Résumé
Introduction : Dengue fever (DF) is a serious and potentially life-threatening infection acquired through the bite of infected Aedes aegypti mosquitoes and responsible for major outbreaks in French Guiana. Dengue transmission is driven by complex interactions between hosts, vectors and viruses that are influenced by environmental, social and climatic factors which results inannual seasonality and multiannual variability. Although the onsets of outbreaks seem to be to be associated with the start of the rainy season, the mechanisms involved in the occurrence of epidemics are not well understood. With the objectives of explaining and predicting dengue outbreaks occurrence in French Guiana, we explored the impact of climatic factors on dengue epidemics at different spatial and temporal scales.Methods : For this analysis covering the 1991-2013 period, we used biologically confirmed DF epidemiological data, climatological parameters including Niño and NAO indices, atmospheric reanalysis gridded data and meteorological stations record including rainfalls, temperatures, relative humidity, sunstroke average and wind force.for. We performed the following analysis. We first assessed the association between incidence of DF and climatic variables to identify temporal predictors of DF outbreaks. Year-to-year disease variability was then investigated by linking atmospheric and oceanic synoptic patterns to the epidemiological regional context. Finally, contextual sequential patterns extraction techniques were used at local scale and weekly time step to identify the most significant climatic factors influencing DF incidence dynamics.Results : From 1991 to 2013, more than 23,600 biologically confirmed DF cases were recorded in French Guiana and 9 major outbreaks were identified. Annual activity was characterized by a seasonal increase of incidence levels during the first quarter of the year. We found that outbreak occurrence can be partly linked with summer Nino conditions, 6 months before the onset of the outbreak. Furthermore, a statistical model based only on climate indicators indicated that 30% of the disease variance from year-to-year is explained by the dry season rainfall, forecasting 75% of the epidemic years as so. At regional scale, the outbreak onset in specific territories was frequently positively or negatively associated with relative humidity, rainfall, and temperature. During epidemic period, local specific weather conditions were not predictive of the epidemic peak, when most of the cases occurred.Conclusion : Using complementary approaches at different geographic scales we explored the role of climate conditions to understand and predict DF outbreaks in French Guiana. Our findings will be useful to timely target appropriate public health interventions to mitigate the effects of major outbreaks, particularly in areas where resources are limited and the medical services may become overwhelmed by large epidemics.