Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Dissertation_Mingyuan_Shao.pdf (5.77 MB)

ETD Abstract Container

Abstract Header

Multivariate Hierarchical Global Rank Test

Shao, Mingyuan
http://rave.ohiolink.edu/etdc/view?acc_num=case1544627557687073

Abstract Details

, Doctor of Philosophy, Case Western Reserve University, Epidemiology and Biostatistics.
Many clinical trials entail multiple endpoints, reflecting diverse aspects of treatment effects, often yielding challenging research questions on how best to analyze and interpret the results. Traditional approaches analyze endpoints individually, leading to increased type I error or larger required sample sizes if adjustment is made for multiple testing. In addition, small-to-mid-size trials are often underpowered on clinical outcomes when the event rate is low. A multivariate hierarchical global rank test, a nonparametric approach that incorporates the same or different types of endpoints into a hierarchical composite endpoint based on relative clinical importance and tests research hypotheses using ranking strategies, has been proposed as a plausible alternative adding to traditional approaches. Various global rank test approaches, with nice applications in the clinical setting, have been published. However, there is still room to improve the precision of these global ranking tests. In this dissertation research, different hierarchical global rank tests are reviewed, especially in the aspect of variance estimation, and some improvements on the global rank test are proposed by developing stratification and weighting strategies based on Cox model risk score ranking. Simulation studies are designed and conducted on a hierarchical composite endpoint of three correlated component outcomes. The types of the three outcomes considered are, in order of hierarchy, time-to-event, continuous or dichotomous, and continuous. The performance of the global rank tests using different methods of variance estimation with and without stratification and weighting in the hierarchical or non-hierarchical setting, are compared with respect to type I error, statistical power, and coverage probabilities of confidence intervals of estimates on the global treatment effect. Finally, the global rank test methods are applied to data from a randomized clinical trial examining the effect of renin-angiotensin-aldosterone system inhibition on progression of coronary atherosclerosis in patients with prehypertension (the AQUARIUS trial). The findings suggest that the proposed stratification and weighting methods boost power while controlling type I errors at the nominal level.
Mark Schluchter (Advisor)
Biostatistics

Recommended Citations

Citations

  • Shao, M. (n.d.). Multivariate Hierarchical Global Rank Test [Doctoral dissertation, Case Western Reserve University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=case1544627557687073

    APA Style (7th edition)

  • Shao, Mingyuan. Multivariate Hierarchical Global Rank Test. Case Western Reserve University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=case1544627557687073.

    MLA Style (8th edition)

  • Shao, Mingyuan. "Multivariate Hierarchical Global Rank Test." Doctoral dissertation, Case Western Reserve University. Accessed MAY 05, 2024. http://rave.ohiolink.edu/etdc/view?acc_num=case1544627557687073

    Chicago Manual of Style (17th edition)

case1544627557687073
320
© , all rights reserved.
This open access ETD is published by Case Western Reserve University School of Graduate Studies and OhioLINK.