The impact of influenza vaccination on infection, hospitalisation and mortality in the Netherlands between 2003 and 2015

Influenza epidemics annually cause substantial morbidity and mortality. For this reason, vaccination is offered yearly to persons with an elevated risk for complications. Assessments of the impact of vaccination are, however, hampered by year-to-year variation in epidemic size and vaccine effectiveness. We estimate the impact of the current vaccination programme comparing simulations with vaccination to counterfactual simulations without vaccination. The simulations rely on an age- and risk-structured transmission model that tracks the build-up and loss of immunity over successive seasons, and that allows the vaccine match to vary between seasons. The model parameters are estimated with a particle Monte Carlo method and approximate Bayesian computation, using epidemiological data on vaccine effectiveness and epidemic size in the Netherlands over a period of 11 years. The number of infections, hospitalisations and deaths vary greatly between years because waning of immunity and vaccine match may differ every season, which is in line with observed variation in influenza epidemic sizes. At an overall coverage of 21%, vaccination has averted on average 13% (7.2-19%, 95% range) of infections, 24% (16-36%) of hospitalisations, and 35% (16-50%) of deaths. This suggests that vaccination is mainly effective in protecting vaccinees from infection rather than reducing transmission. As the Dutch population continues to grow and age, the vaccination programme is projected (up to 2025) to gain in impact, despite a decreasing infection attack rate.