products: Sequential Trial Design with ExpDesign Studio

Sequential Trial Studio is one of the components in ExpDesign Studio. It can also be used as stand-alone software.

Input Features Output Features
  • Different efficacy endpoints: Mean, median, proportion, odds ratio, or survival
  • Different number of experiment groups: One, paired or two groups
  • Equality test, or equivalence tests
  • O'Brien-Fleming's, Pocock's boundaries, Tsaits-Wang's boundary family, symmetric or triangle inner boundaries
  • Permit stopping for null or alternative hypothesis or both
  • One-sided or two-sided test
  • Balanced or unbalanced designs
  • Even or uneven size fractions for interim analyses
  • Parallel design or crossover designs
  • Input examples
  • And more...
For sequential trial design and monitoring:

  • Decision boundaries:
    • Stop boundary on sample mean scale
    • Stop boundary on sample p-scale
    • Stop boundary on standard z-statistic scale
  • Stop boundary plot
  • Expected and maximum sample size
  • Stopping probabilities
  • Design report
  • Conditional and predictive probability, and futility index
  • Repeated confidence interval
  • And much more...

Example: Two-sided test for paired mean difference with 4 analyses permitting early stopping for both Ho and Ha using Pocock's boundaries and triangle inner stopping boundaries.

A trial is to be designed to investigate the efficacy of an experiment drug, ABC, in asthma patients. Each patient in the study will take the placebo in the first period for 6 weeks and ABC in the second period for another 6 weeks. The two treatment periods will be separated by a washout period. The primary efficacy endpoint is the % change in FEV1. It is estimated that the mean % decease in FEV1 is 4 after taking the placebo and 11 after taking ABC. The standard deviation for the difference is 18%. You can generate different sequential designs for this trial. The following is a design with 4 analyses permitting stop for claiming either the null hypothesis or the alternative hypothesis. ExpDesign also generates detailed or brief reports.

Brief Report
This experiment design has 3 interim analyses and a final analysis. The sample sizes for the 4 analyses are 22, 44, 66, and 88, respectively. The sample size ratio between the two groups is 1. The maximum sample size for the design is 88 and the expected sample size is 51 under the null hypothesis and 54 under the alternative hypothesis. The calculation is based on a level of significance, alpha = 0.05, power = 0.8, mean difference = -6, and standard deviation = 18.

The decision rules are specified as follows:

At stage 1:
Accept null hypothesis if p-value > 1
Reject null hypothesis if p-value <= 0.0187
Otherwise, continue.

At stage 2:
Accept null hypothesis if p-value > 0.6121
Reject null hypothesis if p-value <= 0.0187
Otherwise, continue.

At stage 3:
Accept null hypothesis if p-value > 0.153
Reject null hypothesis if p-value <= 0.0187
Otherwise, continue.

At stage 4:
Accept null hypothesis if p-value > 0.0187
Reject null hypothesis if p-value <= 0.0187

How to Monitor a Group Sequential Trial with ExpDesign Studio

Monitoring a sequential trial is necessary because the deviation of analysis schedule and efficacy variable estimation, unpredictable safety factors, financial Constraints or other factors.

ExpDesign Studio provides conditional power, predictive power, futility index, and repeated confidence interval for trial monitoring.