US influenza vaccination policy is problematic. First, influenza vaccine effectiveness is low (e.g., 27%) in the elderly, the group most likely to die. Second, interference between successive doses has recently been described. Third, there are many vaccine formulations, with differing valences, efficacies, administration routes, and allowable ages of use, complicating policy recommendations. Fourth, the tension between the timing of vaccination, annual epidemics, and duration of immunity is clear: if waning immunity occurs with early vaccination and a late epidemic occurs, protection may be reduced whereas an early epidemic may occur before vaccination is completed. To address these challenges, we will use complimentary computational modeling techniques: Markov cohort decision analysis (DA), equation-based dynamic transmission modeling (EBM), and agent-based modeling (ABM). DA provides a clear visual framework for the breadth of strategies under consideration and is relatively quicker for initial analyses. EBM adds to this the dynamics of disease transmission and indirect (herd immunity) effects of vaccination strategies. ABM, conducted on supercomputers, adds further detail through simulating autonomous persons and their spatial and temporal demographics and social interactions during disease spread through a population. Because ABM is computationally intensive, strategies considered by ABM will be narrowed using DA and EBM. Using all three modeling techniques offers a balance of clarity and the complexity of reality, as well as the opportunity to perform validity comparisons between techniques.
Aim 1 : Determine the optimal vaccine selection strategy that minimizes disease burden and resource use in various age groups in 1) the US population and 2) various medical practice populations.
Aim 2 : Determine the ideal timing of annual vaccination, weighing the potential impact of early vaccination, waning immunity, and epidemic timing, interference, and missed vaccination opportunities..
Aim 3 : Using ABM, compare the trade-offs of effectiveness, duration, herd immunity, side effects, achievable vaccination rates, and cost of inactivated vaccines to those of potential universal vaccines in different US locations/populations and determine universal vaccine characteristics that favor its adoption. The research team is experienced in modeling, possesses diverse skill sets, has worked together, has access to epidemiologic data in the Influenza Vaccine Effectiveness Network, and has a strong publication record in vaccination issues, encompassing modeling, cost- effectiveness analysis, and policy.
Flu vaccine policy for the nation must account for the dynamic nature of flu and the growing number of flu vaccine choices. We will use multiple mathematical modeling methods, including use of a supercomputer, to make better flu vaccine policy choices.