From alternative hosts to alternative interventions: identifying drivers of epidemic dynamics for Japanese encephalitis virus in Bangladesh

Some mosquito-borne pathogens that cause disease in humans, like dengue, are transmitted between humans. However, there are many mosquito-borne viruses that are transmitted between other animals, for which humans are infected incidentally. These viruses are difficult to control; human vaccination cannot eliminate them. In addition, because they can usually infect more than one species, they can be present across a range of settings. One exemplar virus of this type is Japanese encephalitis virus (JEV), which is the most common cause of viral encephalitis in Asia. The disease is usually associated with rice fields and pig farming. Rice fields contain suitable habitat for larvae of vector mosquitoes and pigs can easily transmit virus to vectors. Despite this, JE disease has been reported from peri-urban areas of Asia in recent years and, in Bangladesh, the disease occurs in regions with low density of pigs. Here, also, there is a high density of cattle. How JE occurs in Bangladesh is unknown; cattle are preferred by vectors of JEV, but they cannot transmit the virus. I will address the hypothesis that alternative host species are responsible for epidemics of JEV in Bangladesh.

In northern Bangladesh domestic ducks, chickens and pigeons are regularly kept together in households in rice-growing areas. Experiments have shown that poultry can transmit virus to mosquitoes, but it is not known whether they contribute to transmission in nature, which bird species is most important to transmission and which vector species are involved. I will first establish the conditions which would permit JEV transmission in the presence of a high density of cattle. To do this, I will develop mechanistic models of transmission. I will use the models to test whether aggregation of poultry in a few households of a village either amplifies or dilutes transmission, compared with the scenario where poultry are evenly spread across a village with cattle. Aggregation of hosts that can transmit virus may be essential for onward transmission in cases where there is high density of hosts that cannot transmit the virus.

The proportion of blood meals on each host species is particularly important to estimate because it influences the rate of transmission from hosts to mosquitoes as well as mosquitoes to hosts. Using bird-baited traps I will quantify the relative preference of potential vector species to ducks, chickens, and pigeons. Focusing on the bird species most commonly attractive to potential vectors I will then determine how household host community composition and density influences the proportion of blood meals on birds. After updating models with data collected from the field, I will fit models to estimates of cattle JEV immune status and proportion of mosquitoes infected with JEV. Using fitted models, I will determine the relative role of each bird by removing them from the system and quantifying the effects on simulated outbreaks. I will also test the ability of models to predict relatively high and low risk areas of JEV transmission and validate these predictions using cattle immune status data from high and low risk villages. Lastly, I will use validated models to assess various control approaches, including potential ecological interventions that focus on reducing contact between vectors and the most important hosts.

An understanding of how transmission is maintained is important to inform potential control strategies and to better determine the environments that can support JEV transmission. There is currently no human vaccination programme in Bangladesh, despite the occurrence of the disease being similar to other countries before introduction of vaccine. The ability to target human vaccination to high-risk areas may make control more financially and logistically feasible. In addition, consideration of other interventions may be more sustainable in the long-term and be effective at reducing human JE when used with human vaccination.

Japanese encephalitis virus (JEV) is usually transmitted between pigs and mosquitoes. I will determine how JEV epidemics occur in northern Bangladesh where density of cattle is high and density of pigs is low. Cattle are often the preferred host for JEV vectors but cannot transmit the virus. I will test the hypothesis that alternative host species are both necessary and sufficient for JEV epidemics in this region, determine whether host density and community composition are predictive of JEV risk and use results to inform control strategies. The objectives to achieve these aims are: i) development of mechanistic models of JEV transmission; ii) field work to determine the relative contribution of alternative hosts to transmission and test model-based predictions of JEV risk; and iii) use validated models to quantify the potential impact of alternative interventions on transmission.

I will use stochastic models, incorporating non-homogeneous mixing between hosts and vectors, to determine the conditions under which onward transmission is possible, accounting for high density of dead-end hosts. Models will be parametrised using data collected on livestock production, vector abundance and feeding patterns, and fitted to estimates of livestock seroprevalence and JEV prevalence in vectors. Models will be validated by testing predictions of the differences in JEV risk between areas of contrasting livestock density and community composition. Ecological interventions, determined through stakeholder engagement, will then be assessed using models. I will use findings to produce policy-ready documents to help inform targeted human vaccination, the potential effects of combining alternative control strategies and assess risk of JEV spread to new regions. This project will contribute specifically to knowledge required to implement JE control in Bangladesh but also more generally on the potential for mitigating JEV risk elsewhere through sustainable approaches to control.