Date of Award

2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Public Health Sciences

College

College of Graduate Studies

First Advisor

Renee' Hebert Martin

Second Advisor

Yuko Y. Palesch

Third Advisor

Marc Chimowitz

Fourth Advisor

Edsel Peña

Fifth Advisor

Viswanathan Ramakrishnan

Abstract

The modified Rankin Scale (mRS), a seven-point ordinal scale ranging from no symptoms to death, is the most commonly used outcome measures in acute stroke therapy trials. Often, one visit is chosen for the primary analysis, and the scale is dichotomized leading to loss of information. Recently, alternative methods for analyzing the mRS have been explored. In addition, acute onset conditions require immediate attention and treatment, posing a challenge to assess baseline outcome measures for clinical trials. Thus, the mRS is not obtainable at baseline. Much of the progression or recovery experienced by a patient suffering from an acute onset disease is expected to occur early on. Moreover, typically, the goal of a treatment or therapeutic action is improvement in patient health compared to their baseline measure. To accurately quantify improvement, a measure of the outcome at baseline is ideal. This dissertation first explores the feasibility of multistate Markov models for the analysis of the mRS which allow for the full ordinal scale as well as the repeated measures data to be incorporated. The operating characteristics (type I error and power) of the multistate Markov model are compared with those from repeated logistic regression. Next, a framework is developed to predict and incorporate the latent baseline mRS score in a piecewise-constant multistate model. The last part of this work applies the piecewise-constant latent baseline model to real acute stroke trial data and compares the results with alternative methods for analysis of the mRS.

Rights

All rights reserved. Copyright is held by the author.

Download

Share

COinS