Projections of epidemic transmission and estimation of vaccination impact during an ongoing Ebola virus disease outbreak in Northeastern Democratic Republic of Congo, as of Feb. 25, 2019

Worden, Lee; Wannier, Rae; Hoff, Nicole A.; Musene, Kamy; Selo, Bernice; Mossoko, Mathias; Okitolonda-Wemakoy, Emile; Muyembe Tamfum, Jean Jacques; Rutherford, George W.; Lietman, Thomas M.; Rimoin, Anne W.; Porco, Travis C.; Kelly, J. Daniel

As this report, written in February 2019, goes to print in June 2019, we note that, as even a casual observer of the situation will know, the outbreak has exceeded the final size projections listed here, and the spread of disease and mortality is still ongoing. The predictions of our stochastic branching process and regression models reflect an assumption that this outbreak can be modeled by the ensemble of past outbreaks located in countries with previous experience responding to EVD outbreaks. This criterion of non-naive countries has the consequence of excluding the massive West African outbreak of 2013-2016 from the distribution of past outbreaks modeled, which may also be justifiable per se as it is a single statistical outlier. However, as the outbreak has exceeded the resulting projections, their assumptions must be called into question. In particular, the range of quenching scenarios considered by the branching process model appears to overestimate whatever damping of transmission may be occurring in the present outbreak, leading to its undersized projections of future case counts. An additional consequence of overestimating quenching is that because vaccine coverage is modeled as a reduction in transmission from the quenched rates assumed otherwise, the resulting estimates of vaccine coverage may be overly low. The ability of the non-mechanistic, data-agnostic Gott's rule projection to predict the unusually large size of this outbreak more accurately than the other models suggests that broad estimates of its kind, reflecting the true extent of ignorance imposed by the unpredictable nature of events such as this outbreak, may be valuable tools in epidemic prediction and decision support despite the natural desire to predict outcomes with more precision.