Modeling the Impact of Land Use Change (Irrigation) on the Transmission Dynamics of West Nile Virus (Wnv) in Tana River County, Kenya

Abstract

West Nile Virus (WNV) is an arbovirus that is transmitted by Culex mosquitoes. Birds are its amplifying hosts while horses, humans and other animals are dead-end hosts. WNV mainly causes fatalities in birds and horses while in humans, it mostly causes subclinical infection. However, fatal meningo-encephalitis is experienced in about 2 to 17% of those infected. Several outbreaks of WNV have been recorded in some countries like Greece, Israel, Romania, Russia and USA. There is a WNV vaccine for horses but not for humans. The aim of this study was to develop a parsimonious epidemiological model to study how irrigation affects the transmission dynamics of West Nile virus in Tana River County in Kenya. The study also evaluated the impact of use of mosquito adulticides repellents and larvicides as WNV control interventions. We formulated a SEIR (Susceptible, Exposed, Infectious, and Recovered) model with compartments for mosquitoes, birds and humans. Parameters were originated from published literature while the meteorological data for the study site was obtained from the meteorological department. Ordinary differential equations (ODEs) generated from the model were used in R programming language to simulate the risk of WNV under various study scenarios. The simulation was driven by mosquito suitability index which is a function of irrigation and rainfall patterns. The model outcome indicated that the irrigation increased both the amount of water and suitability of the habitat for mosquito breeding about five-fold. This resulted in about three-fold increase in vector density and risk of WNV. Irrigation therefore increased the risk of WNV transmission in Tana River by about 200%. The comparative efficacy analysis of the vector control interventions showed that use of mosquito adulticides was the most effective method followed by repellents and lastly larvicides. The model also showed that WNV epidemics may be seasonal in nature for rainfall situations where they are likely to occur about one month after peak rainfall. However, for irrigated situations, the risk of WNV is likely to be a perennial phenomenon. This model may be used as a framework to guide decision making on the timing and choice of WNV control intervention. However, the model needs to be developed further by incorporating more factors to improve on its accuracy.