The Statistical and Practical Implications of Mis-Stratified Randomization in Clinical Trials.
Saved in:
| Title: | The Statistical and Practical Implications of Mis-Stratified Randomization in Clinical Trials. |
|---|---|
| Authors: | Zhang, Zhiwei1 (AUTHOR) Zhiwei.Zhang6@gilead.com |
| Source: | Statistics in Biopharmaceutical Research. Oct-Dec2025, Vol. 17 Issue 4, p627-637. 11p. |
| Subjects: | Clinical trials, Randomization (Statistics), Simulation methods & models, Treatment effectiveness, Measurement uncertainty (Statistics), Minimum variance estimation |
| Abstract: | Stratified randomization is widely used in clinical trials and commonly followed by a stratified analysis adjusting for the stratification factor. In stratified randomization, the stratification factor is often assessed with error or misclassification, at least initially, resulting in mis-stratified randomization. Prior studies of mis-stratified randomization have compared specific treatment effect estimators and test statistics under specific parametric regression models for outcome generation. This article provides a broader perspective on this problem without imposing parametric assumptions on the outcome generation mechanism and without restricting attention a priori to specific estimators. Under quite general conditions, we show that mis-stratification does not introduce a bias in treatment effect estimation but may attenuate the variance reduction effect of stratified randomization. If the stratification error is non-differential, there is an efficiency advantage in adjusting for the true (i.e., error-free) stratification factor as opposed to the working (i.e., error-prone) stratification factor. However, the key to efficient estimation is to adjust for all important covariates, particularly those unrelated to stratification. For variance estimation, the working stratification factor is more relevant to consider than the true stratification factor. These results and observations are verified empirically in a simulation study. [ABSTRACT FROM AUTHOR] |
| Copyright of Statistics in Biopharmaceutical Research is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Stratified randomization is widely used in clinical trials and commonly followed by a stratified analysis adjusting for the stratification factor. In stratified randomization, the stratification factor is often assessed with error or misclassification, at least initially, resulting in mis-stratified randomization. Prior studies of mis-stratified randomization have compared specific treatment effect estimators and test statistics under specific parametric regression models for outcome generation. This article provides a broader perspective on this problem without imposing parametric assumptions on the outcome generation mechanism and without restricting attention a priori to specific estimators. Under quite general conditions, we show that mis-stratification does not introduce a bias in treatment effect estimation but may attenuate the variance reduction effect of stratified randomization. If the stratification error is non-differential, there is an efficiency advantage in adjusting for the true (i.e., error-free) stratification factor as opposed to the working (i.e., error-prone) stratification factor. However, the key to efficient estimation is to adjust for all important covariates, particularly those unrelated to stratification. For variance estimation, the working stratification factor is more relevant to consider than the true stratification factor. These results and observations are verified empirically in a simulation study. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 19466315 |
| DOI: | 10.1080/19466315.2025.2463938 |
