Skip to main content

Table 4 Simulation results for dropping treatment arms based on the second rule (R2) and the absolute bias for such arms in the estimated treatment effect at the time of dropping decision – all priors on pk (k=A,B,C) are non-informative beta(1,1) priors, when decision threshold is set at 0.05

From: Evaluation of a multi-arm multi-stage Bayesian design for phase II drug selection trials – an example in hemato-oncology

Sample size True benefit Posterior mean estimate biases Average sample sizes % Early stopping
n dB dC pA pB pC A B C B C
40 −0.20 0.00 0.0446 0.0410 −0.0036 34.7688 19.0005 34.1757 83.18 % 20.90 %
  −0.15 0.00 0.0407 0.0196 −0.0049 35.4230 23.9111 34.1465 65.96 % 21.12 %
  −0.10 0.00 0.0384 0.0076 −0.0045 36.0178 28.2528 33.9857 47.31 % 21.51 %
  −0.05 0.00 0.0325 0.0004 −0.0039 36.8988 31.7086 34.2757 31.84 % 20.67 %
  0.00 0.00 0.0289 −0.0046 −0.0034 37.5750 34.3312 34.2628 20.68 % 20.55 %
  0.05 0.00 0.0262 −0.0053 −0.0046 38.1404 35.9159 34.2368 13.61 % 21.01 %
  0.10 0.00 0.0212 −0.0068 −0.0038 38.6097 37.1804 34.2661 8.79 % 20.81 %
  0.15 0.00 0.0184 −0.0079 −0.0043 39.0075 38.0515 34.3098 5.79 20.69 %
  0.20 0.00 0.0163 −0.0098 −0.0051 39.2281 38.5864 34.2477 4.01 % 20.85 %
100 −0.20 0.00 0.0458 0.0362 −0.0144 80.7525 22.0750 80.0766 99.05 % 26.36 %
  −0.15 0.00 0.0427 0.0103 −0.0153 81.9307 34.6011 79.9595 91.50 % 26.58 %
  −0.10 0.00 0.0418 −0.0063 −0.0152 83.5055 51.2404 79.0897 71.48 % 27.30 %
  −0.05 0.00 0.0344 −0.0142 −0.0153 87.6493 67.8997 80.1142 46.17 % 26.49 %
  0.00 0.00 0.0277 −0.0161 −0.0157 90.8189 79.7603 79.7672 26.62 % 26.82 %
  0.05 0.00 0.0247 −0.0149 −0.0154 93.1423 86.5031 79.4970 15.96 % 26.65 %
  0.10 0.00 0.0202 −0.0130 −0.0157 95.3857 91.6455 78.7749 9.26 % 27.54 %
  0.15 0.00 0.0154 −0.0119 −0.0155 96.8064 94.2754 79.4074 6.15 % 27.04 %
  0.20 0.00 0.0112 −0.0100 −0.0147 97.8693 96.1853 80.1669 4.03 % 26.14 %
  1. pA, pB and pC refer to the posterior means of response probability in arms A, B and C, respectively; Δ=0. Bold data refer to the null hypothesis of absence of any treatment difference (dB=dC=0), or to the alternative hypothesis of an expected true 0.15 benefit of treatment B (dB=0.15)