Avoidable errors in the modelling of outbreaks of emerging pathogens, with special reference to Ebola

As an emergent infectious disease outbreak unfolds, public health response isreliant on information on key epidemiological quantities, such as transmissionpotential and serial interval. Increasingly, transmission models fit toincidence data are used to estimate these parameters and guide policy. Somewidely used modelling practices lead to potentially large errors in parameterestimates and, consequently, errors in model-based forecasts. Even moreworryingly, in such situations, confidence in parameter estimates and forecastscan itself be far overestimated, leading to the potential for large errors thatmask their own presence. Fortunately, straightforward and computationallyinexpensive alternatives exist that avoid these problems. Here, we first use asimulation study to demonstrate potential pitfalls of the standard practice offitting deterministic models to cumulative incidence data. Next, we demonstratean alternative based on stochastic models fit to raw data from an early phase of2014 West Africa Ebola virus disease outbreak. We show not only that bias isthereby reduced, but that uncertainty in estimates and forecasts is betterquantified and that, critically, lack of model fit is more readily diagnosed. Weconclude with a short list of principles to guide the modelling response tofuture infectious disease outbreaks.