For the random selection of the households to be surveyed, we used satellite imagery from Google, Bing, ESRI, and Yandex Satellite, through the plugin QuickMapServices in QGIS, as described in the original publication “Novel transdisciplinary methodology for cross-sectional analysis of snakebite epidemiology at national scale”, which is available at PLOS Neglected Tropical Diseases. The same publication also describes the functions used in QGIS and a summarized R script to randomly select the primary sampling units using Probability Proportional to Size (PPS) sampling.
We used the Integrated Nested Laplace Approximations (INLA) method in the study, a robust Bayesian modeling method that borrows strength from neighboring areas and observations. This allowed us to confidently estimate the spatial risk of snakebite for humans and animals.
We were able to identify the hotspots of snakebite risk in the Terai region of Nepal. We also highlighted very important factors that either significantly increase or decrease the risk of snakebite for a person living in a given area of 1–2 sq. km per year. Although poverty was already recognized to be an important factor increasing the risk of snakebites, the use of the Poverty Probability Index (PPI) and its importance as a standardized indicator of poverty was unprecedented.
The eastern and middle Terai regions suffer more incidents of envenoming than the western Terai region. We offered a possible explanation related to the distribution of the krait species (Bungarus genus) and the apparently eastern-predominant custom of sleeping on the floor. From this research, it was again evident that snakebite is a multifactorial disease involving not only humans, domestic animals and snakes, but also behavioral, socioeconomic, cultural, religious and environmental factors.
We could not compute the spatial risk for a few domestic animals. Nonetheless, the result for animals (as previously for humans) revealed some strong risk-increasing and risk-reducing variables that could be used for awareness campaigns in rural communities, such as having straw storages near human dwellings or animal sheds — especially since recent publications from our team showed an important total incidence of 61.5 (51.0–74.1) cases per 1,00,000 animals and significant economic losses for affected households. However, the most important variable was the average minimum temperature of the coldest month, which we considered to be linked to the biology of snakes and possible animal holding practices.
People working in agricultural fields are more exposed to encounters with snakes that hunt rodents and other animals in these fields. In our study, higher NDVI (normalized difference vegetation index) values indicated a reduction in snakebite risk. As discussed in the paper, one possible explanation is the difference between the average NDVI value of forests (higher) and that of crops (lower). However, in other circumstances and countries, the same high NDVI values have been associated with higher snakebite incidence.
As per the study, there are some covariates directly or indirectly related to poverty that were found to have an influence on the risk of snakebites. These covariates are: Poverty Probability Index (PPI), distance to water, food storage, and straw storage. Despite a complex dynamic, the snakebite risk of a household inevitably increases with the number of its members.
