Grizzle Linear Model for Analysis of Clinical Trials Using Crossover Designs Open Access

Abstract

Background: The Melatonin and Metabolic Syndrome (MetSyn) study was a double-blind, placebo-controlled, crossover, phase II randomized clinical trial. Its main purpose was to investigate the effects of melatonin supplementation on treating patients with metabolic syndrome. This thesis analyzes results of this crossover trial with an emphasis on examining the univariate and multivariate effects of carry-over and treatment on the main components of metabolic syndrome.

Methods: In addition to fitting the traditional linear model (Grizzle) to detect the effects of carry-over and treatment on various measures, we tested linear combinations of the five main components of metabolic syndrome derived from a principal component analysis.

Results: Only one variable considered, a measure of sleep efficiency, showed a significant effect of carry-over (F=6.29, p=0.02). A significant treatment effect was detected for the average change of clinic SBP comparing melatonin with placebo (F=6.86, p=0.01). However, no treatment effects were found on any other measures. Interestingly, the first component retained from the principal component analysis reflected mainly changes of triglycerides and waist circumference, while the second represented mainly the changes in HDL and SBP average. Further, one of the two principal components retained from the principal component analysis exhibited a significant treatment effect (F=4.70, p=0.04).

Conclusions: Given that no significant carry-over effect was found among most of the measures of interest, the crossover design of this phase II clinical trial was considered appropriate. The significant treatment effect on clinic SBP suggested a promising benefit of melatonin supplementation in SBP improvement. Since the principal component analysis yielded two distinct principal components, it indicated the five major components of the syndrome may interact through different mechanisms and this should be considered in future studies. Thus the use of the data reduction method (principal component analysis) provided a multivariate solution to this crossover trial analysis that further showed the additional advantage of melatonin in improving HDL.

Table of Contents

Chapter 1 Introduction. 1

Chapter 2 Review of Literature. 4

2.1 The Clinical Problem. 4

2.1.1 Metabolic Syndrome. 4

2.1.2 Melatonin Treatment. 5

2.1.3 Melatonin Efficacy in Treating Metabolic Syndrome. 6

2.2 The Statistical Problem. 8

2.2.1 Crossover Trials. 8

2.2.2 Illustration of Comparisons. 10

2.2.3 The Grizzle Model. 11

2.2.4 Principal Component Analysis. 13

Chapter 3 Results. 16

Chapter 4 Conclusions. 21

Bibliography. 23

Appendix. 28

About this Master's Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Rollins School of Public Health
Department	Biostatistics and Bioinformatics
Subfield / Discipline	Biostatistics - MPH & MSPH
Degree	MSPH
Submission	Master's Thesis
Language	English
Research Field	Biology, Biostatistics Health Sciences, Public Health
Keyword	crossover trials Grizzle model melatonin supplementation principal component analysis metabolic syndrome
Committee Chair / Thesis Advisor	Kelley, Mary, Emory University
Committee Members	Zhang, Rebecca, Emory University
Partnering Agencies	Emory University schools, faculty or affiliated programs Hospital or other health care provider University, college or educational institution (other than Emory)

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Grizzle Linear Model for Analysis of Clinical Trials Using Crossover Designs ()	2018-08-28 10:38:47 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions