Vaccination is an important epidemic intervention strategy. Resource limitations and an imperative for efficient use of public resources drives a need for optimal allocation of vaccines within a population. For a disease causing severe illness in particular members of a population, an effective strategy to reduce illness might be to vaccinate those vulnerable with a vaccine that reduces the chance of catching a disease. However, it is not clear that this is the best strategy, and it is generally unclear how the difference between various vaccine strategies changes depending on population characteristics, vaccine mechanisms and allocation objective. In this paper we develop a mathematical model to consider strategies for vaccine allocation, prior to the establishment of community transmission. By extending the SEIR model to incorporate a range of vaccine mechanisms and disease characteristics, we study the impact of vaccination on a population comprised of individuals at high and low risk of infection. We then compare the outcomes of optimal and suboptimal vaccination strategies for a range of public health objectives using numerical optimisation. Our comparison shows that the difference between vaccinating optimally and suboptimally is dependent on vaccine mechanism, diseases characteristics, and objective considered. Even when sufficient vaccine resources are available, allocation strategy remains important as allocating suboptimally could result in a worse outcome than allocating limited vaccine resources optimally. This work highlights the importance of designing effective vaccine allocation strategies, as allocation of resources can be just as crucial to the success of the overall strategy as total resources available.

No abstract

Since the emergence of SARS-CoV-2 (COVID-19), there have been multiple waves of infection and multiple rounds of vaccination rollouts. Both prior infection and vaccination can prevent future infection and reduce severity of outcomes, combining to form hybrid immunity against COVID-19 at the individual and population level. Here, we explore how different combinations of hybrid immunity affect the size and severity of near-future Omicron waves. To investigate the role of hybrid immunity, we use an agent-based model of COVID-19 transmission with waning immunity to simulate outbreaks in populations with varied past attack rates and past vaccine coverages, basing the demographics and past histories on the World Health Organization (WHO) Western Pacific Region (WPR). We find that if the past infection immunity is high but vaccination levels are low, then the secondary outbreak with the same variant can occur within a few months after the first outbreak; meanwhile, high vaccination levels can suppress near-term outbreaks and delay the second wave. Additionally, hybrid immunity has limited impact on future COVID-19 waves with immune-escape variants. Enhanced understanding of the interplay between infection and vaccine exposure can aid anticipation of future epidemic activity due to current and emergent variants, including the likely impact of responsive vaccine interventions.

Early case detection is critical to preventing onward transmission of COVID-19 by enabling prompt isolation of index infections, and identification and quarantining of contacts. Timeliness and completeness of ascertainment depend on the surveillance strategy employed. We use rapid prototype modelling to quickly investigate the effectiveness of testing strategies, to aid decision making. Models are developed with a focus on providing relevant results to policy makers, and these models are continually updated and improved as new questions are posed. The implementation of testing strategies in high risk settings in Australia was supported using models to explore the effects of test frequency and sensitivity on outbreak detection. An exponential growth model is firstly used to demonstrate how outbreak detection changes with varying growth rate, test frequency and sensitivity. From this model we see that low sensitivity tests can be compensated for by high frequency testing. This model is then updated to an Agent Based Model, which was used to test the robustness of the results from the exponential model, and to extend it to include intermittent workplace scheduling. These models help our fundamental understanding of disease detectability through routine surveillance in workplaces and evaluate the impact of testing strategies and workplace characteristics on the effectiveness of surveillance. This analysis highlights the risks of particular work patterns while also identifying key testing strategies to best improve outbreak detection in high risk workplaces.