Assessing Vaccine Effects on Infectiousness in COVID-19 Vaccine Trials

Abstract

Vaccines have played an important role in the public health response to the COVID-19 pandemic. Clinical trials for several candidate vaccines demonstrated high efficacy at pre- venting severe / symptomatic disease. These vaccines have helped lower the burden of disease in the countries where they have been widely distributed (e.g., United States). The next generation of COVID-19 vaccine trials are designed to evaluate other vaccine effects (e.g., ef- ficacy against infection). Understanding and quantifying how vaccines affect infectiousness is important for making sound public health policy, and analyzing such effects in a clinical trial requires selecting an appropriate endpoint and vaccine efficacy parameter. Many analyses of vaccine effects on infectiousness condition on infection (e.g., comparing infected vacci- nated people to infected unvaccinated people), but conditional analyses have shortcomings. Most glaringly, they can lack power when efficacy against infection is high and may suffer from post-randomization selection bias. An alternative approach is to explore endpoints and vaccine efficacy parameters that do not condition on infection (i.e., unconditional analyses).In this thesis, we define and discuss multiple methods for unconditional analyses of vaccine effects on infectiousness. As these methods commonly rely on proxies of infectiousness rather than directly observed secondary transmission, we also discuss several infectiousness proxies derived from measures of viral shedding. Finally, we use simulations to demonstrate the operating characteristics of a number of these methods and infectiousness proxies in a clinical trial whose design is based on an ongoing COVID-19 vaccine trial.