Skip to main content

Table 1 Bayesian group sequential designs explored for the PARAMEDIC2 study

From: Bayesian group sequential designs for phase III emergency medicine trials: a case study using the PARAMEDIC2 trial

Interim analysis (i)

Stopping boundaries

Number of patients recruited

adrenaline better

placebo better

B1 interim analysis schedulea

B2 interim analysis schedule

B3 interim analysis schedule

1

0.9999

0.99999

50

50

500

2

0.9998

0.99999

400

300

1000

3

0.9997

0.99998

1100

600

1500

4

0.9996

0.9998

1800

1000

2000

5

0.9995

0.9997

2400

1450

2500

6

0.9994

0.9996

3100

1900

3000

7

0.9993

0.9995

3800

2650

3500

8

0.9992

0.9994

4500

3650

4000

9

0.9991

0.9993

5200

5000

4500

10

0.999

0.9992

5900

6500

5000

11

0.998

0.999

NA

7000

5500

12

0.996

0.998

NA

7500

6000

13

0.994

0.997

NA

NA

6500

14

0.992

0.996

NA

NA

7000

15

0.99

0.994

NA

NA

7500

Final analysisb

0.977

0.977

Max 8000

Max 8000

Max 8000

  1. aApproximately where the interim analyses occurred since these were on a time basis (3-monthly) and the recruitment was simulated using a Poisson process
  2. bIf the trial did not stop early, then the final analysis was performed once 8000 patients had been recruited and followed up; if the trial stopped early, then the final analysis was performed once the recruited patients had been followed up