Vaccine Efficacy Assessment for COVID-19

This Summary Document provides guidance on the assessment of Vaccine Efficacy (VE) for use in COVID-19 interventional trials using an adaptive-designed number-of-events approach. This concept is under development for the WHO's Solidarity trial. It is recommended that all COVID-19 vaccine interventional trials prospectively collect incident COVID-19 cases. The primary outcome measure of interest for establishing VE, is COVID-19 clinical disease as opposed to SARS-CoV-2 infection. Clinical trials that assess incident COVID-19 cases should use standard criteria to establish the clinical diagnosis of COVID-19 disease [defined in a separate summary document] and consider implementing a committee for clinical endpoint adjudication of COVID-19 cases on a continuous basis.
An assumption on background incident rates of clinical COVID-19 endpoints in the placebo arm is impractical and may be counterproductive. During a public-health emergency, incidence rates differ from one day to the next, from location to location, and differ from one risk group to the other. Moreover, COVID-19 spread amongst the population is managed by classic public health interventions such as social distancing, quarantine and isolation, further complicating the assumptions required to estimate the number of persons to be randomized in the interventional trial.
An alternative approach is a design in which power and precision are not dictated by the size of the trial but rather by the overall number of cases identified for the primary endpoint. This means that the size of the trial cannot be accurately predicted ahead of time and that enrolment should continue until the number of required cases has been reached. Sponsors can define a pre-defined minimum sample size based on the size required for the safety database.
Ideally, early stage clinical development trials for COVID-19 vaccine candidates should prospectively collect incident disease cases to aid exploratory assessment of VE. Standardized consistent endpoint definitions for clinical disease across early and advanced stage trials would facilitate pooled analysis.