Statistical considerations for a trial of Ebola virus disease therapeutics

Abstract

The 2014 West African outbreak of Ebola virus ravaged Liberia, Sierra Leone, and Guinea, causing hemorrhagic fever and death. The need to identify effective therapeutics was acute. The usual drug development paradigm of phase I, followed by phase II, and then phase III trials would take too long. These and other factors led to the design of a clinical trial of Ebola virus disease therapeutics that differs from more conventional clinical trial designs. This article describes the Ebola virus disease medical countermeasures trial design and the thinking behind it.

Keywords: Barnard’s test; Bayesian methods; Fisher’s exact test; beta-binomial distribution; conditional power; emerging infectious diseases; group-sequential monitoring; non-informative prior.

Figure 1

Comparison of rejection regions with 10/arm. If the number of deaths in the control arm is greater than or equal to the boundary value, treatment is declared beneficial.

Figure 2

The rate of spending of type 1 error rate of the barely Bayesian method (black curve) compared to the O'Brien-Fleming (blue curve) and Pocock (red curve) methods for the monitoring schedule depicted as hash marks on the x-axis.

Figure 3

Differences in 28-day mortality probabilities, p_A – p_B, at the first hypothetical analysis with n_A = n_B = 6 and the second hypothetical analysis with n_A = 12, n_B = 11. The 99.8% credible intervals are shaded.

Figure 4

Ratio of 28-day mortality probabilities, p_A/p_B, at the first hypothetical analysis with n_A = n_B = 6 and the second hypothetical analysis with n_A = 12, n_B = 11. The 99.8% credible intervals are shaded.

Figure 5

Cumulative power by different times in the trial for the barely Bayesian method compared to Fisher's exact test coupled with the Haybittle-Peto procedure.