Estimation of the efficacy of seasonal influenza vaccine : vaccine development, statistical modelling and trial design

Influenza is an infectious seasonal disease against which yearly vaccination is recommended, especially for high risk populations. For a new vaccine, efficacy is classically assessed in large phase III trials. Unfortunately, in the past years, many trials have led to unexpected results. This work first aims to identify the particularities of the context of influenza with regards to the vaccine development. We confirm the link between post vaccination antibody response and vaccine efficacy without being able to identify an absolute threshold of protection. Combining simulations and analytical results, we then substantiate the limits of the actual statistical regression models used to estimate vaccine efficacy. To reach this goal, we had to develop a simulation framework for the generation of phase III clinical trials time-to-infection data. Our methodology brings together particularities of influenza and other infectious diseases as well as historical data. Finally, we propose a new methodology for the assessment of vaccine efficacy, taking into account the effect of strain heterogeneity on vaccine efficacy. By using hierarchical modelling and predictive intervals, we provide better insight on the range of future VE across seasons and geographical regions.