Cost-Effectiveness of Prophylactic Zika Virus Vaccine in the Americas

Affan Shoukat; Thomas Vilches; Seyed M. Moghadas


Emerging Infectious Diseases. 2019;25(12):2191-2196. 

In This Article


Simulation Model

We adopted a previously established agent-based simulation model for the dynamics of Zika virus infection, incorporating both vector and sexual transmission.[19,20] For infection dynamics, the human population was divided into susceptible, exposed and incubating, infectious (symptomatic and asymptomatic), and recovered categories (Appendix Figure 1 We stratified the mosquito population into susceptible, exposed and incubating, and infectious groups. We parameterized the model with country-specific demographics (age and sex distributions and fertility rates), and calibrated it to attack rates[17,18] estimated for the 2015–2017 outbreaks (Appendix Tables 1–4, Figures 2–4). These attack rates were considered to be the proportion of the population that was infected (representing the level of herd immunity) at the start of simulations for each country in the evaluation of vaccination scenarios. We compiled parameters specific to Zika virus infection in both human and mosquito populations, along with costs associated with the disease and vaccination (Appendix Tables 5, 6). Further details of the model and its implementation are provided in the Appendix; for reproducibility, the computational model can be accessed at

Infection Outcomes

We considered microcephaly and Guillain-Barré syndrome (GBS) as outcomes of infection. The risk for microcephaly was highest (5%–14%) for infections occurring during the first trimester of pregnancy (which ends at 97 days) and decreased to 3%–5% for infections occurring during the second and third trimesters.[21–23] We set a probability of 0.798 for survival past the first year of life for infants with microcephaly.[24] Life expectancy of infants with microcephaly who survived the first year of life was reduced by 50%, from 70 years to 35 years, on average.[25] The risk for GBS with Zika virus infection in adults was 0.025%–0.06%.[26]

Vaccination and Cost-effectiveness

We implemented vaccination scenarios corresponding to the recommended strategies in the VTPP.[16] The vaccination coverage was set to 60% for women of reproductive age at the onset of simulations. For pregnant women in the same age group, the vaccination coverage was set to 80% initially and continued at 80% throughout the simulations. We also considered a vaccination coverage of 10% for other persons 9–60 years of age. In the absence of efficacy data, we assumed that a single dose of vaccine provides a protection efficacy of 60%–90% against infection, which was sampled for each vaccinated person and implemented as a reduction factor in disease transmission. Infection following vaccination (if it occurred) was assumed to be asymptomatic. Furthermore, we assumed that vaccination has no effect on the risk of microcephaly in pregnant women if infection occurred.

For cost-effectiveness analysis, we considered both short- and long-term medical costs specific to each country (Appendix Table 6).[9] Short-term costs included physician visits and diagnostic tests for symptomatic Zika virus infection in pregnant women. For microcephaly in infants and GBS in adults, we considered lifetime direct medical costs related to hospitalization, treatment, and other associated outcomes. We quantified the long-term sequelae of microcephaly by disability-adjusted life-years (DALYs) with disability weight (i.e., severe intellectual disability) extracted from the Global Burden of Disease study.[27] For given vaccination costs per individual (VCPI), we calculated the incremental cost-effectiveness ratios (ICERs) and averaged them over simulations (Appendix). Both DALYs and direct lifetime costs were based on a 3% discounting rate annually.[9,25] For cost-effectiveness analysis, we considered the World Health Organization standards of using the per capita gross domestic product (GDP) as a threshold of willingness to pay.[28] The vaccination program was considered very cost-effective for ICER values up to the per capita GDP and cost-effective for ICER values up to 3 times the per capita GDP. We also considered a range of willingness to pay values to inform decisions on vaccine cost-effectiveness in settings in which the per capita GDP threshold may not be applicable. Using a nonparametric bootstrap method, we generated cost-effectiveness acceptability curves for each country and performed cost-effectiveness analysis from a government perspective. All costs are reported in 2015 US dollars.

We ran 2,000 Monte Carlo simulations of Zika virus infection dynamics with a scaled-down population of 10,000 persons for each country. Each simulation was seeded with a single case of Zika virus in the latent stage and run for a time horizon of 1 year with a daily time-step, beginning with a high-temperature season. For each simulation, we recorded the daily incidence of infection and disease outcomes and used them for cost-effectiveness analysis, as well as estimating the percentage reduction of microcephaly attributable to vaccination. DALYs were calculated for the lifetime of each case of microcephaly. Only epidemic curves that had ≥1 secondary cases by the end of simulations were considered in the cost-effectiveness analysis.