Fig. 2

Trial sizes for a single effective intervention among 4 candidates (A, with enrolment of separate trials summed) or among 4 to 20 candidates (C and D, same colors as in A) with either 2 (A-D), 4 (E) or 8 (F) allocations per patient. B shows the total trial size ratio compared to the parallel-arm design (each intervention analyzed versus control) and the associated gains, with the same color scheme as in A. E and F show improvements of the distributive and capped factorial design as more interventions (4 and 8 respectively) are tested per patient (same colors as in A). The efficacy assumption is that a single intervention brings the probability of a good outcome from 0.5 to 0.7, and the other interventions have no effect. Alpha risks (nominal 5%) are Bonferroni-adjusted for multiplicity, and 90% power is sought. *the full factorial design is shown for reference but ignores maximum allocations