Responding to yellow fever outbreaks in West and Central Africa: Rapid prioritization assessment for the pre-emptive vaccination campaigns

06 Jul 2018
Jean K, Hamlet A, Dorigatti I, Gaythorpe K, Imai N, Cibrelus L, Benzler J, Garske T

Introduction: Recent yellow fever (YF) outbreaks in Africa, such as 2016 in Angola, or 2017 in Nigeria, have demonstrated the ongoing threat of large-scale urban YF outbreaks. Moreover, outbreaks of other mosquito-borne diseases (Zika, chikungunya) have increased awareness of the potential for rapid, international spread of arboviruses. Swift outbreak response, for YF primarily in the form of vaccination, is therefore essential to prevent local and international spread. However, limited global stockpiles means that the prioritization of geographic areas for pre-emptive vaccination campaigns is required (i.e. vaccination campaigns targeting areas that are at risk of disease introduction and spread but as yet unaffected). Conducting a rapid prioritization assessment based on the risk of disease spread is thus highly valuable to inform decisions on vaccination activities in a context of emergency.

Objective: To develop a method for rapid risk assessment of YF spread to prioritize sub-national administrative units (hereafter called province) for pre-emptive campaigns. Specific requirements for this method are:

  • speed of implementation in the context of emergency response;
  • transparent methodology to allow discussions with and feedback from decision-makers.

Method: We developed a heuristic method to quantify the risk of YF spread by integrating multiple data streams: population sizes, estimates of existing vaccine-induced population-level immunity, recent incidence of yellow fever cases in the province and travel flows between provinces. The resulting risk score primarily reflects the expected absolute number of yellow cases in the respective province, accounting for local cases and the risk of case importation. Based on their risk score, provinces are ranked according to priority for vaccination and target population sizes are estimated. This baseline, quick-to-compute, risk assessment can be refined by integrating additional elements. For example, the presence and population size of large urban centres at the province level may be relevant to characterize the risk of urban outbreaks, and of international spread in the case of highly connected urban centres. Similarly, independent estimates of the local transmission potential of the disease produced by a mathematical model can be combined into this risk metric.

Results: This heuristic method has been used in collaboration with the World Health Organization in the context of urgent response to yellow fever outbreaks that affected Angola and the Democratic Republic of Congo (DRC) in 2016 and Nigeria in 2017. Results were provided in a form that allowed decision-makers to easily and interactively adjust the relative weights of different factors and visualize the effect on the results. Moreover, the transparency of the method allowed decision-makers to provide feedback and to request the integration of additional elements considered as relevant for outbreak control (for instance: trans-border movements in Angola-DRC). This risk assessment contributed to inform decision for mass vaccination campaigns conducted in the DRC in 2017 and for campaigns (still under consideration) in Nigeria in 2018.

Conclusion: By integrating different, mostly publicly available, data streams, we developed a risk assessment method that can be quickly implemented in the context of yellow fever outbreak response. The transparency and flexibility of the method enhanced interactions with decision-makers to refine estimates.