Infectious Disease Modeling Methods as Tools for Informing Response to Novel Influenza Viruses of Unknown Pandemic Potential

The 2009 influenza A (H1N1) pandemic was one of the most closely tracked and studied epidemics in history. Traditional epidemiological methods, such as outbreak investigations and laboratory-based surveillance, were rapidly used to inform policy decisions [1–4]. These methods were enhanced by newer computational techniques such as bioinformatics and digital surveillance methods [5]. Simultaneously, substantial contributions to the literature were made in the area of infectious disease modeling (IDM) [6–10]. This article is a guide to the way in which (IDM) can contribute to policy discussions and decision-making in preparation for, or during, an influenza pandemic. During an outbreak of influenza with pandemic potential, public health leaders ask a range of questions to inform situational awareness, help assess severity [11] and guide decisions that aim to control the spread and impact of disease. Critical questions include: What is the case-fatality ratio? What is the case-hospitalization ratio? When will disease incidence reach its peak? Who in the population should be prioritized for vaccination or antiviral treatment? How transmissible is the disease? What is the basic reproduction number (R0)? The accuracy with which these questions can be answered is time/data-dependent; as time passes and the outbreak progresses, more data become available to analyze. As such, a good compendium of methods for influenza outbreak analyses will be clear about the data requirements of each method and precisely when, during an evolving pandemic, the method might be most useful. Despite, and perhaps because of, the large accumulation of knowledge regarding the population dynamics of influenza, it has become difficult for those in public health agencies to know how modeling methods may be employed during an outbreak. In particular: What questions can be answered by a specific modeling method? At what stage of a pandemic might a particular modeling method be used? What data are required by the modeling method? In this article, we define IDM methods as techniques that include the mechanism of transmission of infection from an infected to an uninfected host. These are distinct from most methods in epidemiology because they explicitly model the transmission process and therefore the cause of infection; they therefore account for the “dependent” nature of infection, in other words, the fact that a major risk factor for infection is the population infection prevalence itself. IDM methods use data collected as part of routine surveillance (eg, clinical case counts, % polymerase chain reaction positives among tested clinically diagnosed cases, time of symptom onset), but they also often suggest new datasets that investigators might collect, such as detailed household-based questionnaires with an emphasis on determining exposure time windows [6, 7]. Furthermore, the mechanistic assumption of person-to-person transmission used in IDM methods is a strong one and can allow estimates to be made with less data than otherwise. Here, we collate and describe some of the most useful and well-tested IDM methods for use prior to, or during, an influenza pandemic. Our intent is to facilitate communication between public health practitioners and modelers during different stages of the response to pandemic threats, by reviewing these methods and linking them to the timeline of a pandemic response.