From: Bayesian adaptive designs for multi-arm trials: an orthopaedic case study
 | Design 1 | Design 2 | Design 3 | Design 4 | Design 5 | Design 6 |
---|---|---|---|---|---|---|
Proportion stopping for efficacy at 200 patients | NA | 0.216 | 0.148 | 0.166 | 0.147 | 0.072 |
Proportion stopping for efficacy at 400 patients | NA | 0.043 | 0.011 | 0.017 | 0.011 | 0.004 |
Proportion stopping for futility at 200 patients | NA | 0 | 0 | 0 | 0 | 0 |
Proportion stopping for futility at 400 patients | NA | 0 | 0 | 0 | 0 | 0 |
Proportion re-executions declared successful at final analysis | 0.855 | 0.894 | 0.835 | 0.865 | 0.877 | 0.23 |
Proportion re-executions tubular bandage (control) declared best at final analysis | 0 | 0 | 0.001 | 0 | 0 | 0 |
Proportion re-executions boot declared best at final analysis | 0.054 | 0.057 | 0.085 | 0.036 | 0.021 | 0.007 |
Proportion re-executions brace declared best at final analysis | 0.437 | 0.402 | 0.43 | 0.451 | 0.481 | 0.432 |
Proportion re-executions below-knee cast declared best at final analysis | 0.509 | 0.541 | 0.484 | 0.513 | 0.498 | 0.561 |
Median (IQR) of the posterior mean estimates for tubular bandage | 54.25 (52.70–55.68) | 53.72 (51.90–55.46) | 54.40 (52.99–55.74) | 53.91 (52.52–55.30) | 53.97 (52.64–55.33) | 52.49 (51.68–52.96) |
Median (IQR) of the posterior estimates of the difference in means between boot and tubular bandage | 5.60 (3.65–7.48) | 6.00 (4.02–8.25) | 5.65 (3.75–7.56) | 4.77 (2.42–6.84) | 4.85 (2.58–7.05) | 6.42 (3.98–8.15) |
Median (IQR) of the posterior estimates of the difference in means between brace and tubular bandage | 8.60 (6.52–10.63) | 8.66 (6.67–10.89) | 7.62 (4.81–10.22) | 8.48 (5.65–10.71) | 8.67 (5.99–10.73) | 9.64 (6.01–11.66) |
Median (IQR) of the posterior estimates of the difference in means between below-knee cast and tubular bandage | 8.70 (6.86–10.91) | 9.69 (7.22–13.29) | 8.06 (5.44–10.53) | 8.79 (6.57–11.39)– | 8.68 (6.58–11.27) | 10.57 (8.69–11.78) |