- Title Information
- Title
- Analysis of Longitudinal Binary Data from Behavioral Medicine
- Name:
Personal
- Name Part
- Dunsiger, Shira
- Role
- Role Term:
Text
- creator
- Origin Information
- Copyright Date
(keyDate="yes", encoding="w3cdtf")
- 2009
- Physical Description
- Extent
- xi, 101 p.
- digitalOrigin
- born digital
- Note
- Thesis (Ph.D.) -- Brown University (2009)
- Name:
Personal
- Name Part
- Hogan, Joseph
- Role
- Role Term:
Text
- director
- Name:
Personal
- Name Part
- Gatsonis, Constantine
- Role
- Role Term:
Text
- reader
- Name:
Personal
- Name Part
- Marcus, Bess
- Role
- Role Term:
Text
- reader
- Name:
Corporate
- Name Part
- Brown University. Division of Biology and Medicine. Biostatistics
- Role
- Role Term:
Text
- sponsor
- Genre (aat)
- theses
- Abstract
- This thesis work is motivated by the specific challenges in analyzing data from clinical trials in behavioral medicine. Specifically, smoking cessation and physical activity
trials. Typically, the data are binary and longitudinal with an appreciable amount of missing responses. Here, we propose methodology to address questions of behavior change, treatment efficacy
and mediation. Although these methods are applied to behavioral data, they are applicable in other clinical trial settings as well. The first part of this thesis examines a method for
identifying distinct patterns of behavior change amongst participants in a smoking cessation trial. A latent class model is proposed which allows for a degenerate class of responses. A mean
model is used to identify behavior change based on both an initial intensity and slope function. Further subdivision based on first order correlation is also proposed. This model finds classes
of responses that are both clinically meaningful and informative. The second part of the thesis examines the challenges in estimating treatment efficacy in smoking cessation trials, which are
often subject to non-compliance. The G-computation algorithm is used to estimate the effect of receiving treatment. This method is first demonstrated in a hypothetical scenario and later applied
to real data. The G-computation algorithm is a tool that is both computationally efficient and particularly well-suited to smoking cessation trials. The third part of the thesis examines methods
for establishing mediation in a clinical trial from behavioral medicine. A critique of standard regression methods is provided as well as a summary of key statistical papers. A multiple
imputation approach is used to estimate the descriptive direct effect of treatment and consequently, the mediated effect of treatment on outcome. A case study is provided in which this estimator
is applied to data from a physical activity trial with a binary mediator. Extensions of this method are highlighted as are the advantages of using a causal method instead of a regression-based
approach.
- Subject (Local)
- Topic
- longitudinal binary data
- Subject (Local)
- Topic
- behavioral medicine
- Subject (Local)
- Topic
- causal inference
- Subject (Local)
- Topic
- ATE
- Subject (Local)
- Topic
- LCM
- Subject (FAST)
(authorityURI="http://id.worldcat.org/fast", valueURI="http://id.worldcat.org/fast/1015154")
- Topic
- Medicine and psychology
- Subject (FAST)
(authorityURI="http://id.worldcat.org/fast", valueURI="http://id.worldcat.org/fast/1013617")
- Topic
- Mediation
- Record Information
- Record Content Source (marcorg)
- RPB
- Record Creation Date
(encoding="iso8601")
- 20091218
- Language
- Language Term:
Code (ISO639-2B)
- eng
- Language Term:
Text
- English
- Identifier:
DOI
- 10.7301/Z0W094DK
- Access Condition:
rights statement
(href="http://rightsstatements.org/vocab/InC/1.0/")
- In Copyright
- Access Condition:
restriction on access
- Collection is open for research.
- Type of Resource (primo)
- dissertations