|
Activity Number:
|
398
- Different Approaches to the Increase of a Sample Size When the Unblinded Interim Estimate of the Treatment Effect Looks Promising
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 1, 2017 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #323358
|
View Presentation
|
|
Title:
|
An Evaluation of Increasing Sample Size Based on Conditional Power
|
|
Author(s):
|
Michael Gaffney*
|
|
Companies:
|
Pfizer Inc.
|
|
Keywords:
|
Adaptive design ;
Conditional power ;
Sample size re-estimation
|
|
Abstract:
|
Properties of sample size re-estimation (SSR) designs similar to the promising zone design considered by Mehta and Pocock (2011) are evaluated under the assumption of a true effect size of 1.1 down to 0.4 of the protocol specified effect size. Six measures are used: 1. The probability of a sample size increase; 2. The mean proportional increase in sample size given an increase; 3 and 4. The mean true conditional power with and without a sample size increase; and 5 and 6. The expected increase in sample size and power due to the SSR procedure. These measures show the probability of a sample-size increase and the cost/benefit for given true effect sizes, particularly when the SSR may either be pursuing a small effect size of little clinical importance or be unnecessary when the true effect size is close to the protocol specified effect size. The results show the clear superiority of conducting the SSR late in the study and the inefficiency of a mid-study SSR. Waiting until late in the study for the SSR yields a smaller, better targeted set of studies with a greater increase in overall power than a mid-study SSR.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2017 program
|
