Skip to main content

Table 4 Virtual re-execution of shortlisted designs using real-world ANZ 9311 data

From: Improving clinical trials using Bayesian adaptive designs: a breast cancer example

Design Trial conclusion Duration (months) Sample size Hazard ratio (95% CI)
Original trial H0 not rejected 53.0 recruitment + 91.2 follow-up 235 1.17 (0.90, 1.52)
Small sample size
  Posterior Early futility 69.9 233 1.14 (0.85, 1.52)
  PPS Early futility 50.6 214 1.03 (0.71, 1.48)
  CPS Early futility 43.2 164 1.00 (0.66, 1.52)
  Goldilocks Early futility 50.6 214 1.03 (0.71, 1.48)
  PPBS Early futility 42.3 158 0.98 (0.64, 1.50)
High power
  Posterior Inconclusive 120.5 233 1.17 (0.90, 1.53)
  PPS Early futility 52.4 228 1.10 (0.77, 1.58)
  CPS Early futility 47.8 189 1.00 (0.68, 1.47)
  Goldilocks Early futility 52.4 228 1.10 (0.77, 1.58)
  PPBS Early futility 47.8 189 1.00 (0.68, 1.47)
  1. PPS Predictive probability of success, CPS Conditional probability of success, PPBS Predictive probability of Bayesian success. The hazard ratios and 95% confidence intervals (CI) are the estimates from the Cox proportional-hazards model, unadjusted for any adaptations, for illustrative purposes only