Although we conservatively assumed the probability of clinical infection to be independent PD0325901 mouse of age, we performed sensitivity analyses to consider age dependence as has been previously considered. We discuss our mathematical model and related assumptions in more detail in the supplementary material (Supplementary material S1). For all simulations, we assumed that that the vaccine was

equally effective against serotypes DENV-1, DENV-3 and DENV-4 (vaccine efficacy = 0.8, after 3 doses) but only partially effective against DENV-2. We also assumed that vaccine-derived immunity does not wane. Rollout of the vaccine consisted of 3 years of catch-up targeting children 2–15 years of age, followed by regular vaccination of 2–5 year olds. The vaccine Selleck Regorafenib was administered in up to three doses that were given on average every six months apart. Vaccination rates in catch-up and routine programs were constant over time and set so that vaccination

coverage would reach 89% among 2–5 year olds and 69% in 2–15 year olds after 5 years. These vaccination rates were chosen to roughly correspond with the rate of vaccination achieved in Thailand with the Japanese Encephalitis three-dose vaccination using a combination of catch-up and routine immunization campaigns. To explore the effects of vaccination at the population level, we compared the cumulative number of clinically apparent dengue cases in the 10 years after vaccine introduction, to the cumulative number of cases over the same period in the counterfactual population (i.e. same population had the vaccine not been introduced). We also isolated overall, direct and indirect vaccine effects as proposed by Halloran et al. [23]. In addition, we defined a counterfactual vaccine effect, comparing the cumulative incidence in vaccinated individuals of the vaccinated population to the cumulative incidence in “vaccinated” individuals

of the counterfactual population (Supplementary material S1). Since timing and of vaccine introduction may impact the short and medium term effects of vaccination, we performed simulations introducing the vaccine at different points in the multiannual dengue cycle. We present vaccine effects that are averages over eight possible introduction years. We calibrated the model, at steady state, to the transmission dynamics of dengue in Rayong, Thailand, a traditionally hyperendemic setting (Fig. 1). To fit the model to the demography of Rayong, we used data from the 2010 Thai Census [24] (Supplementary Fig. S2.1). To estimate transmission parameters, we used age-specific incidence data from the Ministry of Public of Public Health (2002–2010) and age-stratified serological data from a seroprevalence study conducted among school-children in Rayong in 2010 [15] and [25].