COVID-19 vaccine trials: The use of active controls and non-inferiority studies

Abstract

Background: Recently emerging results from a few placebo-controlled randomized trials of COVID-19 vaccines revealed estimates of 62%–95% relative reductions in risk of virologically confirmed symptomatic COVID-19 disease, over approximately 2-month average follow-up period. Additional safe and effective COVID-19 vaccines are needed in a timely manner to adequately address the pandemic on an international scale. Such safe and effective vaccines would be especially appealing for international deployment if they also have favorable stability, supply, and potential for implementation in mass vaccination campaigns. Randomized trials provide particularly reliable insights about vaccine efficacy and safety. While enhanced efficiency and interpretability can be obtained from placebo-controlled trials, in settings where their conduct is no longer possible, randomized non-inferiority trials may enable obtaining reliable evaluations of experimental vaccines through direct comparison with active comparator vaccines established to have worthwhile efficacy. Methods: The usual objective of non-inferiority trials is to reliably assess whether the efficacy of an experimental vaccine is not unacceptably worse than that of an active control vaccine previously established to be effective, likely in a placebo-controlled trial. This is formally achieved by ruling out a non-inferiority margin identified to be the minimum threshold for what would constitute an unacceptable loss of efficacy. This article not only investigates non-inferiority margins, denoted by δ, that address the usual objective of determining whether the experimental vaccine is “at least similarly effective to” the active comparator vaccine in the non-inferiority trial, but also develops non-inferiority margins, denoted by δo, intended to address the worldwide need for multiple safe and effective vaccines by satisfying the less stringent requirement that the experimental vaccine be “at least similarly effective to” an active comparator vaccine having efficacy that satisfies the widely accepted World Health Organization–Food and Drug Administration criteria for “worthwhile” vaccine efficacy. Results: Using the margin δ enables non-inferiority trials to reliably evaluate experimental vaccines that truly are similarly effective to an active comparator vaccine having any level of “worthwhile” efficacy. When active comparator vaccines have efficacy in the range of 50%–70%, non-inferiority trials designed to use the margin δo have appealing properties, especially for experimental vaccines having true efficacy of approximately 60%. Conclusion: Non-inferiority trials using the proposed margins may enable reliable randomized evaluations of efficacy and safety of experimental COVID-19 vaccines. Such trials often require approximately two- to three-fold the person-years follow-up than a placebo-controlled trial. This could be achieved, without substantive increases in sample size, by increasing the average duration of follow-up from 2 months to approximately 4–6 months, assuming efficacy of the active comparator vaccine has been reliably evaluated over that longer duration.