Searching for job candidates with the right knowledge and experience ? Well help you find them . Advertise your position vacancy in Environmental Health Perspectives

Recent El Ni&o events have siimulated interest in the development of modeling techniques to forecast temes of dimate and related health events. Previous studies have do_cnted associations between specific dimate variables (particularly temperature and rainfall) and outbreak of arboviral disease. In some countries, such diseases are sensitive to El Nifio. Here we describe a climate-based model for the prediction of Ross River virus epidemics in Australia From a literature search and data on case notifications, we determined in which years there were epidemics of Ross River virus in southern Australia between 1928 and 1998. Predictor variables were monthly Southern Oscillation index values for the year of an epidemic or lagged by 1 year. We found that in southeastern states, epidemic years were well predicted by monthy Southem Oscillation index values in January and September in the previous year. The model fore s that there is a high probability of epidemic Ross River vus in the southern states ofAustraa in 1999. We condude that epidemics of arboviral diase can, at least in principle, be preited on the basis of climate relationships. Key wor arboviruses dimate, prediction, vectorbome dis. Environ Heah Prect 107:817-818 (1999). [Online 3 September 19991 hnp:IIeIpnet1.nubs.nib.gev/docs/1 999/I107p817-88maelzerlabssrac html

Recent El Ni&o events have siimulated interest in the development of modeling techniques to forecast temes of dimate and related health events. Previous studies have do_cnted associations between specific dimate variables (particularly temperature and rainfall) and outbreak of arboviral disease. In some countries, such diseases are sensitive to El Nifio. Here we describe a climate-based model for the prediction of Ross River virus epidemics in Australia From a literature search and data on case notifications, we determined in which years there were epidemics of Ross River virus in southern Australia between 1928 and 1998. Predictor variables were monthly Southern Oscillation index values for the year of an epidemic or lagged by 1 year. We found that in southeastern states, epidemic years were well predicted by monthy Southem Oscillation index values in January and September in the previous year. The model fore s that there is a high probability of epidemic Ross River vus in the southern states of Austraa in 1999. We condude that epidemics of arboviral diase can, at least in principle, be preited on the basis of climate Recent El Niflo events and the prospect of anthropogenic dimate change have stimulated interest in the development of modeling techniques to forecast adverse impacts of climate extremes. In the short term, such forecasts may help mitigate adverse impacts by improved adaptation responses; in the longer term, the accuracy of forecasts can be used to assess objectively the validity of the models on which they depend. Arboviral diseases such as dengue and Ross River virus are transmitted by mosquitoes and the spread and persistence of these pathogens may therefore be influenced by climate. Although the importance of climate trends in the emergence of vectorborne diseases is controversial (1), transmission of vectorborne diseases is associated with El Nifno events in some historical datasets (2-7). Here we describe a model for the prediction of Ross River virus epidemics, based on the Southern Oscillation index-the normalized difference in surface atmospheric pressure between Tahiti (French Polynesia) and Darwin (Australia). Arboviral diseases have been an increasing problem in recent decades, both in the Pacific (8) and elsewhere (9). The reasons for this are not clear but may indude increased population density, urbanization, and travel, altered land use, and global dimate change. In Australia, Ross River virus is the most frequent cause of arboviral disease and there were more than 35,000 cases notified between 1991 and 1997. More than half of those infected experience symptoms, and at least half of these develop polyarthritis and a prolonged postviral syndrome lasting from months to years (8). Ross River virus has an endemic pattern in the tropical northern Australian states (Northern Territories and Queensland). In the southern states, including New South Wales, Victoria, and South Australia, there have been periodic epidemics documented in the medical literature and, in recent years, reported to health authorities.

Method
In the Southern Hemisphere, the warmer months are October-March; therefore, epidemics of Ross River virus are usually spread over 2 calendar years. Although epidemics are typically initiated in early summer (October-December), the majority of cases are notified in late summer Uanuary-April of the following year). For this reason, we designate the second year as the year of the epidemic. From a literature search and data on case notifications for 1991-1998, we determined that there were epidemics in southeastern states (New South Wales and Victoria) in 1928, 1943, 1945, 1956, 1965, 1971, 1976, 1980, 1984, 1989, 1993, 1996, and 1997. For years in which notification data were available, an outbreak was deemed to have occurred where more than 600 cases were notified between January and April.
We modeled the occurrence of Ross River virus epidemics from 1928 to 1998 using logistic regression. Epidemic years were classified as 1 and other years as 0. Predictor variables were monthly Southern Oscillation index values for the year of an epidemic or lagged by 1 year. Southern Oscillation index data were obtained from the Department of Primary Industries (1J). We tested the model using separate data for the state of South Australia, and calculated a forecast for the probability of an epidemic in 1999.

Results
We found that in southeastern states, epidemic years were well predicted by monthly Southern Oscillation index values in January and September in the previous year. Epidemics were negatively associated with the Southern Oscillation index in January [odds ratio (OR), 0.87; 95% confidence interval (CI), 0.78-0.97] and positively associated with the value for September (OR, 1.23; CI, 1.08-1.39). This relationship can also be seen by calculating the January-September Southern Oscillation index couplet difference. In previous studies, couplet differences (differences in the Southern Oscillation index between two single months) have proved to be useful composite variables (11,12. Figure 1 shows the occurrence of Ross River virus epidemics (vertical bars) for 1928-1998 and the Southern Oscillation index couplet differences for January-September in the previous year. The model forecasts a high probability of an epidemic of Ross River virus in the southern states of Australia in 1999. The model had good predictive ability in data for the state of South Australia, 1956-1996, with a positive predictive value of 88% and a negative predictive value of 91%, using a forecast probability of > 50% to define epidemic years.

Discussion
We demonstrated a relationship between epidemics of Ross River virus and monthly values of the Southern Oscillation index in historical data for southern Australia (1928Australia ( -1998. A number of previous studies have documented associations between specific dimate variables (particularly temperature and rainfall) and outbreaks of arboviral disease (13)(14)(15)(16). Rainfall affects vector breeding in particular (because breeding requires standing water), but may also influence adult survival (by increasing humidity) and disease transmission (by affecting the growth of vegetation and the abundance of animal reservoir hosts) (17). In Australia, the El Niflo phenomenon explains much of the interannual variation in dimate and the Southern Oscillation index is a useful proxy for temperature and rainfall.  1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 Year (epidemic years are vertical bars) For the purposes of the present study, the Southern Oscillation index has two major advantages over the use of more specific climate variables. First, it avoids the need to make estimates of climate variables for specific geographical areas. There is no reason to suppose that political boundaries define biologically relevant regions for aggregation of dimate data. Second, in many parts of the world (including Australia), rainfall can be predicted several months in advance on the basis of El Nifno. A rapid rise in the Southern Oscillation index in the austral autumn, followed by sustained high values over the winter, are subsequently associated with a high probability of increased rainfall (18). This sequence occurs when La Nifia conditions follow the decay of an El Nifio event in late summer (as happened in 1998). Our results are therefore consistent with previous studies that report an effect of rainfall on Ross River virus epidemics.
The data reported here suggest that it is feasible to forecast epidemics of vectorborne disease using climate-based models. Such forecasts could be used to target antimosquito measures more effectively in both time and place, with potentially important public health benefits. In the future, the accuracy of the model will provide an objective test of the hypothesis that vectorborne disease transmission is influenced by interannual climate variation.