Science Spotlight

Reconstructing Zika’s travels in the Americas

From the Vaccine and Infectious Disease Division and the Center for Inference and Dynamics of Infectious Diseases

            Zika virus (ZIKV) is an RNA virus of the family Flaviviridae. Infection with ZIKV causes mild disease consisting of fever, rash, and conjunctivitis; however, it is asymptomatic in 80% of infected individuals. Transmission of the virus is mainly by mosquito bite, but may also be sexually transmitted or transmitted from mother to fetus. The virus was first discovered in Uganda in 1947 and had been considered a neglected tropical disease until a major outbreak causing microcephaly in Brazil in 2015 caused the World Health Organization (WHO) to take notice; this lead to ZIKV being declared a public health emergency by the WHO in 2016. During this period the outbreak spread to over 200 countries and territories. Even with the increased concern there are many epidemiological facts about the recent ZIKV outbreak that are still unknown, including when and where the virus was introduced to Brazil and the spread of the virus prior to public health documentation with the WHO declaration. In a recently published PNAS paper researchers created a data driven stochastic spatial epidemic model to analyze the spread and magnitude of ZIKV infection in the Americas. This model used seasonal and population factors to create outbreak projections starting in 2013 and extending into February 2017. This research was supported by the Center for Inference and Dynamics of Infectious Diseases (CIDID), led by Dr. Betz Halloran at Fred Hutchinson Cancer Research Center (Vaccine and Infectious Disease Division). In regards to the model, Dr. Halloran said, “This simulation model of the spread of Zika virus in the Americas is the most advanced, state-of-the-art model available to date. It includes human movement data, human census data, mosquito data, socioeconomic data, and climate data among other sources of data. It was validated against independent data including the imported cases in the US. “

Posterior distribution of ZIKV in 12 Brazilian transportation hubs between April 2013 and June 2014 with the likelihood of the epidemic peaks replicating in Colombia.
Posterior distribution of ZIKV in 12 Brazilian transportation hubs between April 2013 and June 2014 with the likelihood of the epidemic peaks replicating in Colombia. graph provided by Dr. Halloran

In order to predict the introduction date of ZIKV into the Americas the group looked at 12 travel hubs in Brazil over a range of dates. From this they found that the posterior distribution (the conditional probability that is given after relevant background is taken into account) suggests that introduction was between August 2013 and April 2014 with a few potential locations including Rio de Janeiro, Brasilia, Fortaleza, and Salvador (see figure). The model generated ZIKV epidemic curves (showing number of cases and time out outbreaks) for 24 countries in the Americas, which are in good agreement with previous models. Data for all 24 countries can be found at Looking specifically at Brazil the model predicts two separate waves of ZIKV infection, one occurring between January and July 2015 in the northeast region and a later one in the rest of the country. The model’s prediction is in agreement with data derived from confirmed cases such as predicting the introduction to Colombia in March to April of 2015. The model was validated against independent data from Brazil, Colombia, and the USA not used for its development.. Overall, the trends replicated but the model produced a greater number of cases, possibly due to (1) under reporting of asymptomatic or mild disease,  (2) the effects of mosquito control (3) spread by non-Aedes mosquitos, and (4) sexual transmission. While these factors likely affect ZIKV spread there was not enough data or understanding to model them.

Looking at the model or collected epidemical data, the virus spread appears to be constrained by seasonality around mosquito season, the main route of transmission. Another limit of infection is access to susceptible hosts. Much like Dengue virus, a similar mosquito borne disease, outbreaks peak once new births introduce more susceptible hosts then ebb as the pool diminishes. It is important to understand the dynamics and projections of future infections to develop effective ZIKV control and vaccine trails. As pointed out by Dr. Halloran, “The model has been used to predict the number of Zika-associated microcephaly cases to be expected in the near future in the different countries. The model can be used for projecting where we might expect transmission in the coming year or so as we plan to select sites for Zika vaccine field trials. We are also using the simulation output to develop designs and statistical methodology for Zika vaccine field trials.” The current model allows for estimates of infection and of the number of babies born to women infected with ZIKV in their first trimester (when birth defects do to ZIKV are highest risk) and the number of microcephaly cases through 2017.


This work was funded by Models of Infectious Disease Agent Study, National Institutes of General Medical Sciences Grant, and the European Commission Horizon 2020 CIMPLEX Grant. Work was also supported by the NIGMS-funded Center for Inference and Dynamics of Infectious Diseases (CIDID), based at the Hutch (

Zhang Q,Sun K,Chinazzi M,Pastore Y Piontti A,Dean NE,Rojas DP,Merler S,Mistry D,Poletti P,Rossi L,Bray M,Halloran ME,Longini IM, Jr.,Vespignani A. 2017. Spread of Zika virus in the Americas. Proc Natl Acad Sci U S A.

Subscribe to RSS

RSS feeds are best viewed in browsers other than Chrome