Methods/Statistics
A novel estimand for quantifying vaccine efficacy against asymptomatic infections Elizabeth Rogawski McQuade* Elizabeth Rogawski McQuade Razieh Nabi David Benkeser
For preventive vaccines that do not confer sterilizing immunity, differences in vaccine efficacy by severity of disease are of interest. Such vaccines may prevent infections entirely in some individuals, but in others may only prevent clinical symptoms or lessen the severity of the clinical disease caused by infection. For these vaccines, it is common practice to report vaccine efficacy against both symptomatic disease and infection. Many studies additionally report a vaccine efficacy measure against asymptomatic infection, comparing the risk of asymptomatic infections in the vaccinated and placebo groups. Such estimates can be misleading because they mix the effects of vaccines preventing asymptomatic infections and of vaccines converting symptomatic to asymptomatic infections. When the latter effect is strong, vaccines can appear harmful with respect to asymptomatic infections (i.e., negative vaccine efficacy). In the context of the COVID-19 pandemic, these types of findings led to considerable confusion in the general population and may have led to increased vaccine hesitancy. In this work, we demonstrate how the typical formulation of vaccine efficacy against asymptomatic infections is not the most relevant estimand for quantifying the biological impact of vaccines on the infection process. Using a causal principal stratification framework, we suggest an alternative estimand for quantifying vaccine impact on asymptomatic infection that is a more natural analogue of the usual vaccine efficacy estimands against infection and symptomatic disease and which excludes vaccine effects that convert symptomatic cases to asymptomatic infections. We describe assumptions under which this estimand can be identified and estimated from randomized and observational studies, and we derive bounds for the estimand that do not require cross-world assumptions. We then apply these methods in the COVE study, a randomized trial of the Moderna COVID-19 (mRNA-1273) vaccine.