Mapping the cryptic spread of the 2015–2016 global Zika virus epidemic

BACKGROUND: Zika virus (ZIKV) emerged as a global epidemic in 2015–2016 from Latin America with its true geographical extent remaining unclear due to widely presumed underreporting. The identification of locations with potential and unknown spread of ZIKV is a key yet understudied component for outbreak preparedness. Here, we aim to identify locations at a high risk of cryptic ZIKV spread during 2015–2016 to further the understanding of the global ZIKV epidemiology, which is critical for the mitigation of the risk of future epidemics. METHODS: We developed an importation simulation model to estimate the weekly number of ZIKV infections imported in each susceptible spatial unit (i.e. location that did not report any autochthonous Zika cases during 2015–2016), integrating epidemiological, demographic, and travel data as model inputs. Thereafter, a global risk model was applied to estimate the weekly ZIKV transmissibility during 2015–2016 for each location. Finally, we assessed the risk of onward ZIKV spread following importation in each susceptible spatial unit to identify locations with a high potential for cryptic ZIKV spread during 2015–2016. RESULTS: We have found 24 susceptible spatial units that were likely to have experienced cryptic ZIKV spread during 2015–2016, of which 10 continue to have a high risk estimate within a highly conservative scenario, namely, Luanda in Angola, Banten in Indonesia, Maharashtra in India, Lagos in Nigeria, Taiwan and Guangdong in China, Dakar in Senegal, Maputo in Mozambique, Kinshasa in Congo DRC, and Pool in Congo. Notably, among the 24 susceptible spatial units identified, some have reported their first ZIKV outbreaks since 2017, thus adding to the credibility of our results (derived using 2015–2016 data only). CONCLUSION: Our study has provided valuable insights into the potentially high-risk locations for cryptic ZIKV circulation during the 2015–2016 pandemic and has also laid a foundation for future studies that attempt to further narrow this key knowledge gap. Our modelling framework can be adapted to identify areas with likely unknown spread of other emerging vector-borne diseases, which has important implications for public health readiness especially in resource-limited settings.