Methods: We use an agent-based model of Liberia and Sierra Leone, calibrated to the current epidemic to simulate 2,000 epidemics (1,000 for each country) of Ebola. Both the distribution of the number of cases of Ebola as well as the length of the epidemics are considered, the latter using Kaplan-Meier survival curves to measure the time until the end of an epidemic.
Results: Both Sierra Leone and Liberia show distinctly multi-modal distributions for epidemic size and the duration of the epidemic. In both countries over 20% of the epidemics end with very few cases, and over half of the epidemics are the relatively small, short-lived outbreaks that previously characterized the epidemiology of Ebola. The remainders are long-lived, severe outbreaks like the one currently ongoing in West Africa, with ~30% of epidemics larger and longer-lived than the current outbreak.
Conclusion: The same model parameters that produce the current outbreak also produce a large number of short, relatively small outbreaks consistent with the previous epidemiology of Ebola. As these simulations only vary based on random chance, the hypothesis that the current outbreak is merely the result of chance, rather than changes to the underlying virus or the structure of the West African population cannot be ruled out. Furthermore, these results suggest that while the current epidemic is undeniably serious, it is far from the worst-case scenario, and the occurrence of other large, long-lasting Ebola outbreaks should be planned for in the future.
B. Lewis, None
M. V. Marathe, None
M. Wilson, None
J. Chen, None
S. Eubank, None