All ETDs from UAB

Advisory Committee Chair

Gary Cutter

Advisory Committee Members

Christopher Coffey

Leslie McClure

Jenifer Voeks

Khurram Bashir

Document Type


Date of Award


Degree Name by School

Doctor of Philosophy (PhD) School of Public Health


Phase III trials often have both efficacy and safety outcomes. Sometimes the safety outcomes are closely associated with the efficacy outcomes. We sought to find a statistical method to treat both efficacy and safety outcomes as bivariate primary outcomes and set up stopping rules at interim stages based on the following criteria: (1) greater safety; (2) greater efficacy; (3) futility. We first examined the internal pilot design to adjust sample size for clinical trials with a bivariate response (efficacy and safety). Three nuisance parameters (two variances and correlation) need to be considered when doing sample size recalculation for the internal pilot design. We adopted Cook & Farewell’s method to test the joint hypothesis with bivariate outcomes and incorporated it into the schema of internal pilot designs. We performed simulations for fixed sample designs and unrestricted and restricted internal pilot designs. The results show the correlation has minimal effect on power, Type I error and sample size. We also show that inaccurate estimation of variances almost has the same effect on power and Type I error as in the univariate case. We also extended Cook & Farewell’s method to monitor efficacy, safety and futility simultaneously in interim analyses by adding the conditional power approach into the bivariate group sequential tests. The simulation results show an improved power while maintaining the overall Type I error rate. Finally, we applied this method in the trial of early-inhaled nitric oxide (NO) therapy in term and near-term infants with iii respiratory failure. We varied the analytical conditions such as the time of doing interim analysis, the weight of efficacy analysis and the type of α spending function. We found that when our method is used, the NO trial should be stopped for futility at the interim stage after more than 50% of primary outcomes are available. This result is the same as the actual result from the original NO trial analysis. We conclude that this method can provide a statistical guideline for the early stopping of a trial due to efficacy, safety and futility.

Included in

Public Health Commons



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.