Vaccines: Populations, Individuals and Models

Abstract

The 2009 H1N1 influenza and 2019 SARS-CoV-2 pandemics have highlighted the need for control measures against emerging infectious diseases. Vaccines are currently the most effective intervention against these pathogens, however decisions regarding who, when, and where to vaccinate can be complicated by competing political, practical, and ethical considerations. While models can potentially help inform these decisions, understanding which parameters will be both relevant and unbiased in an infectious disease model can be challenging. Here we describe considerations decision makers and modelers face in allocating vaccines, recommending vaccination policies, and quantifying the impact of vaccine effectiveness on different disease outcomes. We first examine optimal allocation decisions across two locations considering population, vaccine, pathogen, and delivery characteristics, and compare this optimal allocation strategy to all possible strategies. We find that, across a range of scenarios considered, pro-rata (i.e., allocation proportional to population size) performs better or comparably to more nonproportional strategies in minimizing the cumulative number of infections. Next, we examine seasonal influenza vaccination strategies and their impact on the joint probability of PCR-confirmed influenza infection across two seasons. Previous studies have shown that vaccination in a prior season can impact the risk of influenza infection in subsequent seasons. We aimed to understand, given this two-season effect, if there were any scenarios under which an individual should not get vaccinated in a specific season to minimize the risk of infection in a future season. We find that for any given season vaccination reduces the mean predicted risk of infection for that season This holds true in both the prior season and current season among individuals of each vaccination and infection history in the prior season. Finally, we discuss challenges in parameterizing infectious disease models, focusing on individual based models. We aim to understand what study designs can translate to model parameters and propose study designs to yield unbiased parameter estimates. We find that parameterizing disease models can be challenging depending on how models are parameterized and that even careful design and analysis of studies may not yield causal parameters of interest. These studies highlight the complexities of vaccination decisions and quantification of these decisions.