Intra- and inter-seasonal autoregressive prediction of dengue outbreaks using local weather and regional climate for a tropical urban environment in Colombia

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 5 February 2014: 11:00 AM
Room C213 (The Georgia World Congress Center )
Matthew D. Eastin, University of North Carolina, Charlotte, NC; and E. Delmelle and I. Casas

Dengue fever transmission results from complex interactions between the virus, human hosts, and mosquito vectors – all of which are influenced by environmental factors. A dengue early warning system aims to anticipate periods when environmental conditions are most favorable for an outbreak so timely decisions regarding public awareness, vector control, and disease prevention strategies can be initiated in a cost-effective manner. The aim of this study is to develop intra- and inter-seasonal models of dengue incidence rate, based on local weather and regional climate parameters, for integration into an early warning system in Cali, a dynamic urban environment in Colombia, South America.

Time series of epidemiological and meteorological data for Cali were analyzed from January 2000 through December 2011. Cross-correlations indicated that dengue fever incidence was positively associated with mean temperature, mean daily temperature range, and the El-Niño Southern Oscillation (ENSO) over time lags ranging from 2 weeks to 4–6 months, while a reverse association was found with mean relative humidity over similar time lags. Rainfall was negatively associated with dengue incidence over lags of 2–6 weeks, but an opposite relationship was found when using longer lags of 3-6 months. Overall, significant dengue outbreaks often occurred during dry periods when extreme daily temperatures were confined between 18ºC and 30ºC – the optimal range for mosquito survival and viral transmission. Two multivariate auto-regressive models were developed to predict dengue risk using the most significant meteorological parameters from 2000–2010. During model validation using 2011 data, the intra-seasonal model skillfully predicted dengue (R2 = 0.921) using the mean local daily temperature range at 2 weeks lag, while the inter-seasonal model provided skillful dengue forecasts (R2 = 0.824) using mean local relative humidity and equatorial Pacific sea-surface temperatures (in the Niño–4 region), both at a lag of 6 months.

We have developed two weather/climate-based forecast models that allow dengue outbreaks to be anticipated from 2 weeks to 6 months in advance. These models have the potential to enhance existing early warning systems, which can ultimately be used to support public health decisions regarding the timing and scale of vector control efforts.