Epidemiological modelling of snakebite dynamics in Colombia disentangles the relationships between rainfall and envenoming cases in the country

Carlos Andrés Bravo-Vega Jr., Biomedical engineering, Universidad de los Andes, Bogotá D.C., Colombia, Juan M. Cordovez and Mauricio Santos-Vega, Biomedical Engineering, Universidad de los Andes, Bogotá D.C., Colombia

Background/Question/Methods
The role of climate forcing on infectious diseases dynamics has typically been addreses via retrospective analyses of spatially-aggregated incidence records over whole political regions. In the case of snakebite, a neglected tropical disease characterized by low treatment coverage and high mortality and morbidity if treatment is not applied at time, climate affects directly venomous snakes’ behavior and ecology thus driving incidence. Here, we combined stochastic epidemiological models and statistical inference methods to address the role of rainfall on the dynamics of snakebite incidence in Colombia. We used process-based models, an iterated particle filter (Partially Observed Markov Process (POMP)) approach and an extensive disease surveillance record. To demonstrate the effect of rainfall in the dynamics of snakebite and how this effect varies in space.

Results/Conclusions
First, we identified six zones, with different rainfall patterns and a disease seasonality in the country and we fitted the models for each of these regions. We found that rainfall drives snakebite incidence in regions with a marked dry season, where during these seasons’ envenoming risk decreases significantly. We show that the response to rainfall forcing is explained by the relationship between the encounter frequency with venomous snakes and precipitation. Interestingly, that encounter frequency also differs between regions, and it is higher in regions where Bothrops atrox can be found (Only two species account for 85% of the cases: Bothrops asper and Bothrops atrox, and their distributions don't overlap). These results underscore the non-homogeneous spatial distribution and temporal dynamics of snakebite risk and how that should be related with important aspects of venomous snakes’ natural history. The knowledge about natural history of neotropical venomous snakes is still poor, and it must be enhanced to strengthen the proposed process-based epidemiological models.