Key data for outbreak evaluation: building on the Ebola experience

Cori, Anne; Donnelly, Christl A.; Dorigatti, Ilaria; Ferguson, Neil M.; Fraser, Christophe; Garske, Tini; Jombart, Thibaut; Nedjati-Gilani, Gemma; Nouvellet, Pierre; Riley, Steven; Van Kerkhove, Maria D.; Mills, Harriet L.; Blake, Isobel M.

However, interpreting the results of such analyses can be challenging, as they might be prone to bias and confounding. Efficacy (which measures the impact of an intervention under ideal and controlled circumstances) and effectiveness (which measures the impact of an intervention under real-world conditions) of an intervention (e.g. vaccine) are best measured in a trial setting (either individual-or cluster-randomized [68 -70] ). However, performing trials to evaluate the comparative impact of different multi-intervention packages is impractical. Dynamic epidemic models, where the interventions of interest can be explicitly incorporated, allow the impact of such intervention packages to be predicted [71] . However, outputs of such models are strongly determined by the underlying assumptions and parameter values. Hence they require careful parametrization, supported by data such as intervention efficacy and the size, infectivity, susceptibility and mixing of different groups [72, 73] . These parameters may not be straightforward to estimate, as we discuss in §3 using examples from the West African Ebola epidemic.