Burden is in the eye of the beholder: Sensitivity of yellow fever disease burden estimates to modeling assumptions

Geographically stratified estimates of disease burden play an important role in setting priorities for the management of different diseases and for targeting interventions against a single disease. Such estimates involve numerous assumptions, which uncertainty about is not always well accounted for. We developed a framework for estimating the burden of yellow fever in Africa and evaluated its sensitivity to assumptions about the interpretation of serological data and choice of regression model. We addressed the latter with an ensemble approach, and we found that the former resulted in a nearly twentyfold difference in burden estimates (range of central estimates: 8.4x104-1.5x106 deaths in 2021-2030). Even so, statistical uncertainty made even greater contributions to variance in burden estimates (87%). Combined with estimates that most infections go unreported (range of 95% credible intervals: 99.65-99.99%), our results suggest that yellow fever9s burden will remain highly uncertain without major improvements in surveillance.