the first time a clear relationship between climate variation and transmission of Ross River virus in Cairns, Queensland, in Australia's far north tropics. They found that changes in relative humidity and rainfall play significant roles in disease transmission. While the incidence of Ross River virus disease is known to be linked to rainfall, high tide, and temperature--mosquitoes need water to breed and spread--this new modeling reveals the kind of highly specific relationships between climate data and disease incidence that are critical in accurately predicting future outbreaks.

The investigators chose their methodology because ARIMA models are particularly useful in analyzing nonstationary time-series data that contain seasonal trends, such as the pattern of changes in rainfall, humidity, and temperature. They used data from the Australian National Notifiable Diseases Surveillance System, Bureau of Meteorology, and Bureau of Statistics. The model was based on data collected from January 1985 to December 1994, and validated with data from January 1995 through December 1996.

They found a statistically significant association between relative humidity and occurrence of Ross River virus disease five months later, and between rainfall and occurrence of Ross River virus disease two months later. They observed no statistically significant association for temperature, possibly because the temperature in Cairns is usually consistently high.

Tong and Hu suggest that a decrease in relative humidity lessens stream flow, creating stagnant pools that are perfect mosquito breeding sites. The reduction in water sources also concentrates thirsty wildlife, increasing the potential for disease transmission from insects to vertebrates gathered to drink. Increased rainfall, on the other hand, helps disperse adult mosquitoes.

The investigators observed that many other factors--virus strain, mosquito population densities and survival rates, human behavior, even housing characteristics--need to be incorporated in the modeling. Noting that it is not yet possible to predict the magnitude of changes in disease patterns that global warming will bring, they underscore the need to develop computer models capable of forecasting epidemics under varying environmental conditions.