Abstract
As an emergent infectious disease outbreak unfolds, public health response is reliant on information on key epidemiological quantities, such as transmission potential and serial interval. Increasingly, transmission models fit to incidence data are used to estimate these parameters and guide policy. Some widely used modelling practices lead to potentially large errors in parameter estimates and, consequently, errors in model-based forecasts. Even more worryingly, in such situations, confidence in parameter estimates and forecasts can itself be far overestimated, leading to the potential for large errors that mask their own presence. Fortunately, straightforward and computationally inexpensive alternatives exist that avoid these problems. Here, we first use a simulation study to demonstrate potential pitfalls of the standard practice of fitting deterministic models to cumulative incidence data. Next, we demonstrate an alternative based on stochastic models fit to raw data from an early phase of 2014 West Africa Ebola virus disease outbreak. We show not only that bias is thereby reduced, but that uncertainty in estimates and forecasts is better quantified and that, critically, lack of model fit is more readily diagnosed. We conclude with a short list of principles to guide the modelling response to future infectious disease outbreaks.
Highlights
The success of model-based policy in response to outbreaks of bovine spongiform encephelopathy [1] and foot-and-mouth disease [2,3] established the utility of scientifically informed disease transmission models as tools in a comprehensive strategy for mitigating emerging epidemics
Recent examples in which model-based forecasts were produced within weeks of the index case include severe acute respiratory syndrome (SARS; [4,5]), pandemic H1N1 influenza [6], cholera in Haiti and Zimbabwe [7], Middle East respiratory syndrome (MERS; [8]), and lately, Ebola virus disease (EBVD) in West Africa [9,10]
For Sierra Leone, the disagreement between fitted model and data is not as great, at least as measured by this criterion. These diagnostics caution against the interpretation of the outbreaks in Guinea and Liberia as simple instances of susceptible–exposed – infectious – recovered (SEIR) dynamics, and call for a degree of scepticism in inferences and forecasts based on this model
Summary
The success of model-based policy in response to outbreaks of bovine spongiform encephelopathy [1] and foot-and-mouth disease [2,3] established the utility of scientifically informed disease transmission models as tools in a comprehensive strategy for mitigating emerging epidemics. An inexpensive and common strategy is to formulate deterministic transmission models and fit these to data using least squares or related methods These approaches seek parameters for which model trajectories pass as close to the data as possible. In general one expects that violation of model assumptions will introduce some degree of bias, in this case since both the raw and cumulative incidence curves generically grow exponentially at a rate determined by R0, estimates of this parameter are fairly accurate, on average, when data are drawn, as here, from the early phase of an outbreak. This apparent precision is an illusion, as figure 1d shows This figure plots the achieved coverage ( probability that the true parameter value lies within the estimated confidence interval) as a function of the magnitude of measurement error and the choice of data fitted.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
