|
Activity Number:
|
419
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #310285 |
|
Title:
|
On Errors in Stratifications
|
|
Author(s):
|
Chunlei Ke*+ and Jianming Wang and Xiang Zhang and Qi Jiang and Alan Rong and Steven Snapinn
|
|
Companies:
|
Amgen Inc. and Amgen Inc. and Amgen Inc. and Amgen Inc. and Amgen Inc. and Amgen Inc.
|
|
Address:
|
One Amgen Center Drive, Newbury Park, CA, 91320,
|
|
Keywords:
|
stratified randomization ; clinical trial ; error in variable ; chi-square test
|
|
Abstract:
|
Stratified Randomization is widely used in clinical trials to achieve balance of the treatment assignment with regard to important prognostic factors. It may happen that the value on the stratification factor provided at randomization is later identified to be incorrect late. Thus patients may be allocated into the strata they should not belong to. We evaluated the impact of such errors on the intended balance, and found that randomization was still balanced at least at the level as in simple randomization. While the analysis based on the factor per randomized is usually chosen, the estimate of treatment effect based on such analysis was found unbiased on average. But the inference based on this based on this analysis approach was shown conservative, and thus inefficient. In contrast, the analysis utilizing the true stratification factor was found proper and more efficient.
|
