Using a metapopulation epidemiologic model to estimate the exceedance probabilities of morbidity and mortality resulting from hypothetical influenza pandemic scenarios

Introduction: Public health preparedness efforts often utilize pandemic scenarios for planning. While these scenarios are often used as benchmarks, the likelihood of these scenarios is not well-characterized. Methods: Utilizing data from historical events and research on currently circulating influenza subtypes, we developed a catalog of over 18,000 possible influenza pandemic scenarios. Scenarios were parameterized by sampling from statistical distributions of start location, number of years between pandemics, basic reproduction number, and case-fatality ratio. These scenarios were simulated using a global, stochastic, metapopulation epidemiologic model, which estimated the morbidity and mortality levels from each scenario. This model explicitly incorporated population movement and mitigation measures. Based on the resultant distribution of modeled morbidity and mortality, exceedance probability metrics were calculated. Results: The probability of an influenza pandemic starting in any given year was estimated to be between 3%-4%, and timing between influenza pandemics was estimated to range from 5 - 150 years. Mortality estimates from the suite of 18,000 scenarios ranged from mild, causing few excess deaths, to catastrophic, causing tens of millions of deaths in the US. We also estimated the exceedance probabilities of the HHS pandemic planning scenarios, causing 209,000 and 1.9 million deaths in the US. Discussion and Conclusion: By quantifying the exceedance probabilities of different influenza pandemic scenarios using epidemiologic modeling, preparedness efforts can be better tailored to address the anticipated levels of morbidity and mortality. Higher-probability scenarios could be targeted for preparedness planning efforts, which could lead to more efficient allocation and use of scarce resources.