Predicting endemic cholera:: the role of climate variability and disease dynamics

Retrospective studies of cholera time series in Bangladesh have established a role of the El Nino Southern Oscillation (ENSO), but also of the non-linear dynamics of the disease itself, through changes in the population levels of immunity in this endemic region. The prediction ability of a semi-mechanistic time series model that incorporates both these elements is examined. Results show that ENSO is a key covariate and confirm the importance of its interplay with immunity levels, now from the perspective of prediction. They support the feasibility of using the model as a forecasting tool: the lack of extreme events between 2001 and 2005 would have been anticipated with 75% confidence half a year ahead with a model fitted to data up to 2000. Long-term change in the transmission rate, the non-mechanistic part of the model, sets limits to the forecasting horizon because of a breakdown in its relationship with river discharge towards the end of the 1990s. We discuss this and other limitations of the approach as well as future directions related to the development of an early warning system for cholera in this region.