Electronic Thesis and Dissertation Repository

Thesis Format



Master of Science


Epidemiology and Biostatistics


Zou, Guangyong.

2nd Supervisor

Jairath, Vipul.

Joint Supervisor


A randomized controlled trial is often used to provide high quality evidence regarding treatment interventions. Due to the complex nature of many diseases, trials usually select multiple primary outcomes to capture the efficacy of the interventions. In this thesis, we conducted a literature search to determine the prevalence of the different types of multiple outcomes that have been used in randomized controlled trials. We also reviewed the corresponding statistical methods used to deal with such outcomes. In addition, we described the benefits of using global tests as a statistical method when there are multiple primary outcomes in order to answer the global question of whether the intervention is effective. As an illustration for the global test, we used data from a previously published trial in ulcerative colitis. The global tests included O’Brien’s OLS, Lauter’s test and O’Brien’s rank-sum test, and all of the tests used produced statistically significant results with p-values less than 0.05. Global tests should be considered when using multiple outcomes as well as additional guidelines surrounding how multiple primary outcomes should be managed.

Summary for Lay Audience

Medical research questions can be answered with many different types of study designs, each with its own strengths and weaknesses. A randomized controlled trial (RCT) is considered to provide the highest quality evidence. A RCT may be used to evaluate the effects of a new drug, treatment strategy, a different surgical protocol or a lifestyle change by comparing groups that receive the intervention of interest to a control group(s). When designing an RCT, it is important to have a clear research question. The primary outcome that is chosen for the study should attempt to answer the research question. Most research teams will choose a singular outcome that encompasses the research question, but sometimes the question may not be able to be answered by a single outcome. In this case, there would be multiple primary outcomes used in order to capture all of the information required. There are multiple ways to deal with multiple outcomes. First, one may choose one outcome as the primary and leave the rest as secondary outcomes. Second, one may use an established scoring system to create a composite outcome based on the multiple outcomes. Third, one may apply simultaneous tests to determine if at least one of the outcomes occurs or is statistically significant compared to a control group. Finally, one may treat multiple outcomes as co-primary outcomes. This thesis reports a literature search aimed at determining the prevalence and approaches used to handle multiple outcomes in randomized controlled trials in the medical literature.

There are many statistical methods that can be used to evaluate a dataset and there is not one specific method that is used when there are multiple primary outcomes present. In this thesis, we demonstrate one type of method that can be used which are global tests. Global tests provide certain advantages over other methods that can be used. To demonstrate how global tests can be used, a dataset from a previously conducted clinical trial was used. The results produced from the global tests were then compared to the originally published results.