An exploration of the role of asymptomatic infections in the epidemiology of dengue viruses through susceptible, asymptomatic, infected and recovered (SAIR) models.

It is estimated that 20-97% of all dengue infections could be asymptomatic. I used SIR models to investigate the epidemiological role of such infections, by adding an asymptomatic class (SAIR models). Upon infection in one of the models, a human becomes either symptomatic or asymptomatic. In the other, a human becomes asymptomatic and may progress to being symptomatic. The robustness of results from these models is examined by incorporating the mosquito-vector into one of the models, followed by simulating epidemic dynamics stochastically. Results from the first two models were very similar, with epidemics typically lasting less than one year. When mosquitoes were explicitly modelled in a high-transmission setting, if the level or duration of infectivity from asymptomatic infections was high relative to symptomatic infections, dengue would become endemic. Under stochastic simulation this effect of asymptomatic infections leading to dengue persisting was no longer guaranteed. Longer durations in asymptomatic infections had a higher chance of causing dengue's persistence in stochastic simulation, indicating that this may be more of a key determinant for dengue's persistence to 10 years than the infectivity of such infections. Otherwise, the level and duration of infectivity from asymptomatic infections had similar effects on R0 and other epidemiological measures. With all models, outbreaks often led to a larger proportion of the population being immune than suggested by monitoring symptomatic dengue infections. This population would be at risk of developing severe dengue in a subsequent outbreak with a different dengue serotype, and would have to be determined via expansion factors.